AI readiness checklist
Before the first worker: name the workflow, list the data, and decide what a person still checks.
Most small businesses are "AI curious" and "data messy" at the same time. That is fine. Readiness is not about perfect databases — it is about knowing which workflow hurts enough to fix first and what must stay accurate while you fix it.
1. Name the workflow capping capacity
Pick one process where delays are visible or errors are expensive: month-end cleanup, document chasing, maintenance triage, collections follow-up. If you cannot describe it in five bullet steps, map it before you buy tools.
2. Inventory data sources
List where truth lives today: spreadsheets, shared drives, CRM, ledger, inbox threads. Note duplicates and manual copy-paste. The data digitalization phase scopes the smallest dataset the first worker needs. It is not a multi-year ERP project.
3. Assign an owner
Every agent reports to someone on the org chart. That person owns the role, the exceptions, and the outcome.
4. Write the rules the agent follows
Client messages, ledger posts, and exceptions need rules in the agent: what it can send, what it can post, and when it hands the work to a person. The agent follows those rules while it works.
5. Log every action
Every send, post, and tool call goes in an append-only log. The team can see what the agent did and step in on exceptions.
6. Set a low-risk entry
A paid Back-Office Worker Scan starts at $1,500. Trades shops can start with a Cash Leak Audit at $500, credited toward setup. You leave with the first job to put a worker on.
Ready to test with real work? Bring one manual process — we will tell you if an agent fits or if digitization comes first.