The 90-day AI roadmap for an SME: what to do before buying more tools

Ninety days is enough to go from "we should use AI" to one workflow running in production with a person who owns it — if you spend the first thirty days measuring and deciding instead of buying. The roadmap below is the one I run in my own companies and recommend to the founders I work with.

Days 1–30: measure and decide

Map the ten conversations that happen most between your business and customers, suppliers and staff. Find the copy step in each. Measure frequency, cycle time, error rate and response delay for two weeks. Write down the approval rules that currently live in people's heads. Rank. Pick one flow. Produce: a one-page decision — what, why, who owns it, how you will know it worked.

Days 31–60: automate one flow end to end

Connect the conversation to the system of record. Automate the preparation and the movement of information. Route the decision to its owner with one tap. Log everything. No AI yet unless the step needs interpretation; most first flows do not. Produce: one workflow in production, with an owner, an error path and a baseline comparison.

Days 61–90: add AI where judgement is variable

Now add AI to the parts of that flow — or the next one — where inputs vary: classifying the enquiry, drafting the reply, summarising the thread. Define what it may read, propose and change. Keep approval gates on anything consequential. Produce: the second flow, and a written set of rules for what AI is allowed to do in your business.

Why tools-first fails

A tool bought on day one shapes the problem to fit the tool. Ninety days later you own a subscription and a pilot nobody uses. The order above is slower on paper and faster in reality.

What to do on day 91

Repeat, with the next flow on the ranked list. The list is the asset; the tools are replaceable.

Next step

If this is the problem you are dealing with, see AI consulting in Dubai, or read 10 business workflows UAE SMEs can automate before hiring more people.