OCTION LABS

Sovereign AI for regulated Canada.

We build the AI stack that public-sector and regulated buyers are legally allowed to use: hardware we own, data they keep, compliance built in. Phase 0 is live.

Oction AgencySovereign AI Infrastructure

OCTION LABS

THE THESIS

We own the three things every other AI vendor only rents: the hardware, the data lifecycle, and the compliance framework.

Regulated and public-sector clients cannot legally hand citizen and patient data to OpenAI, Anthropic, or a US cloud. They still need AI. We are the only model built to sell to them on infrastructure we control, in province, with an audit trail.

OCTION LABS01 / 14

OCTION LABS

THE PROBLEM

Regulated industries want AI and cannot legally buy what is on the market.

Municipal governmentprovincial data-sovereignty rules restrict cloud LLMs for citizen data.
HealthcarePHIPA, HIA, Law 25, and PIPEDA push patient data toward local processing; the US CLOUD Act makes cross-border cloud a documented risk.
Finance and legalSOC 2 and client-confidentiality rules block general SaaS AI tools.
Public safetyNIST 800-53 mandates controlled supply chains and audit trails.

The gap: every vendor pitches "secure." None can prove sovereign infrastructure. Buyers want AI but cannot legally buy what is available.

OCTION LABS02 / 14

OCTION LABS

WHAT WE DO

Three models on one vertically integrated stack. Each makes the next more defensible.

Phase 0, Sovereign Compute Pilot (live): grant-eligible proof of concept. Portable compute deployed on client sites to validate data sovereignty, a NIST 800-53 baseline, and the procurement path before production scale.
Model 1, Sovereign AI Deployment (operating): bespoke software plus managed on-premise infrastructure for regulated clients. We own the hardware. Clients own the data. Compliance is built in, not bolted on.
Model 2, Proprietary Data Assets (scaling): anonymized, permissioned operational data from Model 1 becomes licensable training material and enrichment APIs.
Model 3, Sovereign Compute Hubs (government-targeted): inference and HPC clusters for government co-location, on multi-year contracts. Physical ownership of the stack is the switching cost.

One line: we build the only AI stack regulated buyers are legally allowed to use.

OCTION LABS03 / 14

OCTION LABS

WHY NOW

Three forces open a 24 to 36 month window.

Regulatory urgencyNIST 800-53 Rev. 5, provincial AI procurement rules, and EU AI Act spillover into Canadian standards are forcing buyers to act.
Sovereignty pressurefederal strategy increasingly favors domestic AI scale-ups; public-sector buyers are actively moving off pure US cloud dependence.
Talent arbitrageCalgary and Edmonton AI ecosystems (Amii, University of Alberta) are under-recruited relative to Toronto and Vancouver.

Miss the window and incumbents (Deloitte, CGI, IBM) fill it with overpriced, underperforming solutions.

Note: prior versions cited a specific power-draw multiplier for the compute hardware. We removed it pending a measured benchmark rather than ship an unverifiable claim.

OCTION LABS04 / 14

OCTION LABS

WHERE THE DEMAND IS

Four segments where regulated buyers need AI and have defined procurement paths to buy it.

Canadian municipal services

Predictable procurement, defined compliance, citizen-data rules

Regulated enterprise (health, finance, legal)

High-value, sticky workloads blocked from public cloud AI

Sovereign data assets

Permissioned operational data that cannot leave the premise

Government compute and grants

Public investment moving toward domestic, sovereign infrastructure

The demand signal is public: Canada Health Infoway's AI Scribe Program is the largest federally funded health AI deployment in Canadian history (10,554 clinicians enrolled, 708,654 encounters by Aug-Sep 2025). Regulated sectors are moving on AI now, and they need a stack they are legally allowed to use.

Sources: Canada Health Infoway (2025).

OCTION LABS05 / 14

OCTION LABS

TRACTION

Built with no outside capital. This is a deployed systems company, not a science project.

Phase 0 liveCyntech Group geotechnical knowledge assistant running on Oction's on-premise stack in a regulated Canadian enterprise.
Autonomous agent mesh operating on dedicated hardware around the clock (Julian, Lucius, Atlas, and specialist agents).
30,000+ indexed knowledge chunks across a triple-layer RAG/CAG/KAG memory system with hybrid search.
Live infrastructurelocal LLM proxy, memory API, vector store, and cache, all healthy.
NIST 800-53 security framework implemented, documented, and audit-ready.
Active buyer pipelinemultiple Alberta municipal conversations and two systems-integrator relationships at pilot stage.

Honest framing: conversations and pilots are in progress; no production enterprise contract is signed yet. We do not claim revenue we do not have.

OCTION LABS06 / 14

OCTION LABS

THE DATA FLYWHEEL

Every Model 1 client makes Model 2 more valuable. Margins expand as we scale.

Deployeach client generates anonymized, permissioned operational data.
Encodepatterns become reusable knowledge graphs, templates, and playbooks.
Modularizeproven infrastructure configs become repeatable deployment modules.
Licenseanonymized datasets and tuned weights become proprietary, licensable assets.

Year 1 looks like services. Year 3 looks like software plus data. The flywheel is the path from project margins to platform margins.

OCTION LABS07 / 14

OCTION LABS

COMPETITIVE LANDSCAPE

Compliance fluency times hardware control times data ownership equals no direct competitor.

Deloitte / CGI / Accenture

Their weaknessHigh hourly labor, long sales cycles

Our advantageFixed-price deliverables, short pilots

OpenAI / Anthropic / US cloud

Their weaknessCannot sign in-province data-residency on owned hardware

Our advantageHardware we own, in province, auditable

Local MSPs

Their weaknessNo AI depth, no knowledge graph

Our advantagePurpose-built agent mesh plus memory stack

Internal IT teams

Their weaknessNo budget, talent, or compliance fluency

Our advantageWe are the outsourced sovereign-AI department

OCTION LABS08 / 14

OCTION LABS

TWO VERTICALS, ONE STACK

The same sovereign stack serves municipal government and healthcare. Healthcare is the higher-value, higher-defensibility wedge.

Municipalpredictable procurement, defined compliance, faster first contracts. The proving ground.
Healthcarelarger contracts, deeper switching costs, strong non-dilutive funding (Infoway, Vector, CIHR, IRAP), and public outcome proof (CHARTWatch cut unanticipated in-hospital deaths 26%, peer-reviewed in CMAJ, Sept 2024).
The healthcare partnership and pilot motion runs on a separate deck. The point: one stack serves both verticals, so we are not dependent on any single segment.

Sources: CMAJ / Unity Health Toronto (2024).

OCTION LABS09 / 14

OCTION LABS

TEAM

A systems and infrastructure company run by operators, multiplied by an always-on agent staff.

Brandon Gill

RoleCo-Founder / CEO

FocusSales, partnerships, capital strategy, government alignment

MJ Dewji

RoleCo-Founder

FocusStrategic oversight, capital deployment, long-term positioning

Leonel Garcia

RolePartner / COO

FocusOperations; healthcare delivery lead; clinical workflow validation and health-authority relationships

Shamir Cheema

RolePartner / CRO

FocusRevenue operations and commercial strategy

Augustin Gentile

RolePartner / Business Development

FocusPartnerships and revenue-generating deals

AI TEAM autonomous agent mesh

Julian Systems & infrastructureLucius Operations intelligenceAtlas Security & forensicsAvery ResearchKai DataOscar ContentFinn Real estate

OCTION LABS10 / 14

OCTION LABS

WHY WE WIN - THE MOAT

The moat is structural, not feature-based.

Hardware we own becomes the client's switching cost; the compute does not leave.
Compliance fluency (NIST 800-53, provincial health and privacy law) is a sales gate competitors cannot fake.
The data flywheel compoundseach deployment makes the next cheaper to deliver and the data layer more valuable.
Canadian-built and in-province is a procurement requirement, not a marketing line, for our buyers.

Cloud AI vendors cannot follow us here without abandoning the cloud economics that define them.

OCTION LABS11 / 14

OCTION LABS

THE BET

Regulated AI is not a feature. It is a market, and it is up for grabs in the next 24 months.

We own the hardware. We own the data lifecycle. We own the compliance framework. Every competitor rents one of the three.

The companies that solve regulated, sovereign AI in the next two years will define the next decade of enterprise and public-sector infrastructure in Canada. We intend to be one of them.

OCTION LABS12 / 14

OCTION LABS

WHERE WE ARE HEADED

From first deployments to a sovereign AI platform Canada owns.

Near termproduction deployments with municipal and regulated-enterprise clients, building on the live Phase 0 pilot.
Buildinga proprietary, permissioned data layer that compounds in value with every deployment.
Long termsovereign compute hubs that keep Canadian inference, data, and compliance in country.

Each step makes the next more defensible. The direction is one stack that regulated Canada can actually use, own, and trust.

OCTION LABS13 / 14

OCTION LABS

CONTACT

Want to know more about what we are building? Talk to us.

Brandon Gill, Co-Founder & CEO: brandon@octionagency.com

Leonel Garcia, Partner & COO: leonel@octionagency.com

Shamir Cheema, Partner & CRO: shamir@octionagency.com

More detail available on request: NIST 800-53 control mapping; sovereign compute hardware specifications; agent mesh architecture; knowledge system technical overview; municipal procurement playbook.

OCTION LABS14 / 14