AI-Ready Before AI: How an SME Actually Prepares

What “AI-Ready” Means When You're Not a Fortune 500

Search for “AI readiness” and you will find frameworks written for companies with chief data officers, model risk committees, and procurement cycles measured in quarters. That is enterprise governance theater, and if you run a twenty-person services firm, a growing SaaS company, or a healthcare practice, almost none of it applies to you. You do not need an AI ethics board. You need to know which of your workflows is bleeding time, whether the data feeding that workflow is trustworthy, and who on your team will own the fix.

I say this as someone who sits on both sides of the question. I run Interactive Intel, an AI consulting practice, and I run it on agentic AI: our lead intake scrapes and structures inbound email, an agentic intake assistant handles client onboarding, our proposals are drafted by automation, and the website you are reading was rebuilt with Claude Code. All of that is in production today, in a firm small enough that I personally see every workflow. So when I talk about AI readiness for small business, I am describing what I had to get right in my own company before any of that worked — not a slide I borrowed from an enterprise playbook.

The honest definition is this: an SME is AI-ready when it can hand a clearly described process, with reliable inputs, to a system — human or machine — and measure whether the output got better. That is it. Everything else is detail. The trouble is that most small businesses fail one of those three clauses, and they usually fail it long before any model enters the picture.

The 10/20/70 Rule — and Why the Model Was Never My Blocker

Our methodology at Interactive Intel is built on a split we call 10/20/70: roughly 10% of the work in a successful AI adoption is the algorithms, 20% is technology and data plumbing, and 70% is people and process. The numbers are a heuristic, not a measurement — but every engagement we run confirms the shape of the curve, and so did my own firm.

When I built our email-scraping lead intake, the model was the easy part. Claude could read an inbound inquiry, extract the company, the problem, and the urgency, and structure it perfectly — on day one. What it could not do was compensate for the fact that Ihad never written down what happens to a lead after it arrives. Who qualifies it? Against what criteria? What does “qualified” even mean for a boutique that takes a limited number of engagements? Until I sat down and documented that intake process step by step — the 70% work — the AI had nothing coherent to automate. The model was never the blocker. The undocumented process was.

This is the pattern I now see everywhere. SMEs come to us asking which model to use, whether they need fine-tuning, whether their data is “big enough.” Those are 10% questions. The questions that decide success are: can you describe the workflow precisely enough that a new hire could run it from your description alone, and is there one person accountable for it? If the answer is no, no amount of model capability rescues you.

A Practical Readiness Checklist for SMEs

Strip away the enterprise framework language and AI readiness at SME scale comes down to four things. This is the same checklist we apply at the start of every engagement, and the same one I applied to my own firm.

Notice what is not on the list: a data lake, a governance committee, a multi-year roadmap. One workflow, documented, measured, owned. That is the unit of AI readiness at SME scale. Once you have done it once, the second and third workflows go dramatically faster — because the muscle you built is organizational, not technical.

What a Real Migration Taught Me About Data Readiness

The data-hygiene item deserves its own confession. Alongside the consulting practice, I ship Medop, a MedSpa operating system, and part of that work has involved migrating a practice off a legacy EHR — Tebra — into a modern stack. I will be straight about it the way I am with clients: some of it worked cleanly, and some of it broke.

What worked was anything that lived in structured fields with consistent meaning — those records mapped over with predictable effort. What broke was everything the old system had allowed to be ambiguous: free-text fields doing the job of structured data, conventions that existed only in staff habit, records whose real-world meaning depended on context no export file carries. None of that is the legacy vendor's fault, exactly. It is what happens in every business where software quietly absorbs years of improvisation.

The lesson for AI readiness is direct: your data is almost certainly messier than you think, and the mess is invisible until something — a migration, an AI agent — tries to read it literally. An AI system consuming your records inherits every ambiguity in them. You do not need to fix all of it before starting. You need to know where the mess is, scope your first workflow around data you can trust, and clean the rest deliberately as you expand. Discovering it mid-project is the expensive version.

What an AI Readiness Audit Actually Covers

This checklist is exactly what our AI Readiness Audit formalizes. It is a two-to-three-week engagement, priced at $5K–$15K depending on the complexity of your operation — and we publish that range because guessing at consulting prices is a waste of everyone's time. Over those weeks we map your core workflows and document the ones that matter, assess the real state of your data (not the org-chart version of it), identify the single highest-leverage workflow for a first AI deployment, and define the metric and the owner for it. You finish with a written readiness report and a prioritized roadmap you can execute with us, with another firm, or on your own.

For some clients the audit's most valuable finding is “not yet” — here are the three process and data gaps to close first, and here is the order to close them in. That outcome costs a few thousand dollars. Learning the same lesson inside a stalled six-figure implementation costs considerably more.

Where to Start

If you take one thing from this: AI readiness for a small business is not a technology project, it is a clarity project. Pick the workflow that annoys you most. Write it down. Look honestly at the data feeding it. Put a number on it and a name next to it. Do that, and the AI part — the 10% — becomes the easy part, just as it was for me.

If you want a fast, structured read on where you stand, start with our free AI Readiness Assessment — it takes minutes and gives you an honest baseline. And if you would rather talk it through with the person who would actually deliver the work, book a call with me directly. There is no junior team behind the curtain; the founder you read is the founder you get.