It works with what you already run.
No rip and replace. BOS connects to your existing systems first and migrates by plan, so value arrives before disruption does.
GENIE AI · Applied Intelligence Lab · Toronto
The data exists. The people exist. The systems exist. What’s missing is the layer that makes them work as one. GENIE AI builds that layer.
The gap
Finance has one figure, operations has another, and the board deck has a third. The meeting is spent reconciling instead of deciding.
Nothing is wrong with the request. It is simply waiting, and everything downstream is waiting with it.
A shift ends, a ticket moves teams, a person goes on leave. The work continues. The understanding does not.
None of this shows up on a balance sheet. All of it is on one.
Introducing BOS · The Business Autopilot
BOS is the operating layer between what your organization intends and what actually gets done. It connects your people, tools, data and agents, coordinates the work across all of them, and keeps a human on every decision that matters.
No rip and replace. BOS connects to your existing systems first and migrates by plan, so value arrives before disruption does.
Give it an objective in plain language. It decomposes the outcome, routes each step to the right person or agent, and tracks it to done.
Permission boundaries live in the architecture, not in a policy document. Every action is attributable, every decision has an owner.
A private, always-current model of how your organization actually runs, deepening with every action. The advantage is yours alone.
Your BOS team, at work
Watch the core BOS team take a sentence to a finished result: dividing the work, running it in parallel, and stopping for you exactly once, where your judgment matters.
Ready
ERP and CRM disagree on Q3 revenue by $41,200. Which is the system of record?
BOS stops here on purpose. It won’t guess on your numbers.
Board pack ready for your review.
Illustrative run. Every action is logged, attributable, and inside the permissions you set.
Impact studies
Healthcare · Asia · Live
Thousands of patient journeys a week, orchestrated using BOS. A lab result lands in the patient record and flags the treating doctor before anyone types a word. The data never leaves the building.
“We want our teams focused on patient care and saving lives, not on moving information between systems.”F. Abbasi · CEO, AIMS
Aviation · 50M+ passengers a year
3 hrs → 30 secThe staffing analysis that ate a manager’s morning now finishes in thirty seconds. A human still approves every fill.
Nonprofit · 350 people
$48,254A year that used to go into coordination, now going into programme work.
Government · Citizen services
ASL + intakeInteractive ASL avatars and citizen intake agents. Market benchmarks put well-implemented service agents at up to 80% of routine inquiries resolved without a human.
The hardest test
Preferred contact: phone. Language: English. Support needed: none. Every field was filled in before you arrived, by someone who never met you.
Press 1, press 2, press 4 to hear these options again. The menu was designed around the business, not around the reason you called.
The answer is right there, in a video. With audio only. No captions, no sign language, no other way in.
Home, support, topics, FAQ, contact, form. Six steps deep and still no answer, only another place to wait.
That feeling is the whole idea. Being understood, instead of processed. In your language, in one step, the first time.
Premieres everywhere at launch. Captioned and signed.
A real person meets an ASL Experience a business built for them, for the first time. On camera, with their consent, reaction unscripted.
Nothing needs explaining, because the moment does it. If BOS gets it right for the person every other system has gotten wrong, everything else it does becomes easy to believe.
Every participant consents before anything publishes. Capabilities shown are labelled released, validated prototype, or planned.
Built on BOS
Every department on one operating system, running on your own servers.
Explore HospitalOS AccessibilityReal-time signing avatars that let services speak ASL, reviewed by qualified Deaf ASL reviewers.
Explore AvatarOS TrustVerified supplier credentials any buyer confirms in one click, current to the minute.
Explore AcredexHow it starts
An engineer on your floor maps how work actually moves. You get a register of what your coordination failures cost, scored by department, in your own dollars. Yours to keep either way.
The auditThe agents the register ranked first, built in your environment and running in production, with your team adopting them as we go.
BOSThe same engineer stays. Two releases a month, adoption owned rather than assumed, impact proven in your numbers.
White GloveFixed-fee audit, credited in full against a deployment. Scope yours in two minutes.
Scope your auditAnswer five questions. Your audit plan builds on screen as you answer.
We email you the scoped plan with pricing and available start dates.
An engineer on your floor maps the operation. Six artefacts, yours to keep.
Want a deeper, real-time read first? Try the live diagnostic below. Or read what the three-week audit involves.
Tap any answer to change it. The plan updates as you go.
The deeper version of the five-question scoper above. Talk it through by text or voice, ten questions instead of five, and the read builds live as you answer: where work breaks down, how ready you are to orchestrate, and the first agent we would deploy. It appears on screen the moment you finish. Email is optional, and the three-week audit stays one click away.
Free. Results on screen instantly. Nothing is emailed unless you ask.
Latest
Why coordination, not automation, is the binding constraint of the agent era, and why the teams that win will be the ones that orchestrate humans, tools, data, and agents as one system.
Automation optimizes tasks. Coordination is where output is actually lost: deciding what happens next, in what order, with which context, across which people and agents. As agents multiply, the constraint moves from doing the work to routing it.
A working model of coordination cost, why point automation tends to increase it, and the design requirements for an orchestration layer that reduces it. Closes with the architecture principles behind BOS.
A governance model for agents in production: permission boundaries, human control points, and audit trails by design.
Autonomy is not a dial from zero to full. Agents should act freely inside explicit boundaries, with human control points at defined stages and every action written to a trail that can be audited after the fact.
The permission model, how control points are placed without turning every step into an approval queue, and what audit-by-design means for compliance under GDPR and PIPEDA.
Measuring what fragmentation actually costs: context loss at tool handoffs, duplicated state across five systems, and the hours managers spend reassembling a picture the organization already had.
We instrumented a 48-person company (our own) running a five-tool stack and measured where context died, how often state was duplicated, and how much time went to reassembly instead of work.
The measurement method, the headline numbers behind the 83% stack-cost reduction, and a framework for estimating the coordination tax inside your own operation.
Design principles for intelligence that expands what people are capable of rather than replacing what makes us human. The philosophical foundation everything else at GENIE AI is built on.
Humans lead. Agents support. The system orchestrates. That ordering is a design constraint, not a slogan: direction, judgment, and intent stay with people at every stage, and the system is built so that this cannot silently invert.
We eat our own cooking. Five tools ran in parallel: chat, docs, tasks, storage, and a separate AI subscription. Five bills, five contexts, and context died at every handoff. BOS replaced all five with one system and three autonomous agents: Lead Gen, Productivity, and Production.
50M+ passengers a year across 14 contracts. People management across shifts and sectors was entirely manual. Managers spent three hours a day gathering and cross-referencing staffing data across spreadsheets, with no real-time visibility. Decisions were always behind.
Operations in 42 countries. Managing projects, vendors, and teams was manual, error-prone, and slow. Compliance was hard to track, and the existing stack could not scale across geographies without adding headcount and complexity.
350 staff, volunteers, and donors. Every status update required a message; every shift required manual follow-up. Coordinators were spending the majority of their time on repetitive admin instead of mission work.
Figures are from live deployments and internal measurement. Customer identities are anonymized where disclosure has not been cleared.
BOS, the Business Autopilot, sits between the people who direct work and the systems that carry it out. You describe outcomes; BOS decomposes them into work, routes that work across humans and agents, and keeps every part synchronized against live data.
Direction, judgment, and intent stay with people at every stage. That constraint is architectural, not aspirational. See Bounded Autonomy for the governance model.
BOS Studio is the shared workspace where your team and your agents operate on the same context. It connects your existing stack, maintains a single live state across it, and turns described outcomes into executable workflows. Connect → Orchestrate → Execute is the full lifecycle.
Agent Mesh is the layer of role-scoped agents activated where they're needed: Ops, Comms, Analytics, Lead Gen, Productivity, Production, and more. Each runs inside the governance boundary of your workspace, on your live data, with its actions written to the audit trail.
Every deployment teaches BOS how work actually flows through an organization: handoffs, dependencies, exceptions. That knowledge is earned through production, not scraped from an API, and it compounds: each workflow map makes the next orchestration sharper.
Agents act within explicit permission boundaries with human control points at defined stages. Every agent action is logged to an audit trail. BOS is built for GDPR and PIPEDA compliance, with SOC 2 Type II certification in progress.
Every engagement starts the same way: a short Coordination Review that maps where work fragments in your current operations and where orchestration pays back first. Talk to our team to begin.
When work moves between a human and an agent, or between two agents, how much context survives? We're measuring loss at each handoff type and testing a shared-state protocol that keeps it at zero.
Inferring how work actually flows through an organization, not how the org chart says it does, from production telemetry. The foundation of the BOS Knowledge Graph.
Staffing decisions in high-tempo environments (airports, logistics, facilities) where the data changes faster than a human can re-plan. Trialing agent-proposed, human-approved scheduling loops.
What happens to bounded-autonomy governance when agents operate at volume? Stress-testing permission systems and audit-trail integrity under production concurrency.
The measurement study behind The Coordination Tax: instrumenting a 48-person company (ours) to quantify context loss, duplicated state, and reassembly time across a five-tool stack.
AvatarOS renders photorealistic virtual avatars that deliver real-time, grammatically correct sign-language communication alongside spoken and written service information. Streamed to any standard browser. No installation, no specialized hardware.
Real-time ASL delivery, rendered in the cloud and streamed to the browser.
Most accessibility tooling is bolted on after the fact. In BOS, inclusive interfaces are a system capability: sign language is a first-class output of the platform, delivered through the same rendering pipeline as every other channel, serving both of Canada's sign languages for the Deaf and hard-of-hearing population.
02 · How it worksThe pipeline converts service content into grammatically correct sign output, rendered on GPU-accelerated cloud infrastructure and streamed to end-user devices through a standard web browser. All compute-intensive work stays in the cloud; the end user needs nothing but a screen.
03 · Managed in BOS StudioEvery avatar persona moves through a managed lifecycle inside BOS Studio, the platform's no-code control layer, from initial design through to retirement. Every stage is version-controlled with a full audit trail, and non-technical staff can manage the entire lifecycle without engineering dependency.
Structures incoming data, validates submissions in-session, and routes applications downstream. Every intake field is reachable through an avatar-guided workflow and full keyboard navigation.
Surfaces eligibility signals, risk flags, missing documentation, and similar-case precedents. Every recommendation requires human confirmation before any action; every suggestion and decision is logged with role attribution.
Coordinates multi-step processes from intake to decision and communication, with mandatory approval gates before status transitions and real-time status visible to both citizens and staff.
Role-based access control is enforced at every layer boundary. Deployments are cloud-native and containerized, multi-cloud compatible, and designed to operate within Protected B environments, with synthetic or anonymized data used by default in testing.
Talk to our team about bringing AvatarOS to your service delivery environment.
Three weeks mapping how your company actually runs. Every workflow, handoff and system captured, every coordination failure ranked by what it costs you, and a build plan for the agents worth deploying first. Everything is yours to keep.
The audit fee is credited in full against a BOS deployment if you continue. If you don't, you still keep every report.
Not a strategy deck. We come in, watch the work, and write down what is actually happening.
A GENIE engineer joins your floor. Standups, shift changes, the handoffs nobody documents. We learn the operation before proposing anything.
Every workflow captured end to end, every handoff and re-entry logged, every coordination failure flagged in real time with the hours it burns.
Six reports, a severity-ranked register of what is costing you most, blueprints for the first agents, and a roadmap. Then you decide, with no pitch attached.
Sign, align, onboard.
Walk the floor. Hear the unsaid.
Map, rank, blueprint.
We present. You decide.
Three weeks of floor time is the difference between guessing what to automate and knowing what pays back first.
Every role, team and owner mapped, with the coordination load each one carries.
Every workflow from top level down to task level, with handoffs and re-entry marked.
Every tool honestly assessed: real usage, overlap, shadow spreadsheets, where state diverges.
Every failure logged by department and process, scored P0 to P2 against hours and dollars.
Working specifications for the first agents worth deploying, with the boundaries they run inside.
A phased plan modelled on your numbers, and the deck to take it to your board.
Nothing is submitted until you choose to. Your plan rewrites itself as you answer.
Prefer to talk it through? Take the guided version, same questions, same plan.
Tap any answer to change it. The plan updates as you go.
| Capability | Management consultancy | AI SaaS tool | Dev agency | GENIE AI |
|---|---|---|---|---|
| Understands how your work moves | Interview sample | No | Only the brief | 3 weeks on your floor |
| Ranks problems by what they cost | Sometimes | No | No | Severity-ranked register |
| Leaves you something usable | Slide deck | A subscription | Code, no map | Six artefacts, yours to keep |
| Can build what it recommends | No | Generic only | Yes, without context | BOS and agents, in production |
| Keeps humans in command | Not applicable | Opaque | Varies | Bounded autonomy by design |
You see the map of your own operation before committing to anything built on it.
Thirty minutes on how your operation runs and whether an audit is even the right move.
FreeThree weeks on the floor. Six artefacts. Yours whether or not you continue.
Fixed feeWe present to your leadership. You decide with the full picture in hand.
IncludedWe build the agents the audit identified, in your environment, in production.
Audit fee creditedBOS becomes your operating layer, retrained on your data as the work changes.
AnnualThe audit fee is credited in full against deployment if you continue. If you walk away, you keep every report and owe nothing further.
Three days gets you observations. Three weeks gets you patterns. A register built from patterns survives contact with your operation; one built from a workshop does not.
You keep all six artefacts, in full, with no restrictions. The maps describe your company, not our product. Plenty of teams use them to fix coordination without buying anything from us.
A consultancy hands you a recommendation. We hand you a map, a ranked register, and working blueprints for the agents that address the top items, and we can build them. The audit is done by the engineer who would deploy the system.
No. Interviews are scheduled around your shifts, and we observe rather than intervene. Your teams are briefed before we arrive so nobody is surprised by someone taking notes.
NDA before access. Only the systems needed for the mapping, on read-only access wherever possible. Nothing is used to train shared models. Details in our Data Policy.
A GENIE engineer, not an analyst. The same person who would lead your deployment, so nothing is lost in a handover between the audit and the build.
Tell us where you are and we will map the right first move, whether that is an audit, a single agent, or nothing at all yet.
An engineer inside your operation, shipping working agents into production every month and staying until your people actually use them. We do not hand over software and leave.
White Glove follows a Business Audit. We only embed where the map already shows what is worth building.
Not a service tier with a nicer invoice. Four structural commitments, each one costing us something real.
An engineer split six ways is nobody's engineer. We cap White Glove at four concurrent engagements, which means we turn work away to protect the ones we have taken.
The engineer who mapped your operation during the audit is the one who builds and trains on it. No discovery call twice, no context lost in a handover between a sales team and a delivery team.
Your team messages the engineer who wrote the agent, in your channel, and gets an answer the same working day. There is no tier-one triage standing between a question and the person who can answer it.
The monthly report measures how many people actually use what we built, per team, against the baseline the audit set. Shipping something nobody adopts counts as a failure on our side, not yours.
A vendor builds, demos, invoices and leaves. Six weeks later the tool is open in nobody's browser and the operation has quietly gone back to spreadsheets. The software was never the problem.
White Glove is the answer to a question we kept being asked after audits: “this is right, but who is actually going to make it happen here?”
Three things happen every month, in this order, for as long as we are engaged.
Not a ticket queue. A named GENIE engineer in your standups, on your floor or in your channels, who already knows your operation from the audit.
Two releases a month into production, built against the register the audit ranked. Nothing sits in staging waiting for a quarterly review.
Hands-on enablement, floor training, and the unglamorous follow-up that turns a deployment into how the work is done. Measured, not assumed.
Pick the next thing worth building.
Into your environment, not a demo.
Live, with the people who use it.
Before and after, in your numbers.
The same person throughout, who already carries your operation in their head from the audit.
Working agents in your environment monthly, not proofs of concept parked in staging.
Training on your floor, in your language, for the people whose work actually changes.
Adoption, hours recovered and cost against the baseline the audit established.
Agents tuned as the operation changes, so accuracy improves rather than drifting.
Your team reaches the engineer who built it, not a shared inbox and a ticket number.
We will not embed into an operation nobody has mapped. The order matters, and it protects your budget as much as our reputation.
Three weeks mapping how the work moves. Six artefacts, yours to keep.
Entry pointThe agents the audit ranked first, built into your environment and live.
BuildAn engineer stays. Two releases a month, adoption owned, impact proven.
You are hereEngagements run in three-month terms and are limited to four at a time, because an engineer who is embedded in six places is embedded in none.
No. Augmentation gives you hands to execute your plan. White Glove gives you an engineer accountable for an outcome, who decides what to build next with you and proves whether it worked.
In almost every case, yes. Without the map we would be guessing at what to build, and you would be paying us to learn your business twice. If you already have a rigorous process map, show us and we will tell you honestly whether it is enough.
Then that is our problem to solve, and it is why the impact report measures usage per team rather than licences sold. Stalled adoption usually means we built the right thing for the wrong workflow, and we fix it in the next cycle.
Both, weighted to your operation. Floor-heavy work like transit, manufacturing and facilities gets on-site time; distributed teams get the engineer in their channels and standups.
Yes, at the end of any three-month term. Everything built stays yours and keeps running in your environment. There is no clause that turns your operation off.
If you have a map already, we can talk about embedding. If you do not, start there and we will tell you whether White Glove is even warranted.
BOS connects your lab machines, biometric HR, finance and IoT devices into one system running on the hospital's own local server and database, so OPD, Emergency, Endoscopy and Pharmacy work from live data and agents handle the coordination in between.
Every question anyone ever asked became its own developer-built report. New question? New ticket.
Biometric terminals sync to one desktop PC. Shifts, schedules and payroll are stitched together by hand.
Payer mix by doctor by service takes three exports joined in a spreadsheet, then re-typed into a deck.
Everything binds to one Knowledge Graph, so a lab result, a roster and a ledger entry are connected the moment they land.
Analyzers post results straight into patient records over HL7 / ASTM. Critical values auto-flag the treating doctor. No transcription, no delay.
ZKTeco terminals sync both ways. Shifts, timetables and monthly schedules included.
Receipts, refunds, discounts and payer panels reconciled against your existing ledger.
Bedside monitors, infusion pumps, cold-chain fridges and biometric terminals on one bus. Thresholds open tasks with the right owner, instantly.
Live metrics with lineage plus forecasts that warn of tomorrow's Emergency surge and pharmacy stock-outs.
Protocols, QC sheets and studies bound to live data, so the numbers update themselves.
Enroll once, push staff and fingerprints to every ward terminal. Pull attendance automatically, every hour.
Morning, evening and overnight blocks with grace minutes and check-in windows, including 21:00 to 08:00 night duty.
Assign shifts per employee, override with temporary duties, export the department roster to XLSX in one click.
“Payer mix by doctor by service, this month?” answered where the decision is being made.
Metric id, source, refresh time and viewer scope travel with the value into documents and tasks.
No published metric, no answer. A plausible wrong number is impossible by construction.
Live queue and load metrics, doctor-wise collections, biometric staff scheduling across morning and evening shifts.
Bedside vitals on one bus, critical lab values flagged to the treating doctor, surge forecasts before the rush.
Procedure scheduling tied to staff rosters and device availability, with iWork protocols bound to live records.
Stock thresholds open reorder tasks automatically. Cold-chain fridges alert before vaccines are at risk.
Radiology is next in the module roadmap. Every department switches on in parallel with your legacy system, never as a cutover.
BOS deploys inside the hospital. Read-only connectors run against a reporting replica, never live clinical systems.
Requester, metric, filters, row count and destination logged. Exports inherit the requester's scope with no exception path.
Version 1 binds financial and operational aggregates only. Patient-level metrics require a separate security review.
A connector that would move data across a contracted boundary fails validation. Residency is recorded in configuration.
AIMS runs a mature desktop HIS, eight ZKTeco terminals and a flat catalog of hardcoded reports: the representative case for BOS Hospital Management.
Departments, devices, lab machines and the current report catalog mapped in one working session.
Read-only connectors to the HIS replica plus the device sync agent on the ward LAN.
HR fx, analytics fx, Lab and IoT switched on ward by ward, in parallel with the legacy system.
One operating system from the lab bench to the front desk, on your own servers. We run the demo against a sample hospital workspace, or an anonymized slice of your own reports if you prefer.
Name the job the way your operation talks about it, and the generator writes the use case the way an enterprise needs it: the job to be done, its trigger, what the agent does, who owns the outcome, where a human signs off, the guardrails it runs inside, and what to measure in week one. Take it to your team, or send it to us.
Most enterprise AI use cases are written as technology wishes: “use AI in finance.” They stall because nobody can build a wish. A use case that names the job, the trigger, the owner, the control point, the guardrails and the measure can be built, costed, and ranked against every other use case on the list. This is the format our engineers deploy from.
Every decision is one an enterprise has to make anyway. Make them here, in order, and watch the use case write itself. Nothing is sent unless you ask.
Pick as many as apply, then continue.
Tap any line to change it, or send the use case to yourself.
The same format, filled in by operations that now run on them.
A written use case tells you what to build. The audit tells you whether it is the first thing worth building, or the fourth.
Right here, in two minutes. Free, and yours whether or not you ever talk to us.
You are hereThree weeks mapping your operation tells you whether this use case is the first one worth building, or the fourth.
Business AuditThe agent goes into your environment, in production, with the control point you specified.
BOS deploymentMainframes gave companies computation. Personal computing gave them tools. Cloud gave them distribution. None of them ran the work itself. A Business Orchestration System does, and that is a different category of thing.
The winners of each era were not the best applications. They were whoever owned the layer everything else had to run on.
SaaS solved the tool problem so completely that it created the next one. The average company now runs on dozens of systems that each hold a piece of the truth and none of the whole. Adding another application cannot fix a problem caused by applications.
BOS does not replace your systems. It sits above them and below your intent, and it is the only thing in the stack that knows how the whole operation actually moves.
Not a database of your records: a model of how work actually moves through your company. Who hands what to whom, which step silently waits on which, where state diverges between two systems that both claim to be right.
This is earned in production, not scraped from an API. Every deployment teaches it another shape of real work, and that knowledge compounds across every deployment after it.
Takes a described outcome and decomposes it into the work that produces it, then routes each piece to whoever should do it, human or agent, and keeps every piece synchronised against live data as conditions change.
Workflow tools execute a path somebody drew. The orchestration engine works out the path, and redraws it when the operation moves.
Agents act freely inside explicit permission boundaries, with human control points at defined stages and every action written to an audit trail. The constraint is architectural, enforced in the permission model rather than promised in a policy.
It is the reason regulated operations can deploy agents at all, and the reason direction and judgement stay with people as autonomy increases.
Most software ships the same product to everyone and improves on a release cycle. An orchestration layer improves on a deployment cycle, because each new operation it maps teaches it a pattern it can recognise in the next one.
A quoting bottleneck in a cable plant and an approval bottleneck in a hospital are, structurally, the same failure. Having seen one, we arrive at the other already knowing where to look. That is a moat you cannot buy, only accumulate.
Not a company with AI features bolted onto its tools. A company whose work is held in a system that can route, execute and improve it.
By 2032, a handful of business operating systems will run most of the world’s work. BOS will be one of them.
Layers consolidate. There will not be a thousand orchestration systems any more than there were a thousand operating systems or a thousand clouds. There will be a few, chosen early, and the operations built on them will compound away from the ones that waited.
Your team, your tools, your data and your agents in one live system. You describe outcomes. BOS decomposes them into work, routes it, and keeps every part in sync.
Product, pricing and sign-up live at runbos.ai. This page is the story of what it is and why it exists.
A 48-person company running five tools in parallel: chat, docs, tasks, storage and a separate AI subscription. Five bills, five contexts, and context dying at every handoff between them.
We replaced all five with one system and three agents. BOS is not a product we imagined for other people. It is the thing we needed to run ourselves, hardened until other operations could run on it too.
A BOS for every part of your work, each with its own colour, silhouette, and role. Every one shows you exactly what it is doing, so you always know when it needs you. Tap a state to see the whole team switch.
Humans lead. Agents support. The system orchestrates.
Everything binds to one Knowledge Graph, so a task, a document, a metric and an agent action are connected the moment they exist.
Ask for a number or an outcome where the decision is being made, and get a governed answer with its lineage attached.
Living documents bound to live data, so the numbers inside them update themselves instead of going stale on write.
Pin any metric where the team already looks. One definition, one source, visible to everyone with the scope to see it.
The whole operation on one board, at the stage each piece of work actually sits, rather than in somebody's inbox.
Specialised agents activated per role, running inside the permission boundary of your workspace on your live data.
Capability switched on per department: HR, analytics, finance, lab, IoT. New surface, same graph underneath.
BOS reads your existing systems rather than replacing them. ERP, CRM, spreadsheets, devices and internal tools become one live state instead of five partial ones.
Studio decomposes what you asked for into the work that produces it, routes each piece to a person or an agent, and re-plans as conditions move.
Agents carry the work inside explicit boundaries, with human control points at defined stages and every action written to the audit trail.
The reason regulated operations can run agents at all is that the limits are architectural. Permissions are enforced at every layer boundary, not described in a policy document.
Sign up, see pricing, and start a workspace. If you would rather we mapped your operation before you switch anything on, start with the audit instead.
Not sure which? The audit tells you what to build first. runbos.ai lets you start building today.
Acredex · A GENIE AI product
Every audit you passed. Every registration you kept current. Every standard you met when nobody was watching. Acredex turns all of it into one live credential that answers for you, everywhere, the second anyone asks. Accredited. Instantly.
Northwind Analytics Inc.
Government of Canada supplier
Sample credential. Demo data.
What we believe
A week in procurement
She runs a fourteen-person analytics firm. The bid is ready. Her proof of standing is a certificate she downloaded in March, attached for the ninth time this year.
He is the procurement officer. Every PDF needs confirming against the registry. Some answers come back today. Some come back after the deadline.
Their certificate still says active. Nobody knows yet. The paperwork is accurate about the past and silent about today.
Now take the week back.
One line of code, and her live badge sits on her website, her RFP, and her procurement profile. She walks into every bid already believed.
Thirty-eight suppliers, thirty-eight clicks, under a second each. His morning goes to choosing the best bid, not chasing paperwork.
When their registry status changed, every badge they had changed with it. Revoked, everywhere, before anyone could rely on it.
What you get back
Try it
One credential, from issue to audit
Claim your registration once. Three automated checks run and your credential issues.
Show your standing on your website, in RFPs, and on procurement profiles.
Buyers click the badge and see live status. No login, no phone call.
When registration status changes, every badge changes with it.
Anyone can check a credential without having to trust Acredex.
Pick your side of the table
The way governments verify suppliers is about to change. Be on the side of it that moves first. Acredex is in operational-test preparation under the Innovative Solutions Canada Testing Stream.
Performance figures are design targets for the Acredex MVP, to be independently validated during operational testing. Compliance references describe the standards Acredex is designed to; formal assessments are in progress. Northwind, Priya and Daniel are illustrative. Built in Canada by GENIE AI Inc.
Releases, product updates, and company notes from the lab. Press inquiries: hello@genieai.io.
One system for people, tools, data, and agents. Early access begins with a free Coordination Review.
TORONTO, ON · Genie AI today opened early access to BOS, the Business Autopilot: a unifying layer that runs a company's people, tools, data, and AI agents as one synchronized system. Teams describe outcomes; BOS decomposes them into work, routes it across humans and agents, and keeps every part synchronized against live data.
BOS enters early access with production deployments already running, including a major Canadian international airport where daily staffing analysis went from three hours per manager to thirty seconds, and Genie AI's own 48-person operation, which replaced five tools with one system at 83% lower cost.
"Automation optimizes tasks. Coordination is where output is actually lost," the company said. "BOS exists so that direction, judgment, and intent stay with people while the system carries the coordination load."
Early access begins with a Coordination Review: a short mapping of where work fragments in a team's current operations and where orchestration pays back first.
About Genie AI. Genie AI is an applied intelligence lab founded in 2023 in Toronto, building intelligent systems on one principle: humans lead, agents support, systems orchestrate. Genie AI participates in Google for Startups, Microsoft for Startups, NVIDIA Inception, and AWS Startups. Press: hello@genieai.io
Every connected tool now reads and writes one live state, and Ops and Comms agents are available to all early-access workspaces.
BOS Studio now maintains a single live state across every connected tool, so decisions are made on current information rather than the last export. Alongside it, two Agent Mesh activations moved from field trial to general early access: the Ops Agent, for shift-based and real-time operations, and the Comms Agent, for status flows that previously ran on manual follow-up. Both operate inside the standard bounded-autonomy governance: explicit permissions, human control points, full audit trail.
Conversations at the edge of human potential, with founders, scientists, and operators building intelligent work.
The first three episodes are in production, covering the orchestration thesis, running a 40-person company like a 400-person company, and why judgment is the last human monopoly. Launching soon; write to hello@genieai.io with the subject "Notify me" to hear first.
Two agent activations replaced an $11,112-per-month stack, freeing coordinators for mission work.
TORONTO, ON · A 350-person nonprofit running staff, volunteers, and donor programs on BOS reduced its coordination stack from $11,112 to $198 per month, with Comms and Ops agent activations absorbing the status updates and shift follow-ups that previously consumed the majority of coordinator time. Figures are from live deployment measurement; the organization's identity is published with permission where cleared.
Program support across the stack BOS is built and scaled on.
Genie AI was accepted into four builder programs through 2025, supporting infrastructure, model access, and go-to-market as BOS moved from prototype to production deployments.
Global policy, with Canada, EEA and UK, and United States addenda
Version 3.0
Effective upon January 1, 2023
Last updated August 8, 2026
Applies to genieai.io, runbos.ai, the BOS platform and all Genie AI products and services
Privacy Officer privacy@genieai.ca
This policy explains what Genie AI collects, why we collect it, who we share it with, how long we keep it, and the rights you can exercise. Where Genie AI processes personal information on behalf of a business customer, the customer agreement and Data Processing Addendum govern, and this policy is descriptive only.
GENIE AI, Inc. ("Genie AI", "we", "us", "our") is a corporation incorporated federally in Canada under the Canada Business Corporations Act, with its principal place of business in Ontario, Canada. We build BOS, the Business Autopilot, an AI native platform that orchestrates people and AI agents to run business operations, together with the vertical solutions and applications built on it.
This Privacy Policy applies to personal information we handle through our websites, applications, platform, agents, APIs, events, support channels, sales and marketing activities, and any other product, service, program or feature we provide, online or offline (together, the "Services").
This policy does not apply to third party products or websites that link to or integrate with our Services, or to the internal privacy practices of our customers. Those are governed by their own policies.
Genie AI handles personal information in two distinct roles. Understanding which one applies to you determines how this policy operates.
+--------------+-------------------------------+-----------------------+ | Role | When it applies | What governs | +==============+===============================+=======================+ | Controller | We decide why and how | This Privacy Policy. | | | information is processed. | | | ("or | This covers visitors to our | | | ganization" | websites, prospects and | | | under | marketing contacts, users who | | | Canadian | create accounts directly with | | | law) | us, billing contacts, | | | | applicants, and partners. | | +--------------+-------------------------------+-----------------------+ | Processor | A business, enterprise or | The customer | | | government customer uses BOS | agreement and our | | ("service | and submits information about | Data Processing | | provider" | its own employees, | Addendum. The | | or | contractors, customers or | customer is the | | "agent") | citizens. We process that | controller and is | | | Customer Content only on the | responsible for | | | customer's documented | notice and consent to | | | instructions. | its own individuals. | +--------------+-------------------------------+-----------------------+
If you are an individual whose information was placed into BOS by an organization, please direct access, correction and deletion requests to that organization. We will support them in responding, and we will refer your request to them if you contact us directly.
By using our Services you accept this policy. If you do not agree with it, please do not use the Services. If anything here is unclear, contact our Privacy Officer before you continue.
We collect only what we need for the purposes described in this policy. The categories below describe what we collect as a controller.
-------------------------------------------------------------------------
---------------- --------------------------------------- ---------------- Identity and First and last name, business email, You, directly. contact mailing address, phone number, job title, employer, preferred language.
Account and Username, hashed credentials, single You, or your access sign on identifiers, multi factor administrator. enrolment, role and permission assignments, organization membership.
Billing and Billing contact and address, plan and You, and payment transaction licence tier, invoice and payment processors. history, tax identifiers, purchase orders. Card numbers are handled by our payment processors and are not stored on our systems.
Usage and Features used, agents invoked, workflow Automatically. telemetry and task events, session duration, performance and error data, API call metadata.
Device and IP address, browser and device type, Automatically. network operating system, language, referring page, approximate region derived from IP.
Support and Messages, tickets, call and meeting You, directly. communications notes, feedback, survey responses, recordings where you are notified and consent applies.
Marketing and Subscription preferences, campaign You, and event engagement, event registration and partners. attendance, content downloads.
Recruitment Resume, work history, references, right You, and to work confirmation, assessment recruiters. results, where you apply for a role.
Customer Content Whatever a customer chooses to place Our customers. into BOS. This may include personal information about that customer's own people. We do not control what is submitted. -------------------------------------------------------------------------
We do not seek sensitive personal information for our own purposes, and we ask that you do not send it to us in support tickets, emails or free text fields. Where a customer configures BOS to process sensitive categories, that processing happens under the customer agreement, on the customer's instructions, and with the additional controls described in section 9.
Our Services are built for business, enterprise and government use. They are not directed to children, and we do not knowingly collect personal information from anyone under the age of majority in their province, state or country. If we learn we have collected such information without valid consent, we will delete it promptly. Contact our Privacy Officer if you believe a child has provided us with personal information.
We use personal information only for the purposes identified when it was collected, for purposes a reasonable person would consider appropriate in the circumstances, or as otherwise permitted or required by law.
----------------------------------------------------------------------- Purpose Legal basis in the Canadian consent EEA and UK model --------------------------- --------------------- --------------------- Provide, operate, secure Performance of a Express or implied and support the Services, contract. consent at the point including provisioning of collection. accounts and running agents and workflows.
Bill, collect payment, Contract and legal Express consent and manage licences, and meet obligation. legal requirement. tax and accounting obligations.
Maintain, troubleshoot, Legitimate interests. Implied consent, monitor, and improve consistent with reliability, quality and reasonable performance of the expectations. Services.
Protect against fraud, Legitimate interests Permitted use and abuse, unauthorized access and legal obligation. implied consent. and security incidents, and enforce our terms.
Develop new features, Legitimate interests. Implied consent for models, integrations and non personal and de products, using Service identified data. Data and de identified or aggregated data.
Send service and Contract and Implied consent, and administrative notices, legitimate interests. required notice. including security, availability and policy change notices.
Send marketing Consent, or Express or implied communications, legitimate interests consent under CASL, invitations, newsletters for existing business with unsubscribe in and offers. relationships. every message.
Respond to lawful demands Legal obligation. Permitted disclosure from law enforcement, without consent where regulators, courts and required by law. government authorities.
Evaluate candidates and Steps prior to Express consent. manage recruitment. entering a contract and legitimate interests.
Support corporate Legitimate interests. Permitted use, with transactions, financing, the safeguards in due diligence, and business section 8. continuity. -----------------------------------------------------------------------
We may use and disclose non personal information, including aggregated and de identified data, for any lawful business purpose. Where we de identify data we maintain the technical and organizational measures to keep it de identified, and we do not attempt to re identify it.
If we intend to use personal information for a purpose that was not identified when we collected it and that you would not reasonably expect, we will identify the new purpose and obtain your consent first, unless the law permits or requires us to proceed without it.
We obtain express consent where the information is sensitive, where the purpose is not obvious, or where the law requires it. We rely on implied consent only where the information is less sensitive, the purpose is clear from the context, and your reasonable expectations support it.
You can refuse to provide personal information, and you can withdraw consent at any time, subject to legal and contractual restrictions and reasonable notice. Withdrawing consent may mean we can no longer provide part or all of the Services to you. Withdrawal does not affect processing that already occurred, and it does not apply to information we are required to retain by law.
To withdraw consent, contact privacy@genieai.ca, or use the unsubscribe link in any marketing message. Service and security notices are not marketing and will continue while your account is active.
We use cookies, local storage, pixels, web beacons and similar technologies on our websites and, in limited form, in the product.
------------------------------------------------------------------------ Type What it does Can you turn it off ---------------- ------------------------------------ ------------------ Strictly Authentication, session integrity, No. The Services necessary load balancing, security and fraud will not function prevention. without these.
Functional Remembers language, region, Yes, with reduced interface and workspace preferences. convenience.
Analytics and Measures traffic, feature adoption, Yes. performance errors and page performance so we can improve the Services.
Marketing and Measures campaign performance and Yes. attribution attributes sign ups to sources. ------------------------------------------------------------------------
Where required by law we present a consent banner and set non essential cookies only after you opt in. You can also control cookies through your browser or device settings, and you can withdraw consent at any time through the cookie preferences link on our website. Blocking cookies may break parts of the Services.
We use third party analytics providers, including Google Analytics. These providers may collect IP address, timestamps, referring pages and return visitor status, and process that information under their own privacy terms. We configure IP anonymization and data retention limits where the provider supports them.
We honour Global Privacy Control and similar recognized opt out preference signals where applicable law requires it. Beyond that, our Services do not currently alter behaviour in response to browser do not track signals, because no common standard for them has been adopted.
BOS orchestrates AI agents that read, generate, route and act on information. Because that is the core of what we build, we set out our commitments explicitly.
Prompts and outputs may be retained for a limited period to deliver the feature, support debugging, maintain audit trails, and detect abuse. Retention periods are set out in section 11 and can be shortened by contract for enterprise and government deployments.
AI generated output can be incomplete, out of date or wrong. Outputs are decision support, not professional advice, and should be reviewed by a person before they are relied on for legal, financial, medical, employment or safety related decisions. Customers configure the approval gates, escalation paths and human in the loop checkpoints appropriate to their use case, and are responsible for doing so.
Genie AI does not use automated decision making, including profiling, to make decisions about individuals that produce legal or similarly significant effects on them, without human involvement. Where a customer configures BOS to support such decisions in its own operations, the customer is responsible for the lawful basis, the disclosure, and the individual's right to an explanation and to human review. On request we will provide customers with information about the principal factors and parameters a configured workflow uses, so they can meet those obligations, including under Quebec's Law 25 and Articles 13, 14 and 22 of the GDPR.
Our terms prohibit using the Services to conduct unlawful surveillance, to infer sensitive characteristics for discriminatory purposes, to generate biometric identification of individuals without a lawful basis, or to make consequential decisions about people without meaningful human review.
We do not sell your personal information, and we do not rent it, trade it, or share it for cross context behavioural advertising.
We disclose personal information only in the circumstances below.
We engage sub processors in the categories below. The current list of named sub processors, their function and their processing location is published at genieai.io/subprocessors and is updated when it changes.
------------------------------------------------------------------------ Category Function Typical processing region ----------------------- -------------------------------- --------------- Cloud infrastructure Compute, storage, networking, Canada, United and hosting managed databases. States
AI model and inference Language model inference and United States, providers embedding generation. Canada
Identity and access Authentication, single sign on, United States multi factor enrolment.
Payments and billing Payment processing, invoicing, United States, tax calculation. Canada
Communications Transactional email, United States notifications, in product messaging.
Support and ticketing Customer support, knowledge United States, base, session diagnostics. Canada
Analytics and Product analytics, error United States, monitoring tracking, uptime and performance European Union monitoring.
Business operations CRM, contract execution, United States, accounting, collaboration Canada tooling. ------------------------------------------------------------------------
Enterprise and government customers can subscribe to advance notice of new sub processors and, where their agreement provides for it, object on reasonable data protection grounds.
We maintain an information security program with administrative, technical and physical safeguards proportionate to the sensitivity of the information we hold. Our program is aligned to recognized frameworks, and our current certification and attestation status is available on request under NDA.
We maintain a documented incident response plan with defined severity levels, escalation paths and post incident review. If a breach of security safeguards occurs:
No method of transmission or storage is completely secure. We work hard to protect your information but cannot guarantee absolute security. If you suspect unauthorized access to your account or a security issue in our Services, contact security@genieai.ca immediately. We operate a responsible disclosure process and will not pursue good faith security research conducted within its terms.
Genie AI is based in Canada. Personal information we collect may be stored and processed in Canada, the United States, and other jurisdictions where we or our sub processors operate. Those jurisdictions may have data protection laws that differ from those where you live, and information located there may be accessible to local courts, law enforcement and national security authorities under the laws of that jurisdiction.
We take the following steps to protect information that crosses borders.
Data residency options are available for enterprise and government deployments, including Canadian region hosting for Canadian public sector workloads. Residency commitments are made in the customer agreement, not by this policy. Contact privacy@genieai.ca to confirm what is available for a specific deployment.
We keep personal information only as long as it is needed for the purposes it was collected for, or as long as required by law, contract, or the resolution of a dispute. When the retention period ends we delete the information, or irreversibly de identify or anonymize it. Our default schedule follows. Customer agreements may specify shorter or longer periods, and where they do, they prevail.
-----------------------------------------------------------------------
--------------------------- ------------------------------------------- Customer Content in BOS For the term of the agreement. On termination, available for export for 30 days, then deleted from production within 90 days.
Account and profile data For the term of the agreement, plus 12 months, unless earlier deletion is requested.
Prompt and output logs 12 months, configurable down to 30 days for enterprise and government deployments.
Billing, invoicing and tax 7 years from the end of the relevant fiscal records year, to meet Canadian tax and corporate record requirements.
Support tickets and 24 months from closure. correspondence
Security, audit and access 12 to 24 months, depending on system and logs regulatory requirement.
Website analytics data 14 months.
Marketing contacts Until consent is withdrawn, or after 24 months of no engagement, whichever comes first. Unsubscribe records are kept indefinitely so we do not contact you again.
Recruitment records 12 months after the competition closes, or longer with the candidate's consent.
Encrypted backups Rolling 35 days, after which deleted records age out. Deletion requests are applied to production immediately and to backups as they cycle. -----------------------------------------------------------------------
Subject to the exceptions in section 12.3, you can exercise the following rights over personal information we hold about you as a controller.
-----------------------------------------------------------------------
---------------- ------------------------------------------------------ Access Obtain confirmation that we hold personal information about you, a copy of it, and an account of the parties to whom, and the circumstances in which, it has been disclosed.
Correction Have inaccurate or incomplete information corrected or completed. If we decline, we annotate the record to show a correction was requested and not made.
Deletion Have information deleted where it is no longer needed, where consent is withdrawn and no other basis applies, or where required by law.
Portability Receive computerized personal information you provided to us in a structured, commonly used technological format, and have it transmitted to another organization where technically feasible.
Withdraw consent Withdraw consent at any time, on reasonable notice, subject to legal and contractual restrictions.
Object and Object to processing based on legitimate interests, or restrict ask us to restrict processing while a dispute about accuracy or legal basis is resolved.
De indexing Ask us to cease disseminating information, or de index a link, where dissemination contravenes the law or a court order, or causes serious injury to reputation or privacy.
Human review Ask for human review of, and an explanation of the factors behind, a decision based exclusively on automated processing.
Non Exercise your rights without being denied service, discrimination charged a different price, or given a lower quality of service.
Complain Lodge a complaint with your privacy regulator. Our response will tell you which one and how to reach it. -----------------------------------------------------------------------
Send your request to privacy@genieai.ca with the subject line Privacy Request. Include enough detail for us to locate the information and verify your identity. We may ask for proof of identity proportionate to the sensitivity of the information requested, and we use that proof only to verify the request. An authorized agent may act for you with written authorization and proof of their own identity.
We may withhold information, in whole or in part, where the law permits or requires it, including where:
If we refuse, we will tell you the reasons, the provision of the applicable legislation we rely on, how to reach the Privacy Officer to discuss it, and how to ask for a review by the relevant regulator.
The following terms supplement this policy for individuals in the identified regions. Where an addendum conflicts with the main body, the addendum prevails for those individuals.
We handle personal information in accordance with the Personal Information Protection and Electronic Documents Act, Quebec's Law 25 and the Act respecting the protection of personal information in the private sector, and the substantially similar provincial legislation of Alberta and British Columbia.
This section applies to residents of California, Colorado, Connecticut, Virginia, Utah, Texas and other states with comprehensive privacy legislation, to the extent that legislation applies to us.
Our Services link to and integrate with third party sites, platforms and applications, including social media, identity providers and business tools that customers connect to BOS. When you follow a link or authorize an integration, that third party collects and handles information under its own privacy policy and security practices. We do not control those practices. Review them before connecting anything, and be aware that social media plug ins can connect your browser directly to the provider's servers when the page loads.
We may update this policy as our Services, our legal obligations, or our practices change. The date at the top shows when it was last revised. If a change materially affects your rights, we will provide notice by posting it prominently on our website, by email, or both, before it takes effect. Continued use of the Services after a change takes effect means you accept it. We maintain prior versions and will provide one on request.
Direct any question, concern, access request or complaint to our Privacy Officer. We take every complaint seriously, investigate it, and respond in writing with the outcome and the steps you can take if you remain unsatisfied.
-----------------------------------------------------------------------
--------------------------- ------------------------------------------- Privacy questions, requests privacy@genieai.ca, attention Privacy and complaints Officer
Security issues and security@genieai.ca vulnerability reports
General customer service hello@genieai.io
Mail GENIE AI, Inc., attention Privacy Officer, Suite 400, 77 King Street West, Toronto, Ontario M5K 2A1, Canada -----------------------------------------------------------------------
GENIE AI, Inc. Federally incorporated in Canada under the Canada Business Corporations Act. This policy is published in English. Une version francaise est disponible sur demande.
Master terms governing the BOS platform, websites and services
Version 1.0
Effective August 8, 2026
Applies to genieai.io, runbos.ai, BOS, BOS Studio, and all Genie AI products, agents and APIs
Contact legal@genieai.ca
Read these Terms carefully. They include a limitation of liability, a disclaimer of warranties, and provisions governing how disputes are resolved. By creating an account, accessing the Services, or signing an Order Form that references these Terms, you agree to be bound by them.
These Terms of Use (the "Terms") form a binding agreement between GENIE AI, Inc., a corporation incorporated federally in Canada under the Canada Business Corporations Act ("Genie AI", "we", "us", "our"), and the individual or entity accessing the Services ("you", "Customer").
You accept these Terms by creating an account, accessing or using the Services, signing an Order Form that references them, or clicking to accept. If you accept on behalf of an entity, you represent that you have authority to bind that entity, and "you" means that entity. You must be of the age of majority in your jurisdiction and legally capable of entering into a contract.
Where documents conflict, the following order applies, from highest to lowest:
These Terms do not modify a signed enterprise agreement. Where no signed agreement exists, these Terms govern in full.
BOS is an AI native platform that orchestrates people and AI agents to run business operations. We provide the Services on a subscription basis in accordance with the Order Form, these Terms and the Documentation.
We may add, modify, deprecate or remove features. Where a change materially reduces the core functionality you subscribed to, we will provide at least 30 days notice, and if the change materially and adversely affects you, you may terminate the affected subscription and receive a pro rata refund of prepaid fees for the unused remainder of the term. Beta, preview and pilot features are excluded from this commitment.
Unless the Order Form states otherwise, BOS is licensed per business rather than per seat. A licence entitles a single legal entity to use the Services for its own internal business operations, for the number of workspaces, environments, agents, workflows, actions or volume limits stated in the Order Form. A licence may not be shared across affiliates, resold, sublicensed, or used to provide services to third parties, except as expressly permitted in the Order Form.
Subject to your compliance with these Terms and payment of applicable fees, we grant you a non exclusive, non transferable, non sublicensable, revocable right during the term to access and use the Services for your internal business purposes, within the scope on your Order Form.
You may not, and may not permit anyone else to:
You must not use the Services, and must not permit your Users to use them, to:
> If you become aware of a violation on your account, stop it and tell > us at abuse@genieai.ca. We may investigate suspected violations and > may suspend access under section 16.2.
As between the parties, you own and retain all right, title and interest in Customer Content. We claim no ownership of it.
You grant us a worldwide, non exclusive, royalty free licence to host, copy, transmit, display, index, embed, process and otherwise use Customer Content solely to provide, secure, maintain and support the Services for you, to comply with law, and as otherwise permitted in the Data Policy. This licence ends when the Customer Content is deleted, except for copies retained in backups until they cycle out and copies we are required by law to retain.
You are solely responsible for Customer Content, for its accuracy, quality and legality, and for having all rights, consents, notices and lawful bases needed for us to process it as contemplated by these Terms. You are responsible for determining whether the Services are appropriate for your data and your regulatory environment, and for configuring retention, approvals and access controls accordingly.
We maintain backups as described in the Data Policy, but you remain responsible for maintaining your own copies of Customer Content that is important to you. We are not a system of record for your statutory retention obligations unless the Order Form says otherwise.
As between the parties, and subject to your payment of fees and compliance with these Terms, you own the Outputs generated for you. We assign to you whatever rights we may have in those Outputs. Because generative systems produce similar responses to similar inputs, Outputs are not unique, similar or identical Output may be generated for other customers, and we retain the right to produce and use such Output. Outputs may not be protectable by copyright in some jurisdictions.
Outputs are probabilistic. They can be incomplete, out of date, biased or factually wrong, and they may describe actions that were not taken or facts that do not exist. Outputs are decision support and not professional advice, and they are not a substitute for legal, financial, tax, medical, engineering, or other qualified judgement. You must evaluate Outputs for accuracy and suitability before relying on them or acting on them.
BOS can be configured to take actions in connected systems, including sending messages, creating records, moving work, and calling third party APIs. You control that configuration. You are responsible for the scope of permissions you grant to agents, for the approval gates and human in the loop checkpoints you set, and for the consequences of actions taken within the permissions you configured. We recommend least privilege scoping, staged rollout, and human approval for any action that is irreversible, financially material, externally visible, or legally significant.
Full detail on how data moves through the platform, including logging, indexing, embeddings and agent memory, is set out in the Data Policy.
The Services connect to third party products, including cloud platforms, model providers, communication tools, commerce platforms and business systems. When you enable an integration, you authorize us to access and exchange Customer Content with that third party as needed to operate the integration, and you authorize that exchange under your agreement with the third party.
Third party services are governed by their own terms and privacy policies. We do not control them, we do not warrant them, and we are not responsible for their acts, omissions, availability, pricing changes, or handling of your data. If a third party changes or discontinues its service, we may need to modify or discontinue the related feature, and that will not by itself be a breach of these Terms.
We may offer free trials, pilots, proofs of concept, and beta, preview or early access features. These are provided as is and as available, without any warranty, service level, support commitment, or indemnity, and may be modified, suspended or discontinued at any time. Our aggregate liability arising from free or beta access is limited to one hundred Canadian dollars.
Do not use beta features for production workloads or with sensitive data. Beta features may be subject to additional terms, and information about them is our Confidential Information.
Unless the Order Form states otherwise, trial data may be deleted at the end of the trial period, and we will give you a reasonable opportunity to export it first.
We and our licensors own all right, title and interest in the Services, including the underlying software, models we develop, orchestration logic, architecture, user interfaces, Documentation, and all improvements to them, together with all associated intellectual property rights. No rights are granted other than the limited licence in section 5. All rights not expressly granted are reserved.
If you give us feedback, ideas, or suggestions, you grant us a perpetual, irrevocable, worldwide, royalty free licence to use them without restriction or obligation to you. We will not identify you as the source without your consent.
We may use Service Data, and aggregated or de identified data derived from your use, to operate, secure, analyze and improve the Services, and to produce statistics and benchmarks, provided that we do not identify you or your Users, and do not disclose Customer Content.
We may identify you as a customer and use your name and logo on our website, in investor materials and in sales materials, in accordance with any brand guidelines you provide. You may withdraw that permission at any time by writing to legal@genieai.ca, and we will stop using your marks in new materials within 30 days. Case studies, quotes and detailed descriptions of your deployment require your prior written approval.
"Confidential Information" means non public information disclosed by one party to the other that is designated confidential or that a reasonable person would understand to be confidential. Customer Content is your Confidential Information. The Services, pricing, roadmap, architecture, security documentation and beta features are ours.
The receiving party will use the same degree of care it uses for its own confidential information, and no less than reasonable care, will use Confidential Information only to perform under these Terms, and will limit access to personnel and advisors who need it and are bound by confidentiality obligations at least as protective.
Confidential Information does not include information that is or becomes public through no fault of the recipient, was rightfully known without restriction before disclosure, is rightfully received from a third party without restriction, or is independently developed without use of the disclosing party's Confidential Information.
A party may disclose Confidential Information where legally compelled, provided it gives prompt notice where lawful and reasonably cooperates in seeking protective treatment. These obligations survive for three years after termination, and indefinitely for trade secrets and Customer Content.
Our handling of personal information is described in the Genie AI Privacy Policy, and the operational detail of how data moves through the platform is set out in the Genie AI Data Policy. Both are incorporated into these Terms by reference.
Where we process personal information on your behalf, you are the controller and we are the processor, and our Data Processing Addendum applies. The DPA governs in the event of a conflict with these Terms in respect of personal information. Enterprise, regulated and public sector customers should execute the DPA before submitting personal information to the Services.
Support scope, response targets and any service level commitment are set out in your Order Form or the applicable support and service level documentation. Self serve plans receive commercially reasonable support through published channels, without a service level commitment.
We aim for high availability but do not warrant uninterrupted or error free operation. Scheduled maintenance, emergency maintenance, third party outages, force majeure events and suspensions permitted under these Terms are excluded from any availability measurement.
These Terms begin when you first accept them and continue until all subscriptions have expired or been terminated.
We may suspend your access, in whole or in part, immediately and without liability where we reasonably determine that: your use poses a security risk to the Services or to another party; your use may harm our systems or subject us to liability; your use violates section 6; your account is more than 15 days overdue after the notice in section 10; or suspension is required by law. We will give notice where practicable, limit the suspension to what is necessary, and restore access promptly once the cause is resolved.
On termination, your right to access the Services ends and all outstanding fees become immediately due. You may export Customer Content for 30 days after termination using the available export tools. After that period we will delete Customer Content from production systems within 90 days, and from backups as they cycle, in accordance with the Data Policy, unless retention is required by law. Sections 2, 7.1, 8.1, 8.2, 10, 12, 13, 17, 18, 19, 21 and 22 survive termination.
We warrant that the Services will perform materially in accordance with the Documentation during the term, and that we will provide them with reasonable skill and care. Your exclusive remedy for breach of this warranty is that we will use commercially reasonable efforts to correct the non conformity, and if we cannot do so within a reasonable time, you may terminate the affected subscription and receive a pro rata refund of prepaid fees for the unused remainder of the term.
Each party warrants that it has the authority to enter into these Terms and will comply with laws applicable to its performance.
> Except as expressly stated in this section, the Services, Outputs > and all related materials are provided on an as is and as available > basis, without warranty of any kind.
To the maximum extent permitted by law, we disclaim all other warranties, express, implied, statutory or otherwise, including implied warranties of merchantability, fitness for a particular purpose, title, non infringement, quiet enjoyment, accuracy, and any warranty arising from course of dealing or usage of trade. We do not warrant that the Services will be uninterrupted, secure, or error free, that defects will be corrected, that Outputs will be accurate, complete, current, lawful or fit for your purpose, or that the Services will meet your requirements or produce any particular business result. Some jurisdictions do not allow the exclusion of certain warranties, so parts of this section may not apply to you.
We will defend you against any third party claim alleging that the Services, when used in accordance with these Terms and the Documentation, infringe that third party's Canadian, United States or European Union patent, copyright or trade mark rights, and we will pay damages finally awarded or amounts we agree in settlement. This obligation does not apply to claims arising from Customer Content, Outputs, your use in breach of these Terms, combination of the Services with anything not provided by us where the claim would not have arisen otherwise, modifications not made by us, use of a superseded version after we provided a replacement, beta or free features, or third party services.
If the Services become, or we believe may become, the subject of an infringement claim, we may at our option procure the right to continue use, replace or modify the Services to make them non infringing, or terminate the affected subscription and refund prepaid fees for the unused remainder of the term. This section states our entire liability and your exclusive remedy for intellectual property infringement.
You will defend us against any third party claim arising from Customer Content, your Outputs and how you use them, your breach of section 6, your violation of law or third party rights, or your failure to obtain required consents, and you will pay damages finally awarded or amounts you agree in settlement.
The indemnified party must promptly notify the indemnifying party of the claim, give it sole control of the defence and settlement, and provide reasonable cooperation at the indemnifying party's expense. No settlement that admits liability or imposes obligations on the indemnified party may be made without its consent, not to be unreasonably withheld. Late notice reduces the indemnity only to the extent of resulting prejudice.
To the maximum extent permitted by law, and except for the Excluded Claims below:
"Excluded Claims" means your payment obligations, either party's indemnification obligations under section 18, breach of section 13, your breach of sections 5 or 6, and liability that cannot be limited or excluded by law, including fraud, fraudulent misrepresentation, death or personal injury caused by negligence, and any liability under applicable consumer protection legislation that cannot be waived.
You acknowledge that Outputs are probabilistic and that you are responsible for reviewing them and configuring appropriate human oversight. We are not liable for decisions made, or actions taken by configured agents within permissions you granted, in reliance on Outputs. These limitations apply regardless of the theory of liability and survive any failure of essential purpose of a limited remedy. The allocation of risk in this section is a fundamental basis of the bargain and is reflected in our pricing.
These Terms are governed by the laws of the Province of Ontario and the federal laws of Canada applicable in that province, without regard to conflict of laws rules. The United Nations Convention on Contracts for the International Sale of Goods does not apply.
The parties will first attempt to resolve any dispute through good faith discussion between senior representatives for 30 days after written notice of the dispute.
If the dispute is not resolved, the parties irrevocably submit to the exclusive jurisdiction of the courts of the Province of Ontario, sitting in Toronto, and waive any objection based on venue or forum non conveniens. Nothing in this section prevents either party from seeking injunctive or equitable relief in any court of competent jurisdiction to protect its intellectual property or Confidential Information.
Disputes are resolved on an individual basis. To the maximum extent permitted by law, each party waives any right to bring or participate in a class, collective or representative proceeding. Nothing in these Terms limits any non waivable right you may have under the consumer protection laws of your province or country.
-----------------------------------------------------------------------
--------------------------- ------------------------------------------- Contracts, legal notices legal@genieai.ca and these Terms
Privacy requests and privacy@genieai.ca, attention Privacy complaints Officer
Security issues and security@genieai.ca vulnerability reports
Abuse and acceptable use abuse@genieai.ca reports
General customer service hello@genieai.io
Mail GENIE AI, Inc., attention Legal, Suite 400, 77 King Street West, Toronto, Ontario M5K 2A1, Canada -----------------------------------------------------------------------
GENIE AI, Inc. Federally incorporated in Canada under the Canada Business Corporations Act. Version 1.0, effective August 8, 2026.
How data moves through BOS: classification, AI handling, residency, retention and control
Version 1.0
Effective August 8, 2026
Applies to BOS, BOS Studio, agents, APIs and all Genie AI products
Audience Customers, security reviewers, privacy officers and procurement teams
Contact privacy@genieai.ca
This Data Policy is the operational companion to the Genie AI Privacy Policy and Terms of Use. The Privacy Policy explains our obligations to individuals. This document explains, at a system level, what data BOS holds, where it goes, who can reach it, how long it stays, and what you control. It is incorporated into the Terms of Use by reference.
Genie AI builds BOS, an AI native platform that orchestrates people and AI agents to run business operations. Because agents read, generate and act on customer data, we publish this policy so customers can evaluate the platform without a discovery call.
This policy applies to all Genie AI products, environments and APIs. It describes our default posture. Enterprise, regulated and public sector agreements may set stricter terms, and where they do, the agreement prevails.
-----------------------------------------------------------------------
------------------------ ---------------------------------------------- Privacy Policy What personal information we handle, why, and what rights individuals have.
Terms of Use The commercial and legal relationship, licence scope, liability and acceptable use.
Data Policy (this How data physically and logically moves document) through the platform, and what you control.
Data Processing Addendum The binding controller to processor terms, including transfer mechanisms and audit rights.
Security Addendum and Control detail, certifications, penetration Trust Centre test summaries and architecture diagrams, available under NDA. -----------------------------------------------------------------------
We classify everything the platform touches into five categories. The category determines how the data is handled, who can access it, and how long it is kept.
------------------------------------------------------------------------
-------------- --------------------------- ----------------------------- Customer Data you submit, connect or Documents, records, prompts, Content generate inside your files, workflow definitions, workspace. connected system data, agent outputs.
Account Data Information identifying you Names, business emails, and your users, and roles, workspace administering your configuration, billing subscription. contacts, licence entitlements.
Service Data Machine generated data Event logs, agent invocation produced by operating the records, latency and error platform. telemetry, API metadata, audit trails, security logs.
Derived Data Data computed from Customer Embeddings, vector indexes, Content to make the Knowledge Graph nodes and platform work. edges, classifications, summaries, agent memory.
Genie AI Data Our own material. Platform code, orchestration logic, prompt libraries, models we develop, documentation, aggregate statistics. ------------------------------------------------------------------------
> Derived Data is the class most often missed in vendor reviews. It is > generated from your content, it is treated as your Confidential > Information, it lives inside your tenant, and it is deleted when the > source content is deleted.
------------------------------------------------------------------------
-------------- ----------------------- --------------------------------- Customer You. Host, process and index it solely Content to provide, secure and support the Services for you, on your instructions.
Derived Data You, as an extension of Use it inside your tenant to Customer Content. power retrieval, orchestration and agent behaviour. Not used for other customers.
Account Data You provide it. We act Provision access, bill, support, as controller for communicate, and meet legal administering the obligations. relationship.
Service Data Genie AI. Operate, secure, troubleshoot, measure and improve the Services. May be aggregated. Not disclosed in a form that identifies you or exposes Customer Content.
Genie AI Data Genie AI. All rights reserved. Licensed to you only as set out in the Terms of Use. ------------------------------------------------------------------------
Every piece of Customer Content follows the same path. You control the entry point, the configuration, and the exit.
------------------------------------------------------------------------
------------ --------------------------------- ------------------------- Ingestion Content enters through upload, Which sources are direct entry, API, or an connected, which scopes integration you authorize. It is are granted, and what is encrypted in transit and written uploaded. to your tenant.
Indexing Content may be chunked, embedded Which repositories and and written to a vector index and workspaces are indexed, to the Knowledge Graph so agents and whether indexing is can retrieve it. enabled per source.
Processing Agents read relevant content, Agent scope, tool call a model, and produce an permissions, model Output. Relevant context is selection where offered, assembled per request, not sent and redaction rules. in bulk.
Action Where configured, an agent writes Approval gates, human in back to a connected system or the loop checkpoints, and notifies a person. least privilege scoping of every connector.
Logging The request, decision path and Log retention window, and outcome are recorded for audit, whether prompt and output debugging and abuse detection. bodies are retained or only metadata.
Retention Content and Derived Data persist Workspace and source until deleted by you, or until level retention settings, the retention period ends. and manual deletion at any time.
Deletion Deletion removes the record, its When to delete, and embeddings, its index entries and whether to export first. its graph nodes from production, then ages out of backups. ------------------------------------------------------------------------
BOS is built from six nodes. Each handles data differently, so we describe them separately.
-----------------------------------------------------------------------
---------------- ------------------------------------------------------ iWork Holds work items, tasks, assignments and status. Stores Customer Content and Account Data. Visible to workspace members according to their role.
BOS Chat Conversational interface. Prompts and responses are Customer Content. Conversation history is scoped to the user and workspace, and is subject to the retention window you set.
Project Fx Project and execution structures, plans, dependencies and artefacts. Stores Customer Content.
Agent Mesh Runs agents. Assembles only the context needed for each request, calls the selected model, and records the invocation. Agents operate within the tool permissions you grant, and cannot reach outside your tenant.
Orchestration Routes work between people and agents. Holds routing Engine rules, approval gates, and execution state. Generates Service Data.
Knowledge Graph Stores entities, relationships and embeddings derived from your content. Derived Data, scoped to your tenant, deleted when the source is deleted. -----------------------------------------------------------------------
Vertical solutions built on BOS, including AvatarOS, CommerceOS, AutoOS and Acredex, inherit this model. Where a vertical introduces an additional data flow, for example commerce order data or generated media assets, it is documented in that product's supplementary data sheet.
This is the section most security reviews focus on, so we state the position plainly.
For each agent invocation we send only the assembled context needed for that request, which typically includes the user's instruction, retrieved excerpts from your content, relevant workspace configuration, and tool definitions. We do not send your full corpus, your user directory, or your billing information. Where redaction or field masking is configured, it is applied before the request leaves our systems.
------------------------------------------------------------------------
retention** --------------------------- ------------------ ------------------------- Prompt and output bodies 12 months Yes, down to 30 days, or metadata only, for enterprise and government deployments.
Invocation metadata (agent, 12 to 24 months Partially, subject to model, timestamp, latency, audit and security token counts, outcome) requirements.
Tool and action audit trail 24 months Extendable where records management rules require it.
Abuse and safety signals 12 months No, retained for platform integrity. ------------------------------------------------------------------------
Agents can retain memory across sessions where you enable it. Memory is Derived Data, scoped to the workspace or user you assign it to, visible to workspace administrators, editable and deletable at any time, and removed when the workspace is deleted. Memory is off by default for new workspaces in regulated configurations.
We do not routinely read Customer Content. Limited human review of specific records occurs only in the circumstances described in section 8, and every instance is logged.
Our personnel do not have routine access to Customer Content. Access is possible only in the four circumstances below, is granted on a least privilege and time limited basis, and is logged with the actor, the reason, the scope and the duration.
All personnel are subject to confidentiality agreements, background screening where permitted by law, mandatory privacy and security training at onboarding and annually, and immediate access revocation on role change or departure. Enterprise customers can request access logs relating to their tenant.
Genie AI is a Canadian company. By default, Customer Content is stored in Canadian or United States cloud regions, and processing may occur in either. Some sub processors, including certain model providers and analytics tools, process data in the United States or the European Union.
We publish the current list of sub processors, including function and processing region, at genieai.io/subprocessors. Every sub processor is subject to security and privacy due diligence before engagement and periodically afterwards, and is bound by written terms that restrict processing to our instructions, impose confidentiality and security obligations at least as protective as ours, and prohibit independent use of Customer Content.
Enterprise and public sector customers can subscribe to advance notice of new sub processors and, where their agreement provides, object on reasonable data protection grounds. We remain accountable for our sub processors' performance.
The full control set is described in the Privacy Policy and the Security Addendum. In summary:
Certification and attestation status, architecture diagrams and test summaries are available under NDA. Contact security@genieai.ca.
Default retention periods are set out in the Privacy Policy and summarized below. Customer agreements may set shorter or longer periods and prevail where they do.
-----------------------------------------------------------------------
--------------------------- ------------------------------------------- Customer Content For the term. On termination, exportable for 30 days, then deleted from production within 90 days.
Derived Data (embeddings, Deleted together with the source content, graph nodes, agent memory) or when the workspace is deleted.
Prompt and output logs 12 months, configurable down to 30 days.
Account Data Term plus 12 months, unless earlier deletion is requested.
Security, audit and access 12 to 24 months. logs
Billing and tax records 7 years, to meet Canadian tax and corporate record requirements.
Backups Rolling 35 days. -----------------------------------------------------------------------
You can export your data at any time during the term, and for 30 days after termination, using the export tools available in the product and through the API. Exports are provided in structured, commonly used, machine readable formats. Where an export requires engineering assistance, we will provide reasonable help and will tell you in advance if a fee applies. We do not hold your data hostage as a commercial lever, and export availability is not contingent on the reason for termination, subject only to payment of undisputed amounts already due.
We use aggregated and irreversibly de identified data to measure quality, tune orchestration, benchmark performance and improve the Services. When we de identify data we maintain technical and organizational measures to keep it de identified, we do not attempt to re identify it, and we contractually prohibit recipients from doing so. Aggregate statistics we publish never identify a customer, a user, or the content of any workspace.
Unless we have agreed in writing to support the category and the appropriate addendum is in place, do not submit to the Services:
If sensitive data is submitted by mistake, delete it from the workspace and contact privacy@genieai.ca so we can help confirm removal from indexes and logs.
Platform controls only work when they are configured. You are responsible for:
We update this policy as the platform and our obligations evolve. Material changes that reduce protections will be notified to customers at least 30 days before they take effect, and prior versions are available on request.
-----------------------------------------------------------------------
--------------------------- ------------------------------------------- Data handling questions and privacy@genieai.ca, attention Privacy privacy requests Officer
Security documentation, security@genieai.ca questionnaires and vulnerability reports
Contracts, DPA and audit legal@genieai.ca requests
General customer service hello@genieai.io
Mail GENIE AI, Inc., attention Privacy Officer, Suite 400, 77 King Street West, Toronto, Ontario M5K 2A1, Canada -----------------------------------------------------------------------
GENIE AI, Inc. Federally incorporated in Canada under the Canada Business Corporations Act. Version 1.0, effective August 8, 2026. Read with the Genie AI Privacy Policy and Terms of Use.
Founded 2023. Built on the belief that intelligent systems should expand what humans are capable of, not replace what makes us human.
We think the greatest transformation in human history is underway, and that it either elevates every person or a privileged few. We are building for the first outcome, one operation at a time, with humans in command of every one of them.
The golden age doesn’t build itself. Join the labs.
Request received. A confirmation is on its way to . We’ll reply within two business days with next steps, starting with a short Coordination Review.