AI Foundations for Small Business · Lesson 6 of 11

Industry Playbooks

Time: ~30 min reading + ~15 min worksheet Goal: Develop fast intuition for "where AI pays off" in the most common SMB verticals. You don't need to be an industry expert — you need to know what to look for.

---

Every industry has a handful of structural patterns that show up over and over: how the business makes money, where the bottlenecks are, what kind of data flows through. Once you see the pattern, AI opportunities stop hiding.

We'll hit nine common SMB verticals. For each:

• The business shape (how they make money and where time goes)
• Top 3 AI plays (ranked by likely ROI)
• Watch-outs (compliance, risk, or cultural traps)

Business shape

Selling expert time. Revenue = hours × rate. Margins come from efficiency of expert delivery and reducing non-billable admin.

Top AI plays

• Document intake and extraction — tax returns, contracts, case files. Manual data entry is pervasive.
• First-draft writing — memos, letters, briefs, engagement letters. Save hours per document.
• Knowledge retrieval — RAG over past cases, past tax returns, past engagements. "Have we seen this before?"

Watch-outs

• Confidentiality is non-negotiable. Attorney-client privilege, CPA-client confidentiality, HIPAA-adjacent issues. On-prem or tightly-controlled cloud only.
• Ethics rules matter. Bar associations and state CPA boards have AI opinions. Don't pretend otherwise.
• Hallucination risk is existential. Lawyers who filed AI-generated briefs with fake cases have been sanctioned. Reviewers are required.

Worth knowing: Mid-size professional firms — family CPAs, small law practices, local consultants — are ideal early AI adopters *if* you handle the privacy story credibly.

Business shape

Reimbursement-driven. Revenue hinges on accurate coding, complete documentation, and getting clean claims out the door. Clinicians hate paperwork.

Top AI plays

• Ambient clinical documentation — AI listens to visits, writes the note. Abridge, Nuance DAX, Suki; also buildable custom. Huge clinician satisfaction impact.
• Prior authorization and claims scrubbing — extract data from notes, identify issues before submission.
• Patient communication assist — drafting replies to portal messages (with clinician review).

Watch-outs

• HIPAA. Business Associate Agreements (BAAs) required for anything that touches PHI. OpenAI, Anthropic, and Microsoft all offer BAA options — *you have to sign up for the right tier.* Default consumer products are NOT HIPAA compliant.
• Clinical liability. AI cannot diagnose or prescribe. Every AI-touched decision must have a licensed human owner.
• Change management is brutal. Clinicians have seen a lot of "technology solutions" that made their lives worse. Be humble.

Business shape

Make-to-order or make-to-stock. Revenue comes from throughput, quality, and managing complex supply chains. Information systems often fragmented (ERP, MES, CAD, shop floor).

Top AI plays

• Quote / configurator assist — complex products with many options. AI reads the customer inquiry, suggests a configuration, drafts a quote.
• Quality / defect analysis — computer vision on inspection photos (classical ML territory). Or LLMs summarizing warranty/defect patterns.
• Predictive maintenance — classical ML on sensor data. Reduce unplanned downtime.

Watch-outs

• Data silos can be severe. Integration is often the real project, not AI.
• Shop floor culture — the people doing the work must be brought along. Top-down AI in manufacturing fails often.
• Safety-critical decisions must stay human.

Business shape

Project-based. Revenue from bid wins, execution margin, and change orders. Cost control is everything.

Top AI plays

• Bid / estimate drafting — AI reads plans or specs, pulls out quantities, drafts estimate. Still emerging but improving fast.
• Daily report / log automation — foreman does a voice memo, AI writes the daily log, flags issues.
• Document extraction from specs, RFIs, submittals — Extractor pattern on PDFs.

Watch-outs

• Tech adoption culture varies wildly. Some contractors are bleeding edge; some still use faxes. Meet them where they are.
• Field connectivity matters. Offline-capable tools win.
• Plans and specs are often drawings. Vision models are needed, and they're not perfect yet.

Business shape

Sell product, keep customers coming back. Margins depend on merchandising, ops efficiency, and marketing.

Top AI plays

• Product description generation at scale — classic Expander use case. Hundreds of SKUs get real descriptions instead of manufacturer defaults.
• Customer service automation — bot on site + email triage.
• Review summarization + voice-of-customer — what are shoppers actually saying about the products and the experience?

Watch-outs

• AI-generated content and SEO — duplicate or low-quality AI content gets penalized. Quality matters.
• Customer service tone — AI reply bots that sound tone-deaf destroy brands. Lean toward suggest-not-send.
• Returns fraud / review fraud are their own AI sub-problems, often needing classical ML.

Business shape

Tight margins, high labor dependency, brutal operational pace.

Top AI plays

• Marketing and content — social posts, menu descriptions, seasonal campaigns.
• Review response automation — AI-drafts replies to every Yelp/Google review, owner approves.
• Scheduling optimization — classical ML or software with AI features (7shifts, Homebase).

Watch-outs

• Owners are busy. If it doesn't work on the first visit, they'll never come back to it. Ruthlessly simple is the only thing that sticks.
• Voice AI for ordering is a whole category — mixed results so far.
• Don't overcomplicate. A $9/month tool that saves an hour a week beats a $500/month custom build.

Business shape

Move things efficiently. Revenue depends on route density, load matching, and minimizing empty miles / overstock.

Top AI plays

• Document intelligence — BOLs, invoices, customs forms. Extractor pattern at scale.
• Load matching / shipment optimization — often AI-assisted, sometimes classical ML, sometimes heuristics.
• Customer / dispatch communication — automated status updates, exception handling, AI summaries of long email threads.

Watch-outs

• Integrations are the project. Everyone uses different TMS/WMS systems. Data plumbing is most of the cost.
• Drivers don't want more apps. Ops people do. Target the right user.

Business shape

Relationship-intensive. Revenue from renewals, referrals, and carefully matched placements.

Top AI plays

• Quote / proposal generation — from carrier docs or past similar policies.
• Claims / application document intake — extracting structured data from submitted forms.
• Renewal email personalization — AI-writes based on the client's specific situation.

Watch-outs

• Regulation — state-level insurance rules, KYC, AML. Be careful.
• Accuracy matters more than speed in most cases. Don't ship AI-drafted anything to a client without review.

Business shape

Mission-driven, often resource-constrained, grant/funding-driven.

Top AI plays

• Grant writing assist — Expander pattern, massive time saver.
• Constituent/member communication — response drafting, newsletter generation.
• Document summarization — board packets, meeting minutes, policy summaries.

Watch-outs

• Transparency expectations are higher. Disclose AI use.
• Procurement processes can be slow. Plan for it.
• Ethics — vulnerable populations, community trust. Go carefully.

When you're in an industry you don't know well, ask these five questions:

• What's the main unit of work? (A case? A job? A ticket? A patient visit? A quote?)
• Where does time disappear? (Documentation? Research? Rework? Waiting?)
• Where does text pile up? (Incoming emails, PDFs, forms, reports.)
• Where does knowledge get lost? (When an expert leaves, retires, or is out sick.)
• What do they lose deals over? (Speed? Accuracy? Presentation?)

Those five questions will surface 80% of the real opportunities in any industry, even ones you're brand new to.

---

• Every industry has structural patterns — learn to see them.
• Professional services, healthcare, logistics usually have the deepest "extractor" opportunities.
• Construction, trades, restaurants need AI that survives rough conditions and busy owners.
• Healthcare and legal have the highest compliance stakes — build your story before you pitch.
• Five questions work in any industry to find opportunities fast.

---

Next up: Module 7 — Data, Privacy, and Risk. The part most people skip and regret.

---

Time: ~15 min Goal: Apply the five questions and the industry playbooks to fresh scenarios.

---

For each industry, answer the five questions briefly (one sentence each).

A. A 40-person home health agency

• Main unit of work: ______________
• Where time disappears: ______________
• Where text piles up: ______________
• Where knowledge gets lost: ______________
• What they lose business over: ______________

Best AI play to lead with: ______________

B. A 15-person wholesale food distributor

• Main unit of work: ______________
• Where time disappears: ______________
• Where text piles up: ______________
• Where knowledge gets lost: ______________
• What they lose business over: ______________

Best AI play to lead with: ______________

C. A 60-person regional architecture firm

• Main unit of work: ______________
• Where time disappears: ______________
• Where text piles up: ______________
• Where knowledge gets lost: ______________
• What they lose business over: ______________

Best AI play to lead with: ______________

For each client, name the single most important compliance / risk issue to address before you recommend anything.

• Solo dermatology practice, 4 employees → ______________
• Small personal injury law firm → ______________
• Retail pharmacy, 3 locations → ______________
• Regional non-profit serving adults with disabilities → ______________
• Small investment advisory firm → ______________
• Dental office, 12 employees → ______________

---

A. Home health agency:

• A patient visit / encounter
• Clinical documentation after visits, scheduling, coordination calls
• Visit notes, care plans, physician orders, progress reports
• In senior caregivers' heads; also lost when clinicians leave
• Staffing reliability, documentation completeness for reimbursement

Lead: HIPAA-compliant ambient clinical documentation. Huge clinician satisfaction + reimbursement impact.

B. Wholesale food distributor:

• An order / load
• Order entry (often phoned/emailed), invoice processing, dispatch
• Order emails, invoices, delivery manifests, credit claims
• In dispatcher's head; long-time sales reps' relationships
• Slow quote turnaround, order errors, late deliveries

Lead: Order intake automation — emails/PDFs → structured orders in the system. Extractor pattern, fast ROI.

C. Architecture firm:

• A project / design submission
• Producing proposals and RFQ responses, specification writing, coordination
• RFPs, specs, drawing markups, consultant coordination emails
• In senior architects' heads; past project knowledge poorly searchable
• Proposal speed, creative differentiation in pitches

Lead: RAG over past proposals + specs + project write-ups. "Have we done something like this before?"

Exercise 2:

• Solo dermatology → HIPAA / BAA required for any AI touching patient data
• Personal injury law firm → Attorney-client privilege and state bar's AI rules; hallucination risk on case research
• Retail pharmacy → HIPAA plus state-level prescription / DEA regulations
• Non-profit serving adults with disabilities → Vulnerable population protections, consent, transparency about AI use
• Investment advisory → SEC / FINRA regulatory constraints; record-keeping and recommendation documentation rules
• Dental office → HIPAA / BAA plus state dental board rules

---

Check yourself: If you're sitting across from an SMB owner in an industry you've never seen before, could you use those five questions to build a credible opportunity map in the first 30 minutes of the meeting? That's the bar.

Previous: AI for SMB: Where the Real Value Lives · Next: Data, Privacy, and Risk (The Boring Stuff That Matters)