How to hire an AI product designer

Five steps, about three weeks, and roughly $2,000 to $3,000 of paid trial work per finalist. Decide the shape first — firm, contractor, or fractional lead — then screen portfolios for twenty minutes, interview one decision for forty-five, pay two finalists for a day of real work, and check references on the work you watched. Most of that budget is the fourth step, and it's the only one that reliably predicts anything.

This is the short spine. Each step below has a longer piece behind it, because the detail is where hiring actually goes wrong, and I'd rather link than repeat. Standing disclosure: I sell fractional design leadership, which is one of the three shapes in step one, and I say below when it's the wrong one.

Step 0 — Decide what you're buying

Before any of the mechanics: an agency, a marketplace contractor, and a fractional lead are three different purchases at overlapping prices, and almost everyone publishing on this topic sells exactly one of them.

Short version. An agency when you need a team on the ground and the work is well-specified. A marketplace contractor when you have a backlog of screens and the decisions are already made. A fractional lead when the constraint is judgment rather than capacity — when what's missing is someone to decide what should exist. If you can't tell which, you're probably buying capacity to avoid making a decision, which is the expensive mistake.

→ The full comparison, including when each beats the others

Step 1 — Screen portfolios for twenty minutes, then stop

Cap it. The artifact is now the cheapest part of an application — there are 203 pipeline-generated product screens on this very site, and a convincing one costs me a command — so a long portfolio review just produces a more confident wrong answer.

Twenty minutes, one question: is there any evidence of judgment here. Look for what was discarded, what constraint bit, who else was named. Skip polish, screen count, and process narrative entirely; all three are cheap now.

→ What still counts as evidence, and what stopped

Step 2 — Interview one decision for forty-five minutes

Not a list of questions — every published list ships with its answers, so anything look-up-able is rehearsed. Pick one decision from something they shipped and stay on it for the whole conversation. Judgment isn't visible until about the third follow-up.

If you ask one thing, ask what they threw away and why it was worse. A discarded direction is expensive to fabricate convincingly: it has to be good enough that a competent person would have tried it and it has to fail for a checkable reason.

→ The seven questions, and what a thin answer sounds like

Step 3 — Pay two finalists for a day of real work

The step that does the actual work, and the one most processes skip because it costs money.

Give them a real ticket in your real repo. Let them use whatever tools they normally use — no bans, no take-home brief, no invented problem. Watch how they turn the crank: how they gather context, what they cut, where they stop and ask. The output matters less than the sequence.

Budget it properly. At $350 an hour — my published ad-hoc rate, and a reasonable market anchor for senior one-off work — a day is roughly $2,800, and two finalists is under $6,000. Set against a mis-hire on a senior design role, that is not a close call. Paying also changes what you're allowed to ask for: unpaid spec work selects for whoever is least busy, which is rarely who you want.

→ How to set up the trial, and the practical objections

Step 4 — Reference the work you watched

Ask referees about the specific decision you interviewed and the specific work you paid for. Generic reference calls return generic warmth. "They told me they cut the wizard flow after the step-three support tickets — were you there for that, and what did they get wrong about it?" returns something you can use.

One question worth asking every referee: what did this person need from the team around them? Nobody is good at everything, and the answer tells you what you'd be signing up to supply.

The whole thing, on one page

Three weeks end to end if you don't let it drift, and the drift is almost always between steps 2 and 3, because step 3 needs a budget line and someone has to approve it.

When to do none of this

When the work is well-specified production. If you know what the screens are, this process is expensive theatre — a marketplace contractor and a clear spec beats it on every axis.

When you're hiring at volume. Deep single-decision interviews are hard to score consistently and easy to run unfairly. Twenty hires needs a rubric, not a conversation.

When you don't yet know what you want built. No hiring process fixes an undecided product. You'll interview well, hire someone good, and hand them a brief that changes in month two — and the failure will get recorded as a bad hire. Buy a small fixed-scope engagement first, get a written answer about what should exist, and hire against that.

That last case is common enough that it's worth saying I sell exactly that — an audit or strategy sprint, fixed scope, two to three weeks, from $6,000, ending in a written result you own. It's also the cheapest way to find out whether you need the hire at all, which is not a sales pitch I'd make if the answer were always yes. hello@product.inc, and tell me which of the three shapes you think you're buying — I'll tell you if I think you're wrong.