Engagement 01 · AI, with no clear product yet

AI Product Prototype

From an AI concept everyone talks about to a working experience you can judge, with an honest read on whether to keep going.

The situation
An AI mandate, demos everywhere, no product
The team
You and me; ML engineers join for production
You leave with
A working prototype and a build, kill, or park call
Timeline
Weeks, not quarters, and scoped in the first conversation

The situation

Leadership wants AI in the product. The team has a demo, or three, and no product. Nobody can say what it is for, who it helps, or whether it is worth what it will cost to run.

How it works, step by step

Four steps, in order, each one something you can check off.

  1. Find where it genuinely helps

    And, just as usefully, where it does not. Most products have one or two places where this technology removes real work, surrounded by a dozen where it adds a slower way to do something that already worked.

  2. Cut to one bet worth testing

    One use, defined tightly enough to build and judge. The demos that go nowhere are usually the ones that tried to show everything at once and proved nothing in particular.

  3. Build a working experience, not a mock

    Something real people can use on real inputs, with the failure cases visible rather than staged away. A demo that only works on the happy path tells you nothing you did not already hope.

  4. Judge it honestly

    Build, kill, or park, with the reasoning and the likely cost of the next step. An engagement that talks you out of something expensive has paid for itself.

What you leave with

  • A prototype real people can use
  • What works, and where it falls over
  • A build, kill, or park call
  • The shape and cost of the next step

Working with one person

You talk to the person doing the work

No account layer, no juniors briefed second-hand, no rewriting of what you said before it reaches the designer. You explain the problem once, to the person who solves it.

The attention is undivided and it is finite

A one-person practice runs few engagements at once. That is the point: yours gets the whole of the attention. It is also the constraint, so start the conversation earlier than you need to.

Where this is not the right fit

You need production ML engineering, model training or infrastructure. I design and build the product around the technology, and work with the people who build the technology itself.

Shape, scope, and cost

Scoped to the problem in front of it, so length and price come out of a first conversation rather than a menu. That conversation is free and short. If the problem is still taking shape, we start with a smaller piece that produces clarity.

Is this the one?

Tell me what is happening with your product. If a different engagement fits better, or none of them do, I will say so.

Start a conversation