Ten questions to ask a design partner, now that every pitch looks polished.

What generative tools have made cheap is the surface of a studio: the deck, the mood board, the confident timeline. What they haven't made cheap is the substance underneath: the people who will actually do the work, the projects they stopped, the factory problems they owned. Each question below is aimed at the substance, and each one works best asked out loud.

• 1. Who, by name, will do my work? Not the pitch team, the hands. Firms that sell seniors and staff juniors pause here.
• 2. Show me something you killed. A firm that's never recommended stopping has never tested its own work honestly. Our gates exist to produce exactly these stories.
• 3. When my product reaches manufacture, who's accountable: you, or a vendor you'll introduce me to? This is the handoff question, and it's where most "end-to-end" claims dissolve.
• 4. What happens when engineering contradicts the concept you sold me? The honest answer involves changed designs and dated evidence, not heroic rescues.
• 5. Which decisions will you make with evidence, and which with taste? Both are legitimate; a firm that can't tell you which is which is running on taste alone.
• 6. What do I own when we're done: files, tooling, rationale? The right answer is everything, documented, in your systems, in formats your own tools can open and search. Anything else is a subscription dressed as a service.
• 7. Walk me through a program that went wrong. Not the anonymised triumph, but the slipped schedule, the failed inspection, what changed afterwards. Scar tissue is the credential.
• 8. How will I know, mid-program, whether we're on track? Listen for artifacts on dates (prototypes, test data, inspection reports) rather than status meetings. A weekly summary drafted by software is fine, as long as every line links to the evidence behind it.
• 9. What won't you do? A firm with no refusals has no position. (Ours are on the studio page, in writing.)
• 10. Why shouldn't I hire you? The only honest answer names the programs they're wrong for. If the answer is a smile, the diligence isn't done, theirs or yours.

Scoring the answers

You're not looking for polish. You're looking for specificity, named people, dated artifacts and comfortable admissions of limits. Ask every candidate the same ten, take notes, and weigh the pauses. Then send the winner a proper brief and judge the first response: the firm that sharpens your problem before quoting it has already started working for you.

The agent-native lens: what to ask about the agents

When the buyer sends an assistant. You may well start the search with one. Asked to shortlist studios for a connected product, an assistant reads what firms publish and compares it. Point it at the checkable parts of these ten: named people on the work, a written process with gates and what is delivered at each, a published list of what the firm won't do, ownership terms in writing. Those it can line up across many websites; a claim to be end-to-end it can't. Then lay the written answers side by side, but hold the interview yourself. An assistant can compare what was written down. It can't hear the pause, and the pause is most of the signal.

When the business runs on agents, yours or the studio's. Studios increasingly run parts of the work through agents: drafting documentation, checking drawings against requirements, chasing suppliers. That deserves questions of its own. Which parts of my program will agents touch, and which named person answers for what they produce? Where do my brief and my CAD go: which tools see them, and is any of it used to train a model? Can you show me the trail of who approved each release, and when? And if your own team runs on agents, extend question six: deliverables in native, versioned formats your systems can read, not only PDFs, so the record of why keeps working after the engagement ends.

The questions outlast the tools

Tools are likely to keep making the pitch easier to polish. They do nothing for the answers to these ten, which come only from work a firm has actually done. And yes, we expect to be asked all ten. It's a better first meeting than a showreel.