Home/Guides/Use Cases
Guide 02 · ~8 min read · July 2026

How small businesses here actually use AI.

Not the demos from the keynotes — the workflows that show up again and again in real Atlantic Canada businesses: contractors, clinics, distributors, professional firms. Six use-cases, each with what it's good for, what it won't do, and the rough effort involved. If a vendor only tells you the first column, keep your wallet closed.

A note on scale before we start: these are small-business use-cases — the 3-to-50-person operations that make up most of the economy out here. None of them require a data science team, and most of them start with software you already pay for. Microsoft's 2025 Work Trend Index puts the time recovered by employees who use AI well at six-plus hours a week; whether you get anywhere near that depends less on the tools and more on picking the right workflow and actually training people. That's what the honesty columns below are for.

Use case 01 · Writing

Email and document drafting with Microsoft 365 Copilot

The lowest-friction starting point, because it lives inside Outlook and Word where your team already works. First drafts of routine emails, summaries of long threads, first-pass proposals and letters — the blank-page tax, mostly eliminated.

Good forRoutine correspondence, summarizing long threads and documents, first drafts your team edits rather than writes from scratch.
Won't doKnow your business facts or prices — it drafts from what it can see, so someone must review before send. It also won't fix a messy file system underneath.
Rough effortDays, not months. Copilot licensing from Microsoft (a per-user monthly add-on), plus a training session ($1,200–2,000) so the habits actually stick.
Use case 02 · Operations

Quote and work-order automation

A request comes in by phone or email, someone retypes it into a quote, then into a work order, then into the schedule. AI-assisted intake turns the request into structured data once, and everything downstream flows from it. We built exactly this for a New Brunswick overhead-door company — dispatch and work orders — and it's running as a live pilot.

Good forRepetitive quoting from known pricing rules, fewer retyping errors, faster turnaround from request to scheduled job.
Won't doPrice unusual one-off jobs, or handle exceptions without a human check — the win is drafting the routine 80%, not replacing the estimator.
Rough effortA custom build, typically mid-range of $8,000–25,000, scoped first by a process review ($850–2,800) so you know the price before committing.
Use case 03 · Follow-through

Meeting notes into CRM and action items

Teams and most meeting tools can transcribe already. The useful step is the one after: turning the transcript into action items with owners, and into a CRM record that doesn't depend on whoever took notes remembering to do it Friday afternoon.

Good forNothing slipping after sales calls and project meetings; consistent records across the team instead of three personal notebooks.
Won't doUnderstand what wasn't said out loud, get every name right in a noisy room, or do the actual follow-up — it removes the clerical step, not the relationship.
Rough effortLight. Often configuration of tools you own plus a training session; a small custom build only if you want automatic CRM writes with review steps.
Use case 04 · Paperwork

Document intake and processing

Invoices, intake forms, purchase orders, inspection reports — documents arrive as PDFs and email attachments, and someone keys them into a system. AI document processing extracts the fields and routes them, with a person confirming instead of typing.

Good forHigh-volume, repetitive document types; getting data into your systems the same day it arrives instead of at month-end.
Won't doHit 100% accuracy — a review step stays in the loop by design. Wildly inconsistent document formats need setup work per format.
Rough effortA custom build, lower-to-mid $8,000–25,000 depending on how many document types and systems are involved.
Use case 05 · Knowledge

An internal chatbot on your own documents

Policies, procedures, product specs, price lists — the answers exist, but they live in a folder nobody searches. A private chatbot over your own documents lets staff ask in plain language and get the answer with a citation to the source file. New-hire onboarding is where this earns its keep fastest.

Good forQuestions that get asked repeatedly, onboarding, and making one person's institutional knowledge available to everyone.
Won't doAnswer what was never written down, or stay useful if the documents rot. The bigger cost is usually cleaning up the documents first — that's the data-hygiene score in the assessment.
Rough effortLower end of the $8,000–25,000 build range for the bot itself; add time for document cleanup if your files need it.
Use case 06 · Reporting

Reporting with Power BI and AI

Monthly reporting that takes two days of copy-paste can become a dashboard that's always current, with AI features for plain-language questions ("how did service revenue compare to last March?") and first-pass written summaries your team refines instead of writes.

Good forRecurring reports, spotting trends earlier, and giving owners a live view instead of a month-old snapshot.
Won't doFix bad or inconsistent source data, and it won't make the decision for you — it makes the picture clearer, not the call.
Rough effortVaries most of the six: from a training session on tools you own to a scoped build for automated data pipelines. (Zaiphr's founder is a Microsoft Certified Power BI Data Analyst Associate — this one's close to home.)

Honest about the limits

Every use-case above comes with the same fine print, so here it is in one place:

  • AI output needs review. Every workflow above keeps a person in the loop somewhere. If a vendor proposes removing the human entirely from customer-facing or financial steps, push back.
  • Messy data in, messy answers out. Several of these depend on your documents and records being findable and roughly consistent. Sometimes the honest first project is cleanup, not AI.
  • Adoption is the hard part. Tools that nobody was trained on become shelfware by month three. Budget for training and a rollout plan, not just the build.
  • Some workflows aren't worth automating. Low volume plus high variance means the setup never pays back. A good assessment says "skip this one" out loud.
  • Set data rules before rollout, not after. Decide what can go into which tools — client data especially — before your team starts pasting things into chatbots on their own.

If you want help figuring out which of these six (if any) fits your business, that's exactly what the $2,000 assessment is for — and government funding may offset part of the cost for Atlantic Canada businesses.

Next step

Which one fits your business?

Tell us what's eating your team's time and we'll give you an honest read — including "none of the above" if that's the truth. We reply within 1 business day.

Responds within 1 business day Moncton, NB · serving Atlantic Canada Honest scoping · no hype