AI fluency became a hiring filter, and portfolios have not caught up
By Allan Leone on
73 percent of design hiring managers now want proficiency with AI tools and 79 percent want experience designing AI products. Those are two different skills and most portfolios show neither.
Figma's hiring research puts it plainly: 73 percent of hiring managers see a growing need for candidates proficient in AI tools, and 79 percent say the same about designing AI products. I keep seeing those two numbers quoted as one thing. They are not.
Using AI in your process and designing a product that contains a model are unrelated skills. One is about your speed. The other is about handling uncertainty in an interface. A portfolio that demonstrates the first tells a hiring manager nothing about the second.
Nobody wants to see your prompts
The most common attempt at showing AI fluency is a case study about the tooling. Generated fifty concepts. Cut research synthesis from days to hours. It reads as a tools review, and the reviewer's next question is whether the output was any good, which the case study never answers.
Speed is table stakes now and it is invisible in the artefact. What is visible is judgement: which of the fifty you kept, and the reason, which is the part that is actually yours.
What the second number is really asking for
Designing an AI product means having answered questions that do not come up anywhere else.
- What the interface does when the model is wrong, and how the user finds out.
- How the thing expresses uncertainty without becoming useless.
- What the first run looks like before there is any data about this user.
- Where the human decides, and how that boundary is made visible.
- What it costs per interaction, and how the design responds to that cost.
A single screen showing your error and low-confidence states is worth more than a whole case study about workflow acceleration. It is also rarer, which is the point.
The honest version
If you have not shipped an AI feature, do not manufacture one. Take a product you know well, find the place a model would plausibly sit, and design the states around the failure. Say clearly that it is speculative. Showing that you understand where it breaks is the signal, and pretending it shipped is a signal too, the wrong one.
The demand is genuinely there. The filter is not asking whether you can operate the tools. It is asking whether you have thought about what happens when the thing in the middle of your product is confidently wrong in front of a customer.
Tags: careers, design, ai