Twelve questions. No consultant required. If you answer honestly, you will know in thirty minutes whether AI can save you money this quarter, or whether you are three steps away from being ready. I run five small businesses solo on a $600-a-month AI stack. This is the same internal checklist I use on every new client before I will take their money.
Most AI advisory firms will charge you between $5,000 and $25,000 for an "AI readiness assessment." They deliver a 40-page deck six weeks later. The deck describes your business back to you in consultant English and recommends a 9-month transformation.
That is theater. You do not need a deck to know if AI fits your business. You need twelve honest answers.
This checklist assumes you run a $1-10M revenue small business. Ecommerce, distribution, professional services, local trades, light manufacturing. Not a 500-person enterprise, not a pre-revenue startup. The tier where AI stops being a fad and starts being payroll replacement.
Examples: customer emails, lead calls, invoice entry, shipment confirmations, review responses, quote requests, support tickets, scheduling calls.
YES = AI can likely automate one or more of these. Proceed.
NO = your business is physical-labor heavy (e.g., a roofing crew on-site). AI can help on the back-office side, but the ROI is smaller. Check Part 2 before continuing.
Invoice processing, quote follow-up, inventory check, review moderation, appointment reminder, and so on.
YES = that's your first target. Repetition is where AI wins.
NO = each task is bespoke? Then you need a knowledge-work AI (assistant), not an automation. Different budget, different ROI.
The "hates" part matters. If a human hates it, the output quality is probably declining anyway. AI doesn't get tired.
YES = that is your first AI deployment. Name the task.
NO = the team is either content or too diplomatic to admit it. Ask each team member privately which task they would pay someone else to do. Their answer is your target.
If yes, AI can probably do it. The test of automatability is: can you write it down clearly?
YES = documentation already exists or is in your head and exportable. AI-ready.
NO = the task relies on tacit knowledge or judgment that has never been codified. That's expensive to automate. Start with a different task.
Examples: Shopify, QuickBooks, HubSpot, Pipedrive, Stripe, Zoho, Airtable, Google Sheets, a Postgres database.
YES = AI can read live data. Every automation becomes 10x more valuable.
NO = data lives in paper, Excel files on someone's laptop, or a custom system with no exports. Step 1 is migrating to something queryable. Budget 2-4 weeks before any AI work.
For ecommerce: orders. For distribution: invoices. For services: billable time. For inventory-based: purchase orders.
YES = enough context for AI to pattern-match. Forecasting, pricing, segmentation become options.
NO = too early for predictive AI. Stick to automation and document handling. Revisit in 6 months.
YES = AI can learn your voice, tone, and edge cases from real examples. That is half the battle.
NO = communication is in one person's inbox or Slack DMs. Step 1 is centralizing into something AI can read. Gmail works. Front works. HubSpot works.
Not an expert. Curious. Someone who tried ChatGPT and liked it. Someone who will click a button you tell them to click.
YES = you have an internal champion. AI rollout has a chance.
NO = the team will quietly sabotage the deployment. AI fails in week 4 because nobody opens the dashboard. Solve this first or do not start.
This is the delicate part. AI does not replace someone usually. It upgrades them from doer to supervisor.
YES = the existing operator becomes the AI supervisor. Headcount stays or shifts to higher-value work.
NO = the person fears being fired. Legitimate fear. Address openly. In my own businesses, nobody was laid off because of AI. The scope of each person expanded.
Not delegated. You. Reviewing outputs, adjusting prompts, approving edge cases.
YES = deployment will work.
NO = hire a Fractional AI Officer. That is literally what the role solves: someone owns the deployment so you don't have to.
Real cost of an operator-layer AI stack for a $1-10M business: $500-1,500 per month. My own is ~$600 across 5 businesses.
YES = proceed. ROI starts around month 2-3.
NO = wait until cash flow allows. Or deploy a single surgical automation for under $200/month. Start smaller.
Most AI projects I see fail not because AI fails. They fail because the business couldn't survive the month 2 cost. If the AI stack is life-or-death to your P&L on day one, the stakes will poison the rollout.
YES = you can run the deployment without panic. Panic kills AI projects more than bad prompts do.
NO = start with a single surgical task (e.g., review responses on Amazon) that costs under $100/month. Prove ROI at tiny scale. Reinvest.
| "YES" count | Your status | What to do |
|---|---|---|
| 10-12 | AI-ready today | Pick one task from question 3. Deploy in 2-4 weeks. Run for 90 days before adding anything. |
| 7-9 | 3 weeks from ready | Fix the "no" items first. Most are data or people gaps. Then redo the audit. |
| 4-6 | Not ready yet | Do the foundational work (centralize data, find a champion, budget allocation) before touching AI. |
| 0-3 | Too early | Focus on operational fundamentals for 3-6 months. AI is a force multiplier on a working business, not a fix for a broken one. |
They try to automate the most complex task first. Big mistake.
The right first deployment is a boring, repetitive, text-heavy task that nobody will miss. Invoice entry. Review responses. Appointment reminders. Lead qualification from an inbound form. Something you already wrote a checklist for.
Prove the stack on something boring. Then go hunt the hard stuff in month 3.
If you score 7+ on the checklist and want a second pair of hands:
Cost: $2,500 for the 2-week audit. $5-8K for the Sprint deployment. $3-5K/month for the retainer afterward. Total Year 1 cost: $40-75K. Typical ROI in headcount avoided: 1.5-3x.
Most AI consultancies I've looked at will spend three months "assessing." I install and run a stack in five weeks. Price is lower. Risk is lower. Decision is reversible. That is the whole pitch.
DIY if: you are a founder-operator who loves systems, has an AI-curious team member, and can give 5-10 hours per week to the build. ChatGPT plus Zapier plus one afternoon with Claude API docs will get you 60% of the way.
Call someone if: you are the only one in the company who can own this, your time is worth more than $300/hour, and you cannot afford a 3-month stall while you learn on your own. At that point a Fractional AI Officer is the cheapest path to production.
Call me specifically if: you are in the $1-10M revenue range, already past the "should we do AI" question, and need someone who runs their own businesses on the stack they sell you. Everything I deploy on a client, I run on one of my five businesses first.
Want me to do this audit with you live?
30-min call. No slides, no pitch. We walk through your answers together.
Book 30 minutesDmytro Negodiuk is a Fractional AI Officer based in Brooklyn. He runs five small businesses solo on a $600/month AI stack. Forbes profiled the practice on April 15, 2026 (Gene Marks, Quicker Better Tech column).