Automation that survives production
Most AI pilots die between the demo and the real world. We build the unglamorous parts that decide whether an agent survives: approval gates, evaluation against labelled data, error and retry policy, cost accounting, and somewhere in Canada to run it.
Past no-code.
Short of a $60k platform.
No-code tools are cheap and genuinely good at moving data between apps. They stop at judgement — reading a document, deciding what it means, handling the exception. Enterprise agent platforms do that, but the tier where it actually works starts around CAD $60,000 a year before anyone implements anything, and processes your data on infrastructure you don't choose.
That gap is where most mid-market companies in Canada and the United States actually live. It's the only place we work.
You should not hire us if a $30/month tool already solves your problem. We will tell you that on the first call, for free, and we would rather lose the engagement than sell you something you don't need.
Fixed scope. Fixed price. Written exclusions.
Every engagement here has a published price, a stated timeline, and an explicit list of what is not included. These six are the build engagements; the ongoing ones are below them. Nothing starts without that in writing. An advisory hour is CAD $195; fixed-scope advisory engagements start at CAD $4,500; minimum build engagement is CAD $9,500. Advisory time is CAD $195 an hour, booked in single hours and credited against a Blueprint or Assessment booked within 30 days — the same rate applies to work outside an agreed scope, though we would rather re-scope than bill it. Prices in Canadian dollars, verified 23 August 2026.
Automation Blueprint
$4,500CAD · Advisory only, no code · Not credited — take the Assessment if you intend to buildIncluded
- Target architecture for one automation, written down
- Feasibility checked against your actual systems and data
- Cost model: build, run, and cost per task at your volume
- Named alternatives, including buying an off-the-shelf tool
- A written recommendation you can take to any vendor
Not included
- Any code, prototype or integration
- Credit against a later build — that is the Assessment
Automation Readiness Assessment
$12,500CAD · Focused (1–2 processes) · Enterprise-wide from $16,500 · Credited in full against a build within 90 daysIncluded
- Process inventory across one department
- Measured task volume and fully-loaded cost per task
- Top 3 candidates ranked by payback period
- Build-vs-buy recommendation naming real alternatives
- Data-residency and privacy memo
- 90-minute walkthrough with your team
Not included
- Any production build
- System integration work
Fixed-Scope Agent Pilot
$22,000CAD · Single integration · Multi-integration + hardening to $55,000 · 40/40/20 milestonesIncluded
- One workflow automated end to end, in production
- Human approval gate before any irreversible action
- Evaluation harness scored against a held-out labelled set
- Per-task cost accounting
- Error, retry and fallback policy
- Runbook and full source handover — you own it
Not included
- Additional workflows (priced separately)
- Ongoing hosting — see Care below
Product Build
from $55,000CAD · A working product with its own users, billing and accounts · Source handed overIncluded
- A web application your customers sign in to, not an internal script
- Accounts, roles and permissions built to your rules
- Subscription billing wired end to end — proration, tax, failed payments
- The core feature set agreed and written down before development starts
- Full source, infrastructure definitions and runbook
Not included
- Native iOS or Android apps — priced separately
- Ongoing operation — that is Resident Care
Platform Build
from $120,000CAD · Several workflows under one approval policy · Scoped after an AssessmentIncluded
- A platform across several workflows, not a single automation
- Your own data model, authentication and role-based access
- Multiple agents coordinating under one approval policy
- Integration with your existing systems of record
- Staged delivery — a working system at the end of each stage
- Full source, infrastructure definitions and runbook
Not included
- A fixed total before scoping — $120,000 is the floor, not the quote
- Ongoing operation — that is Resident Care
Engine Licence
from $50,000CAD one-time, plus from $2,000 CAD / month · You licence the engine; we keep maintaining itIncluded
- Access to an engine we already built and run in production
- Implementation of the interface and integration around it
- Versioned API, changelog, and notice before anything breaking
- We keep the engine current as sources and standards change
- Exit terms written down before you sign
Not included
- Ownership of the engine — this is a licence, not a transfer
- Exclusivity, unless separately agreed and priced
After launch: keep it running, keep building, or both.
Keeping a system running and continuing to build on it are separate jobs, priced separately, and you can buy either without the other. Every tier exits on 60 days notice. Prices in Canadian dollars, verified 23 August 2026.
Resident Care & Hosting
from $2,500CAD / month · Essential $2,500 · Operate $5,000 · Assured $9,000 · LLM inference billed at costIncluded
- Hosting in a Canadian or U.S. region, or your own infrastructure
- Monitoring, alerting and incident response
- Model and cost optimisation as pricing shifts
- Monthly accuracy review against your labelled set
- A named contact — the person who built it
Not included
- New feature development (quoted separately)
- Third-party model API costs (billed at cost)
Continuous Build
from $5,500CAD / month · A queue of feature work, one task at a time · Not hostingIncluded
- A standing queue of feature work, delivered in order
- Two to four business days per task, typically
- The same person who built the system
- Re-scored against your labelled set as features land
- Pause or leave on 60 days notice
Not included
- Parallel workstreams — the queue is sequential by design
- Hosting and incident response — that is Resident Care
Priority Build
from $12,000CAD / month · Front of the queue, and architecture time with it · Not hostingIncluded
- Your work goes to the front of the queue, ahead of Continuous Build
- Larger features, not just sequential single tasks
- Standing architecture time — decisions reviewed before they are built
- A staging environment kept current for your team to try things on
- Named contact, same-day acknowledgement on weekdays
Not included
- A guaranteed ship date on unscoped work — we scope first, always
- Hosting and incident response — that is Resident Care
Where the volume actually is
Five industries we know the paperwork of
The engineering is the same. What differs is which documents matter, which rules must hold, and what a regulator or an insurer will ask you afterwards.
Build vs buy — our actual take
We are not the right answer for every problem. Here is where we genuinely think each option wins, including the ones that cost us the engagement.
We also implement on these platforms. If the honest answer for you is n8n or Copilot Studio, we'll build it there and charge less, because it takes us fewer hours.
| Option | Use it when | Don't when |
|---|---|---|
| Zapier / Make | Simple triggers between SaaS apps, low volume, no judgement required. | You need reasoning over documents, or volume makes per-task pricing hurt. |
| n8n (self-hosted) | You have technical staff, want to own the runtime, and need Canadian hosting cheaply. | Nobody in-house will maintain it. An unmaintained workflow engine is a liability. |
| Microsoft Copilot Studio | You're deep in Microsoft 365 and your processes live in Teams, SharePoint and Dataverse. | Your systems are outside the Microsoft estate, or you need custom model routing. |
| Enterprise agent platforms | You have the budget, the volume and a procurement team that wants a big vendor's certifications. | You're mid-market. Entry tiers are too small and the next tier is a five-figure annual commitment. |
| Offshore AI agency | You want the largest possible engineering team per dollar, you have someone in-house to manage scope, and your data carries no residency or sector obligation. | Your data must stay in a named jurisdiction, your contract has to be enforceable where you are, or you need working hours that overlap yours. Check the vendor's registered office, not the flag on the marketing site. |
| Us | Volume is real, judgement is required, data residency matters, and you want one accountable builder. | A $30/month tool solves it. We'll tell you on the first call. |
How we measure accuracy
Most AI automation accuracy figures are self-graded: the system writes its own success criteria, then scores itself against them. The number goes in a slide and nobody can reproduce it. We do it the way it's done in engineering.
Why it matters commercially
A held-out evaluation set is the difference between "the vendor says it's 98% accurate" and "we measured 94% on 200 of our own invoices, and here are the 12 it got wrong." The second one survives a procurement review. The first one doesn't.
It also gives you a real decision: at your measured accuracy and your volume, does the exception queue cost less than the manual process? Sometimes the answer is no — and you should know that before you spend.
You help label the set
Before we build, you help us label a held-out set of real cases from your own operation — including the awkward ones.
Held out during the build
The agent never sees them during development.
Scored against that set
Accuracy is measured against that set, and the set stays yours.
Re-run every month
We re-run it every month under a Care retainer.
Questions worth asking first
Each answer restates what is set out above — what an engagement costs, when you should not hire us, how accuracy is judged, and where the work runs.
What does an engagement cost, and is there a minimum?
Every engagement on this page has a published price, a stated timeline, and an explicit list of what is not included — nothing starts without that in writing. An advisory hour is CAD $195; fixed-scope advisory engagements start at CAD $4,500; minimum build engagement is CAD $9,500. The two-week Automation Readiness Assessment is $12,500 focused on one to two processes, credited in full against a build within 90 days; the fixed-scope agent pilot is $22,000 for a single integration. Prices in Canadian dollars, verified 23 August 2026.
What if a $30/month tool already solves our problem?
Then you should not hire us. We will tell you that on the first call, for free, and we would rather lose the engagement than sell you something you don't need. We also implement on these platforms — if the honest answer for you is n8n or Copilot Studio, we'll build it there and charge less, because it takes us fewer hours.
How do you measure accuracy?
Not by self-grading. Before we build, you help us label a held-out set of real cases from your own operation, including the awkward ones. The agent never sees them during development, accuracy is measured against that set, the set stays yours, and we re-run it every month under a Care retainer.
Where does it run, and do we get the source?
Hosting is in a Canadian or U.S. region, or on your own infrastructure. The Canadian regions are AWS ca-central-1 (Montreal) or Azure Canada Central (Toronto); we can also build inside your own cloud tenancy, where you hold the keys, the logs and the bill, or fully on-premise on your own hardware. The region is selected with you and named in the contract. The pilot includes a runbook and full source handover — you own it.
Are you SOC 2 or ISO 27001 certified?
No. Quintessentia Network Inc. is not SOC 2 certified and is not ISO 27001 certified. We are a small, owner-operated Canadian firm serving clients in Canada and the United States. If your procurement process requires either certification from every vendor, we will not pass it today — and we would rather you learn that here than three meetings in. The trust page states it plainly, along with what we offer instead.
Start with the assessment
Two weeks, fixed price, and half of it comes back if you go on to a pilot. You'll get a ranked list of what's worth automating — and an honest answer if the answer is "nothing yet."
Book a scoping call