Design and craft as the competitive moat in the age of AI

Executive overview

As AI makes code cheaper to produce, software differentiation shifts entirely to design and craft. Figma's CEO Dylan Field argues that "good enough" is now mediocre — to win in software you must get to excellent.

Field walks through Figma's expansion from a single design tool to a multi-product platform, the philosophy behind each product launch, and how AI is reshaping roles across product teams.

The companies that treat design as a strategic priority — not a finishing step — will be the ones that win.

Keeping momentum after the Adobe deal collapsed

• The acquisition process ran 16 months; various regulators blocked it.
• Figma continued shipping during the process rather than coasting toward a close.
• Field communicated regularly with the company, signalling clearly when "the path was narrowing."
• Post-collapse, Figma ran a programme called Detach — employees could take three months of severance with no hard feelings; just over 4% did.
• The reset allowed people who were tired or misaligned to leave, and recommitted the rest to a hard-charging pace.

Maintaining startup speed at 13 years old

• Select problems people are genuinely motivated by; move on if things aren't converging.
• Challenge timeline assumptions from first principles — distinguish real constraints from padding.
• Flatter organisations move faster; fewer layers reduces path dependency and false requirements.
• Monitor tech debt actively; systematic slowdowns often trace back to infrastructure that needs fixing before you can accelerate again.

Going from one product to two (and then many)

• FigJam was internally pushed for years before COVID made the need undeniable.
• One month before launch it felt like it was "lacking soul" — a board meeting resolved to make fun the differentiator.
• A single design sprint produced ~20 ideas; a few became definitional features (e.g. Cursor Chat).
• Going from one to two products is the hardest transition; two to N is significantly easier.
• Subsequent products — Slides, Sites, Draw, Buzz, Dev Mode — each follow a workflow-tracing logic: find what people are already doing in Figma Design, extract it into the right surface.

How Figma decides where to expand next

• Follow the workflow, not the TAM ranking.
• Figma Design itself had no obvious large market by Bureau of Labor Statistics data at founding (~250k designers worldwide); the thesis was that design would become the differentiator as software expectations rose, expanding the market.
• Each product solves a specific workflow gap: FigJam for brainstorming, Buzz for brand-consistent bulk asset production, Sites for publishing designs, Dev Mode for design-to-code handoff.
• Don't pack unrelated use cases into one surface — complexity hurts both the power user and the casual user.

Time to value and removing blockers

• Get users to an "incredible moment" as fast as possible — in Figma Design that means a collaborative multiplayer experience; in FigJam a shared brainstorm; in Make a working prototype.
• Figma once had a dedicated Blockers team that removed friction one issue at a time; retention and activation improved visibly with each fix.
• Every product needs a balance: table-stakes fixes keep users; a glimpse of the vision keeps them believing.
• Don't ship an MVP with nothing awesome in it — show where you're going even if you can't fully deliver it yet.

Figma Make and AI-assisted product building

• Figma Make lets anyone prompt their way to a working prototype or app, without needing design skills.
• Primary use case today: prototyping for product teams. PMs build their own explorations instead of delegating to designers, freeing designers for deeper work.
• Design system consistency inside Make is a current priority — ideas should be evaluated on their merits, not dismissed because they look visually inconsistent.
• Round-trip interoperability (Make → Figma Design → back to Make) is key; most AI outputs need hand-refinement to reach the final bar.
• The earlier "Make Design" launch was pulled at Config after QA revealed it could reproduce recognisable third-party designs (e.g. Apple's weather app). Lesson: wide-surface AI features require rigorous evals, not just vibes.
• Visual output quality is a core differentiator; Figma Make ranked near the top of an academic pairwise comparison study.

What good taste looks like — and how to develop it

• Taste = a point of view developed through deliberate exposure, reflection, and willingness to make judgements about what is good and bad.
• Build your repertoire: understand the canon that led to a thing, then decide where you agree or disagree philosophically.
• Few people create new aesthetic frameworks (tastemakers); many more can match and execute within one.
• The best product designers can toggle between their own taste and the brand's taste — both skills matter.
• Cross-domain exposure accelerates taste development; musicians who move into design are a recurring pattern.

How product building will change in five years

• Role boundaries are merging: 56% of non-designers already engage heavily in at least one design-centric task (up 12 percentage points year-on-year in Figma's own research).
• 72% of respondents cited AI tools like Make as a top driver of role expansion.
• 53% still believe deep knowledge is required to do a task well — AI enables generalism but doesn't replace expertise.
• AI productivity gains at Figma are "mild to moderate" on established codebases; demand for engineering headcount has not decreased.
• Design grows more important as code becomes cheaper to produce; designers who step into leadership roles will have an outsized advantage.
• The useful frame: see AI as a growth opportunity, not a cost-cutting lever.