The shapes of work we ship. No client names.
No logos wall, no case-study deck. Client work is confidential by default, so we publish the shape of each pattern instead.
So we publish what we can: each pattern as we scope it, tagged by service group — bottleneck, solution, how it may be funded, measurable impact. Sector and revenue band only. If one looks like a problem on your desk, that’s the conversation to start.
Want to see what a finished deliverable looks like? Read the sample AI Roadmap we wrote for a fictional property manager.
- Typical AI build
- 2–8 weeks
- Price
- Fixed, agreed before kickoff
- Training and handoff
- Included in every build
- Post-launch support
- 30 days, every build
Pull structured data out of long, messy documents.
The bottleneck
A regulated intake team re-keying 38 fields out of 12-page contracts by hand — multi-hour reviews, fatigue errors late in the day, and a backlog that grew faster than headcount could clear it.
How it may be funded
Invoiced line by line, so the client’s accountant can decide whether any part counts as capital investment under the Productivity Super-Deduction. Much of a build may simply be a regular business expense.
Itemized invoices; your accountant’s callThe solution
An extractor that reads each document, scores how sure it is of every field, and sends anything uncertain to a person. It reads and drafts, and never writes back to your systems without a person’s sign-off. It can run on Canadian cloud regions where the documents call for it, and is tested against the client’s real historical documents before production.
The measurable impact Illustrative
On the illustrative test set below: per-document review time falls from hours to roughly 90 seconds, field-level accuracy lands in the low-to-mid 90s, and only the edge cases reach a person.
- Field accuracy
- 93–96%
- Time per document
- ~90 s
- Sent to a person
- 5–12%
- Test set size
- 1,200
All four figures are illustrative of the pattern, not drawn from a single named engagement.
Five more patterns.
-
Classification & routing for high-volume intake.
Bottleneck. A logistics operator hand-triaging thousands of calls, forms, and emails into the wrong queues.
Solution. A classifier that sorts each item into a fixed set of categories for a person to act on, and flags anything it’s unsure about. Read-only; Canadian hosting where it matters.
-
Internal knowledge search grounded in your real documents.
Bottleneck. A professional-services firm where senior staff fielded the same policy questions all week — answers buried across folders, tickets, and heads.
Solution. A read-only assistant that answers only from the client’s own documents, cites its sources, and hands off to a person rather than guessing when it isn’t sure. Canadian hosting where it matters.
-
Workflow automation across the tools you already pay for.
Bottleneck. A wholesale-distribution back office shuttling data between email, docs, a CRM, Slack, and the spreadsheet nobody admits is load-bearing.
Solution. Workflows built in n8n, Make or code, with AI steps only where they earn it. Nothing is sent outside without a person’s approval, and there’s an off switch and an audit log. Canadian hosting where it matters.
Itemized invoices; your accountant’s call -
Hands-on AI literacy on your team’s real work.
Bottleneck. A finance-sector team using consumer AI tools ad hoc — no shared rules, real exposure to leaking data into prompts they didn’t control.
Solution. A full-day workshop, then a six-week enablement program, on the client’s own tools and workflows — safe, PIPEDA-aware prompting, practiced on real artifacts, not toy prompts.
-
Websites and small web apps, ready for AI later.
Bottleneck. A services firm whose site and internal tools couldn’t host the automations they wanted next — everything bolted on after launch.
Solution. Fast sites in the client’s own repo and domain, built to target WCAG 2.1 AA and structured to take an assistant or search later without a rebuild — real components, real states.
Every AI system runs on one security stack.
The AI builds above aren’t secured case by case — they sit on one architecture, the four layers below. It’s what keeps every system limited to reading and drafting, with a person approving what it does, and hosted in Canada where it matters, without a custom security story each time. Workshops and websites don’t need it; anything with a model in it does.
Canadian Compute (Data Residency)
Canadian cloud regions where your data calls for it.
Bounded Autonomy (Read-Only by Default)
It can read and draft, but can’t change anything without a person’s sign-off.
Human-in-the-Loop (The Approval Gate)
The AI proposes; your team executes.
Recoverability (The Kill Switch)
Audit logs that can’t be altered, and an instant switch back to manual.
A note on discretion.
Systems like these sit inside operations, presales, legal, and support stacks — places a company doesn’t advertise because the competitive edge lives there. We’d rather have the long, referral-driven relationship than the screenshot.
What’s here is what we’re free to publish: the shape of the work, how it’s financed, and the rules we run it by. Instead of client screenshots, we show the work in two ways you can check: the illustrative demo on our homepage and a full sample AI Roadmap.
If one of these resembles your problem, let’s talk.
A free 30-minute call. We read your brief before we meet. If it’s not a good fit, we’ll point you toward someone better — and probably name them.