← Insights

What 'AI-ready' actually means for a small business

By Fred Lundin5 min read
Small Biz Data framework

Building your business data foundation.

The journey from a messy data state to a structured foundation — establishing permanent standards and security protocols for AI readiness.

The discovery phase
Mapping & cleaning.
Inventory every system and file.

Map every data source — spreadsheets, email inboxes, POS, and even paper files.

Restructure for a "CPA-grade" chart of accounts.

Target under 50 active accounts so financial reports stay clear and functional.

Designate a "source of truth" for customers.

Pick one system (CRM or billing) that "wins" whenever customer records overlap.

Risk
Critical action
Data loss
Every system has a cloud or local backup.
Access control
Revoke credentials for former employees immediately.
Privacy
Stop using free-tier AI tools for sensitive data.
The governance phase
Standardizing & securing.
Implement "side of the road" naming standards.

Use YYYY-MM-DD for dates and consistent formats for filenames and addresses.

Document the top 3 "hurt most" processes.

Identify workflows that would break if a key employee left tomorrow.

Establish a same-day departure protocol.

Revoke access to every system the day an employee leaves — without exception.

Turn MFA on for every business-critical account.

Payroll, accounting, email, banking. If the account can move money, it needs a second factor.

Vet AI vendors on data-handling.

Confirm your inputs aren't used to train the vendor's models before pasting anything sensitive.

"AI-ready" is one of those phrases that sounds like a strategy and usually isn't. Vendors mean different things by it. Consultants mean different things. And the version that matters for a 12-person plumbing company is not the version that matters for Goldman Sachs.

Here's what it actually means when you run a small business.

Your data lives in one place per subject

Not literally one database. But if I ask "who are our top ten customers by revenue?", you should be able to answer that from one place, not by exporting three spreadsheets and reconciling them.

Your data is machine-readable

Dates look like dates. Money looks like money. Categories are consistent. If your top revenue category is spelled "Consulting", "consulting", and "consulting svcs" across three systems, no AI tool is going to summarize your revenue for you accurately.

You know who touches what

If a staffer pastes a customer list into ChatGPT, you should know (a) whether that was allowed, (b) what happened to that data, and (c) whether any AI vendor is training their model on your customer list.

You have at least one process written down

The magic of AI at small scale is context. A generic model can write "an email". A model with your invoicing process written down can write your invoice reminder email — the one that matches how your business actually operates.

You have a low-stakes pilot in progress

The businesses that succeed with AI aren't the ones with the biggest strategy decks. They're the ones who picked something small — categorizing bank feed transactions, drafting follow-up emails, summarizing weekly numbers — and got a win.

That's the whole picture. Everything else is a vendor pitch.

Ready to make your data actually useful?

Book a free 30-minute assessment. We'll walk through your current data situation, spot the highest-leverage fixes, and give you a clear path forward — no commitment.

No commitment · No technical jargon · 30 minutes