Tier-1 Tradesperson, Tier-2 Engineer, Tier-3 Innovator: Which One Do You Actually Need

Most bad engagements aren't bad work. They're well-executed work at the wrong tier. A buyer who needed a tradesperson hired an innovator, paid for six months of exploration, and got a deliverable that solved a problem they didn't have. A buyer who needed an innovator hired a tradesperson and got the same prompt-pack everyone else has. Tier mismatch is the most expensive scoping error we see, and it's almost always invisible until the invoice lands.

The three tiers, in plain language

Alianna Maren, writing about careers in AI, splits the field into three tiers: tradesperson, engineer, innovator. She is talking about which depth of study is load-bearing for each kind of worker, not how to scope a project. We are borrowing the same three labels for the buy side, because the symmetry is useful: if those are the three depths of practitioner, those are also the three depths of work somebody has to do for you. (Maren's original piece on tier-specific focus is here; one-line frame from us: she splits AI work into three career tiers and tells each tier what depth of study is actually load-bearing for them. Class E.)

For the buyer, the same three tiers describe the work itself:

• Tier-1 (tradesperson work). The pattern is known. The tools exist. Someone competent applies them well. A working chatbot on top of your existing knowledge base. A retrieval pipeline over your support tickets. A weekly report that pulls from five dashboards instead of you copying numbers by hand. Most "AI" buys belong here and the honest version is cheap and fast.
• Tier-2 (engineer work). The pattern is known in pieces but not in combination. Integration is the hard part. Your data lives in three systems with different access models; the model output has to feed a downstream process that already exists; the failure modes have to be caught before they reach a customer. Slower, more expensive, more durable.
• Tier-3 (innovator work). The pattern does not yet exist. You are asking a question nobody has a recipe for, and the work is partly to invent the recipe. Months, not weeks. Real research risk. Often the right answer at the end is "this isn't yet doable; here is what would have to be true." That is a valid Tier-3 outcome and the buyer should expect it.

The mismatch is almost always upward

In our intake we see far more buyers paying for Tier-3 work when Tier-1 would have closed the problem than the reverse. (Class C — from the post-mortems on inherited engagements we have unwound; the pattern is consistent.) Two reasons:

1. The vendor's incentives. Tier-3 work has a bigger invoice, a vaguer end-state, and a longer runway. A vendor who is paid by the month and scoped by the deliverable will, absent a written exit condition, naturally drift any engagement upward into the tier with more billable months.
2. The buyer's vocabulary. "AI strategy" and "transformation" are Tier-3 framings. They are not wrong, but they describe almost every problem in Tier-3 terms even when the actual unmet need is a Tier-1 retrieval pipeline that could be built in three weeks.

The downward mismatch — Tier-3 problem sold as Tier-1 work — happens too, usually as a starter pack that the buyer outgrows in month two. It is less expensive but more disappointing.

How to tell, before you buy

Three questions, in order:

1. Has anyone, anywhere, shipped a working version of this? If yes, you are buying Tier-1 or Tier-2 work. The honest vendor will tell you which one and why. If no, you are in Tier-3 territory and the next two questions matter more.
2. Where does the hard part live — in the model, in the integration, or in the question itself? Model-hard is rare and usually Tier-3. Integration-hard is Tier-2 and is most of what enterprises actually need. Question-hard ("we don't yet know what we're optimizing for") is Tier-3 and the workshop is the right first hour.
3. What is the smallest version that would tell you the bigger version is worth doing? If you can answer this in one sentence, you are ready for a Tier-1 or Tier-2 scope with a written exit condition. If you cannot, the first paid engagement should produce that sentence, not the bigger build. (Class C.)

What we do with the tiers

We accept Tier-1 work and price it like Tier-1 work. We accept Tier-2 work, the majority of what we ship, and write the exit condition into the contract so it cannot drift upward. We accept Tier-3 work rarely, only when the buyer has named the question precisely and the budget matches the months it will actually take. We will tell you which tier you are in during the workshop, and we will tell you when we think the honest answer is "you do not need us; you need a competent Tier-1 build your existing team can do." That happens more often than the rest of the industry would suggest.

If you want the four-question pre-flight that decides whether the workshop itself will be useful, see are you ready for the workshop. For the longer walkthrough of how scoping plays out, see the workshop-readiness buyer's guide. When you have a candidate problem and want to know which tier it is in, book the workshop and bring it.