There are two ways to buy AI work, and most operators don’t realize they’re choosing between them until they’re already locked into the wrong one.
The first is an AI Office — a senior partner on retainer, an ongoing relationship, a monthly cadence. The second is a project-based engagement — fixed scope, a defined deliverable, an end date. Both can produce results. They’re different tools for different problems. Picking the wrong one is the single most common reason AI engagements stall out in the mid-market.
This guide is for COOs, CFOs, and CEOs at founder-led, family-run, and PE-backed companies ($5M–$100M revenue, strongest fit $5M–$30M) who are deciding how to structure their first real AI investment and want a framework before they sign.
The Two Models in Plain Language
AI Office (retainer)
- Cost: $2,500–$10,000/month, depending on tier — Sherpa ($2,500), Operator ($5,000), or Embedded ($10,000).
- Length: Month-to-month, no minimum term. No lockup.
- What you get: an ongoing roadmap, monthly working sessions, small builds (at the Operator and Embedded tiers), quarterly briefings, and a clear path to scale into larger work when a specific opportunity earns it.
- Best for: building AI capability over time, running several workflows across quarters, and owner-led businesses that don’t have internal AI leadership yet.
- The promise: the retainer is priced to pay for itself — we target a 3X-or-better payback. If we’re not earning our keep, you’re on month-to-month and free to walk.
Project-based AI consulting
- Cost: typically starts at six figures once scoping, contracting, and a full build are accounted for.
- Length: a few months to a defined end date.
- What you get: a single deliverable scoped at the start — a deployed workflow, a strategy document, a platform implementation.
- Best for: one specific build with clear, real scope; a company with the internal capability to operate the result; a relationship that doesn’t need to continue afterward.
The Decision Framework
Five questions settle it. Answer them honestly before you read any proposal.
1. Do you know exactly what you want built?
If you can describe the build specifically and the scope is real, a project can work — or, more likely, a Value Sprint inside an AI Office relationship handles it, since most well-defined builds fit in a few weeks under ~$95K. If you have a sense but want to refine it, that’s discovery work, and discovery is what the retainer is built for. Start with the AI Office.
2. Will you need more AI work after this one?
If it’s genuinely one-and-done, a project is fine. If you’ll likely run more AI work over the next 12–24 months — and most operators do, once the first win lands — the AI Office produces far better economics, because the same senior partner amortizes across every build instead of being re-scoped and re-sold each time.
3. Can you operate the result without help?
If you have the internal team to run, monitor, and evolve what gets built, a project can hand it off cleanly. If you don’t — or you’re not sure — the retainer carries that operational support, which is exactly where most one-off projects quietly fail after the consultants leave.
4. How much can you commit up front?
A six-figure project has to pencil against a specific, sizeable build. If you can’t commit that kind of capital up front, the math usually doesn’t work. An AI Office at roughly $30K–$60K in Year 1 is the better fit, and it ships real workflows along the way.
5. How fast do you need to start?
An AI Office kicks off in about two weeks. Project-based engagements typically take six weeks just to scope and contract before any work begins. If you need movement this quarter, the retainer is the faster door.
The Common Path
For most $5M–$100M B2B service businesses, the route that actually works looks like this:
- Start with the AI Office (roughly $30K–$60K in Year 1) to build internal pattern recognition and ship two or three small workflows.
- Add Value Sprints ($2K–$25K each, up to ~$95K) when a specific opportunity emerges that needs a bigger build than the monthly cadence covers.
- Step up to a multi-quarter program ($100K+) when three to five related opportunities cluster into something coherent and worth committing to. This is where our 12-month KPI guarantee applies — if the agreed business KPIs aren’t met, we keep working at our cost until they are.
- Graduate to Managed Solutions when a system is in production and needs to be run and evolved on an ongoing basis.
Here’s the trap. Most companies that open with a large up-front project find themselves needing an AI Office afterward anyway. The project ends, the workflow is live, but the team needs ongoing senior judgment to keep it useful. The relationship that should have come first becomes the year-two catch-up — and you pay twice for the lesson.
Red Flags in Project-Based AI Consulting
If you’re weighing a project proposal, watch for these:
- The vendor wants to scope it with limited input from you. A build scoped to the vendor’s convenience rather than your operation is a build that misses.
- The deliverables are mostly documents. AI work that produces decks instead of running systems tends to die on the shelf.
- No KPI committed in writing. Strategy without measurement is theater.
- No plan for what happens after the end date. The deliverable lives or dies on adoption. If the vendor isn’t owning that, you are — and you may not know it yet.
Red Flags in AI Office (Yes, These Too)
A retainer isn’t the right answer for everyone, and not every retainer is honest. Watch for:
- No commitment to paying for itself. If a partner won’t stand behind a payback target, ask why.
- Annual prepay instead of month-to-month after the minimum. Long lockups misalign incentives; the relationship should have to earn each month.
- No published pricing. Quote-only forms are a red flag in any AI engagement.
- Senior people who sell but don’t deliver. Bait-and-switch is a real pattern — seniors in the sales meeting, juniors on the account. Ask who is actually in the room each month.
What This Means for Your Decision
The honest read for most mid-market operators we work with:
The AI Office is the right starting point. Lower commitment, faster start, a senior team, incentives aligned to a payback you can measure, and capability that compounds. Add Value Sprints as specific builds surface. Consider a fixed-scope project only after the retainer has produced two or three wins and you have a build too large to fit inside a Sprint.
If you’re staring at a project proposal and wondering whether an AI Office would fit better, that’s exactly the conversation we have on intro calls — bring the proposal.