AI systems built around your operations.
Custom AI for contracts, field data, finance and logistics. Run it on your own hardware or in a dedicated Canadian cloud. You own everything we build.
Canadian-hosted, air-gapped option, you own the code.
Daily progress
Aggregated from daily reports and sensor exports
Progress stalled for roughly 18 hours on days 4 to 5, then resumed below the planned rate. Flagged for an engineer to review.
- Police reportPresent
- PhotosPresent
- Repair quotePresent
- Policy activeMissing
The policy lapsed on Mar 2 and the loss date is Mar 5. Summary and policy excerpt attached for the adjuster.
Northline Industrial Ltd.
34 pages · 4 clauses extracted
7.1 Payment terms
Invoices are payable within sixty (60) days of receipt by the Customer, without set-off.
12.3 Term and renewal
This Agreement renews automatically for successive twelve (12) month terms unless either party gives written notice at least ninety (90) days before the end of the then-current term.
18.2 Limitation of liability
Each party's aggregate liability under this Agreement shall not exceed the fees paid in the preceding twelve (12) months.
22.1 Governing law
This Agreement is governed by the laws of the Province of Alberta and the federal laws of Canada applicable therein.
Liability is limited to 12 months of fees. Your standard terms ask for 2x annual fees, so a large loss would be mostly uncovered.
Propose a cap of 2x annual fees, with carve-outs for data breach.
Built for the work your industry actually does.
We build AI around the work your team already does: your documents, your systems, your rules. Then we hand over the code, the model files and the keys.
See how it could work in your business.
Interface concepts with sample data. Click around. Every real project is built around your own systems.
The choice is yours. We build it.
On a server at your site, or on a dedicated Canadian-region cloud server. Same engineering, either way.
| On-premise | Private cloud | |
|---|---|---|
| Data custody | Runs on a server at your site. Nothing needs to leave your network. | Runs on a dedicated server in a Canadian region, in your own cloud account or one we set up for you. |
| Cost model | Runs on a server you own, or one we help you set up. Hardware cost depends on the models you need. | No hardware to buy. Monthly hosting cost, scoped to your usage. |
| Per-query cost | No per-token API bills when using local models. Hardware and power are your costs. | Depends on users, data volume and which model you choose. |
| Maintenance | Needs a physical server, power and cooling at your site. Updates are applied on a schedule we agree on. | We maintain it with you. No hardware to own or depreciate. |
| Internet needed | Local models work without an internet connection. | Needs an internet connection to your Canadian-region cloud server. |
| Latency | Responses stay on your local network. | Depends on your connection and the model used. |
| Best for | Sensitive documents, remote sites, and teams with strict data rules. | Multi-site teams and fast rollouts. |
We work with open-weight models that run on your own hardware, and with commercial models where your data rules allow it. The right mix depends on the task, and we say so plainly when a smaller local model is not enough.
What you own.
Everything we build for you is yours at handover. No licence to renew, no black box.
The code
Every connector, pipeline and interface we write is handed over, with the documentation your team needs to run it.
The model files
Open-weight models and any fine-tuned weights sit on your hardware or in your cloud account, not ours.
The configuration
Deployment settings, prompts and integration setup are delivered alongside the code.
No lock-in
Moving it in-house or to another vendor is a planned option, not a rebuild. Hosting fees on private cloud continue while you run it there; the software is still yours.
From audit to production in six weeks.
Fixed milestones, founders doing the engineering, no vendor lock-in.
Week 1 · Plan
Workflow audit
Scoping operational bottlenecks, security boundaries, and raw data silos.
Deliverable: Architecture blueprintWeek 2 · Design
Security topology
Architecting on-premise hardware specs vs. private cloud data isolation.
Deliverable: Infrastructure contractWeeks 3–4 · Build
Pipeline synthesis
Coding custom ingestion connectors and fine-tuning model weights.
Deliverable: Staging prototypeWeeks 5–6 · Deploy
Production cutover
Planned rollout with full code and configuration handover.
Deliverable: Complete IP transfer
Also from SieveAI
opAI, in development.
Our own product for oil and gas engineers: an offline workspace with an AI assistant that does the clicking. Ask for the analysis, and watch it work.
opAI
Fitting a hyperbolic decline to the best well in the group.
What does SieveAI build?
Custom AI systems fitted to your documents, data and workflows: contract review, field reporting, invoice audit, claims intake, support drafting and similar. Every project is built for one organization, not resold as a product.
Should we run on-premise or in the cloud?
On-premise keeps data in your physical custody and works without internet. Private cloud is a dedicated, single-tenant server in a Canadian region with nothing to maintain. We recommend one during scoping based on your data rules, sites and budget.
Who owns the code and the models?
You do. The code, configuration and model files are handed over at the end of the project. There is no lock-in.
How long does a project take?
Our plan runs six weeks from workflow audit to production cutover. We confirm timing against your systems and data during scoping.
Is our data used to train models?
No. Your data is not used to train any model.
Do you work outside Alberta?
We are based in Edmonton and work with organizations across Canada.
What does it cost?
Every project is custom, so there is no list price. After a short scoping call we propose fixed milestones and a cost for each.
Tell us what slows your team down.
We will tell you honestly whether software can help, and whether it should run on your hardware or in a Canadian cloud.