AI Company Brain
Ask your company a question. Get the answer, with a source.
Every SOP, policy, price list and past quote in one place your team can just ask — in plain English. New hires stop interrupting your best people, and your best people stop being the bottleneck.
For businesses with real internal knowledge and turnover or growth pain: clinics, firms, contractors with crews, hospitality groups, anyone onboarding seasonal staff.
In plain English
What you get.
Answers with a receipt
Every answer names the document it came from, so nobody has to take the AI’s word for it.
One place to ask
In a browser, or right inside Slack or Microsoft Teams — wherever your team already talks.
Onboarding that doesn’t eat your senior staff
A new hire’s first fifty questions get answered without interrupting the one person who knows everything.
A backup for what’s in people’s heads
The knowledge your long-tenured staff carry stops being a single point of failure the week they take holidays.
The short version on money: builds start at $4,500, plus a Care Plan from $299/mo to keep it running. Fixed-quoted before we start — full numbers further down, or on the pricing page.
Go-live in 3–6 weeks
How it works — from scattered binders to cited answers.
Week 1
Discovery.
We identify the authoritative sources, set the access rules, and collect the top questions staff actually ask.
Weeks 1–3
Ingestion & build.
We pipeline your SOPs, handbooks and drives (and scoped email/ticket history), build the assistant, and wire the chat interface.
Weeks 3–5
Testing.
Answer-quality evaluation against a real question set, hallucination guardrails, and "I don't know" behaviour instead of guessing.
Weeks 5–6
Rollout.
Staff rollout session, admin guide, supervised launch, and a first-week answer-quality review. Then the Care Plan keeps it sharp.
Client effort: Your effort: medium — gathering and de-duplicating source documents is the real work, and it's mostly yours. Budget a few hours across a couple of weeks.
What's included
What it does — and what's included.
| Phase / item | What you get |
|---|---|
| Discovery | Identify authoritative sources, access rules ("who can see what"), and the top questions staff actually ask. |
| Build | A private internal assistant over your own documents, with a chat interface (web and/or Slack/Teams) and a source citation on every answer. |
| Ingestion | Pipeline for your SOPs, handbooks, drives, and — where scoped — email and ticket history, with refresh on update. |
| Access control | Role-based permissions so answers respect who's asking (PIPEDA-aware). |
| Testing | Answer-quality evaluation against a real question set, hallucination guardrails, "I don't know" behaviour. |
| Training & docs | Staff rollout session, admin guide for adding/removing sources. |
| Go-live | Supervised launch, first-week answer-quality review. |
Not included: writing your SOPs for you (we can scope that separately), public-facing use (that's Website AI Chat), records digitization at scale. Common add-ons: email/ticket ingestion, multi-department knowledge silos, analytics on what staff ask most — a goldmine for finding process gaps.
Under the hood
Why a company brain is harder than a search box.
Pointing an AI at a shared drive takes an afternoon. Making it right, current and safe is the actual work — and it lives in these categories.
- Deciding what’s true. Most businesses have three versions of the same policy in three places. Before anything is indexed, someone has to declare which one is authoritative. We run that call with you, and it is usually the most valuable hour of the whole build.
- Answer quality, measured. We build a question set from what your staff really ask, then test answers against it before launch. "I don’t know" is a correct answer, and we make sure it happens instead of a confident guess.
- Permissions that hold. Answers respect who is asking — payroll doesn’t surface for the seasonal crew. Access rules are built in, not a setting we hope someone configures later.
- Messy sources, handled. Scanned PDFs, a decade of email, ticket history, three folders all called FINAL. Wrangling that into something answerable is most of what you are paying for.
- Staying current. Sources refresh as they change and answers get re-tested under the care plan. A brain that quietly went stale is worse than no brain, because people trust it.
- Knowing what your team keeps asking. The question log is a map of where your own process is unclear. Owners routinely find it more useful than the answers themselves.
Named as categories on purpose. Which of these your business actually needs, in what order, and what it would save is the deliverable of the paid AI Opportunity Audit — or the free 20-minute assessment call. We publish the shapes, not the recipes.
Works with what you've got
It reads what you already have.
We build on top of your existing stack — no rip-and-replace.
- Document sources: Google Workspace, Microsoft 365, SharePoint, shared drives, or a docs export.
- Chat interface: web, and/or right inside Slack or Microsoft Teams.
- Email & tickets: ingested where scoped, refreshed on update.
- Access control: role-based permissions, so answers respect who's asking.
Prerequisites: documents in reasonably accessible form, clarity on access rules, and a decision-maker to declare which sources are authoritative.
The honest part
What this isn’t — and when it’s not worth it.
Every build has edges. Here are ours — better read now than discovered in week three.
- It isn’t a public chatbot. This one is for your staff. The customer-facing version is Website AI Chat.
- It won’t write your SOPs. It answers from what exists. If the knowledge lives only in one person’s head, somebody has to write it down first — we can scope that separately, but we won’t pretend the AI does it for you.
- It answers; it doesn’t act. Raising the ticket, sending the email, updating the record — that’s Workflow Automation layered on top, and we keep the line clear on purpose.
- When it’s not worth it yet: a team of three who all know everything doesn’t need this. Turnover, seasonal hiring, or crews who can’t reach the office are what make it pay.
What it looks like in practice
An example engagement (illustrative).
A fictional-but-realistic engagement to show the shape of a build — not a case study, and not a guarantee. Real proof arrives with our founding clients.
Monashee Property Management, Revelstoke Illustrative
400+ units, seasonal staff, and answers scattered across binders, a shared drive, and two long-tenured managers' memories. The company brain lets any staffer ask "what's the pet policy at Glacier View?" or "what's our snow-removal SLA?" and get a cited answer. Setup $8,900, Growth care. Onboarding a seasonal hire drops from weeks to days.
Transparent pricing
Starting at $4,500 to build. From $299/mo to keep it sharp.
Moves up with: number and messiness of sources; email/ticket ingestion; permission complexity; number of departments; whether we clean and structure documents first. A single tidy SOP set sits near $4,500; a multi-source, permissioned, email-inclusive brain approaches $12,000.
- AI Company Brain — build
- Starting at $4,500. Typical range $4,500–$12,000, fixed-quoted to your setup.
- Care Plan — keep it running
- Requires a Care Plan, from $299/mo (Core); Growth for larger corpora or teams.
- Payment terms
- 50% to start, 50% at go-live under $10,000; 50/25/25 at $10,000 and up.
- Usage
- AI usage (and SMS where the build sends messages) passes through with a monthly allowance included, itemized on every invoice — never marked up in secret.
Questions
AI Company Brain FAQ.
What can it actually answer?
Anything in the sources you declare authoritative — SOPs, handbooks, policies, price lists, past quotes, and (where scoped) email and ticket history. Every answer cites its source, and when it doesn't know, it says "I don't know" instead of guessing — that behaviour is tested before launch.
Who can see what?
You decide. Role-based permissions are built in, so answers respect who's asking — payroll documents don't surface for the seasonal crew. It's PIPEDA-aware by design.
How much work is it for us?
Medium — gathering and de-duplicating your source documents is the real work, and it's mostly yours. Budget a few hours across a couple of weeks; we handle everything else.
Isn't this what Microsoft Copilot already does?
Partly, and only inside Microsoft. A per-seat assistant searches the silo it lives in. A company brain reads across the shared drive, the email archive, the ticket history and the scanned paper SOPs, respects your own access rules, and cites the source on every answer. If your knowledge genuinely all lives in one tidy Microsoft 365 tenant, we'll tell you to save your money.
What does it cost?
Builds start at $4,500 and are fixed-quoted before we begin (typical range $4,500–$12,000 by the number and messiness of sources). Every AI build includes a Care Plan from $299/mo — Growth for larger corpora or teams. Full detail on the pricing page.
Pairs well with
Most builds don't stay solo.
Every system you add sits under the same single Care Plan — the tier grows only when your setup does.
Is it worth it for your business? Let's find out — free.
Book a free 20-minute AI assessment. We'll tell you what this build would do for you, what it'd cost — and if it's not worth it yet, we'll say so.
Prefer to talk now? Call (236) 837-3550 — our own AI receptionist picks up, 24/7. Yes, the same one we sell: consider the call a live demo. It books, answers, or takes a message — and a real BC-based human follows up within one business day, usually faster.