You do not build an AI strategy. You rebuild each capability in your firm so that AI does the work and your people supervise the judgment, in the order your P&L dictates. A separate AI strategy sitting alongside the business is how firms end up with tools nobody uses.
This question barely existed in Collective 54's records two years ago. It has been asked 28 times and it is the fastest-rising question we track, which is the clearest signal we have that founders have moved from wondering about AI to trying to build with it.
Most AI strategies are a list of tools and a training plan. They fail the same way for the same reason: they treat AI as a topic rather than as a change in how the work gets produced. The tools get bought, a few enthusiasts adopt them, the rest of the firm carries on, and margin does not move.
AI is not a department. It shows up inside every function, which makes the real question not what is our AI plan but which capability do we rebuild first, and what does good look like when we do.
The AI-Native Boutique Firm Map organises a firm's capabilities the way a P&L already works: revenue is created, delivery cost is paid, gross margin is produced, sales and overhead are funded, and what remains is EBITDA.
Revenue minus Cost to Serve equals Gross Margin, minus Overhead, minus Sales, equals EBITDA.
That gives the sequence. Name the most pressing constraint honestly, find the capability that owns it, and rebuild that one.
One capability at a time, in a logical order, until the machine works. Not everything at once.
Read the Era Framework in full
A tool rollout has none of these. That is the whole difference, and it is why the same software produces a margin change in one firm and a shelf of unused licences in another.
If the firm has no repeatable service, sequencing capabilities is premature. Define what you sell first. And if the binding constraint is demand rather than delivery, no internal rebuild will fix it, so start with the revenue capabilities and be honest about which problem you actually have.
Do not write an AI strategy. Pick the capability where the firm hurts most, rebuild it so AI does the work and a named person supervises the judgment, prove the measure moved, then take the next one. The order comes from your P&L, not from a tool list.
With whichever capability owns your most pressing constraint, read off your P&L. If the constraint is margin, start with service design and delivery. If it is revenue, start with pricing, lead generation and referrals. If it is execution, start with operations. Rebuild one capability at a time rather than everything at once.
No. AI is not a department, it is a change in how the work gets produced, and it shows up inside every function. A standalone AI strategy is usually a list of tools and a training plan, which is why the tools get bought and margin does not move.
Four things are true of it. The work is defined well enough that someone other than its inventor could do it. The continuous part is performed by software rather than by a person remembering. A named human owns the judgment. And there is a measure that moves, so you can tell whether it worked.
Because buying a tool is not the same as rebuilding a capability. Without a defined process, software owning the continuous work, a named owner of the judgment, and a measure that moves, a tool is an addition to how the firm already works rather than a change to it.
Sources: Greg Alexander, The AI-Native Boutique Firm (Advantage Books, January 2027), for the AI-Native Boutique Firm Map and the Era Framework. The Boutique (2020) does not address this question directly; its argument that scale requires repeatable work is the precondition for everything above, but the mechanism is new.
Collective 54 is the private community for founders and executives of boutique professional services firms between $5M and $50M in revenue. Members work these answers against their own numbers.