The UK Design Council published the Double Diamond in 2005. A simple visual description of design consisting of two sequential diamonds:

Diamond 1 — Discover & Define (the right problem)

Discover (diverge): research, understand, gather
Define (converge): frame the problem worth solving

Diamond 2 — Develop & Deliver (the right solution)

Develop (diverge): generate options, prototypes
Deliver (converge): pick a solution, ship

The logic: most startups skip directly to the solution without understanding the problem. Double Diamond forces you to pause in the problem phase before solving.

Why this matters more than ever

AI makes building solutions fast. Building used to be the bottleneck — now the bottleneck is what you build.

This means: the phase where you define the right problem is more critical than ever. If you build the wrong thing well, you waste more time in the AI era than before — because cycles run faster.

Discover — diverge on the problem phase

Classically: weeks of user interviews, market analysis, competitor research.

AI-assisted: hours. Ask AI for industry summaries, generate user interview scripts, identify 10 problems your target segment might face.

Critical: AI helps generate lists. Your work is to actually talk to people. AI can't be your user.

Define — converge on one real problem

This is where most fail. You gathered ten problems in Discover. Only one is likely the one worth building for.

A good Define output: one sentence describing the problem, with:

Target segment (who has it)
Situation (when/where it appears)
Impact (what goes wrong if not solved)
Current solution (what people do now — and why it doesn't work)

AI helps compress. Give it ten problems → ask "Which is structurally most likely to be worth solving? Reasoning." The decision is still yours.

Develop — generate solution options

The second diverge. You have a problem, now generate options.

Classic Design Sprint style: draw four different solutions. With AI: ask for 10 different solutions, each based on a different assumption ("fast integration into existing stack", "new UI", "AI agent that does it for you", etc.).

Deliver — pick and ship

The second converge. Comparison table: build difficulty, user value, competitive edge, your fit.

Pick one. Build the MVP. Ship to production.

Innovaidor and Double Diamond

Innovaidor naturally drives the whole conversation in a Double Diamond shape. You start with a seed (Discover), spar it (Define), generate methods (Develop), take it to handoff for building (Deliver). You can also activate the Double Diamond method explicitly — AI walks you through all four phases and prevents skipping straight to a solution.

Pitfalls

1. Skipping straight to Diamond 2. Most common. You have an idea, you start building. You haven't validated you're solving the right problem.

2. Discover phase too long. The opposite extreme: months of research without decisions. A time budget matters — e.g., a week at most for an early-stage founder.

3. Develop with one option only. You're not generating enough alternatives. Force yourself to list at least 5 different solutions before choosing.

Closing

Double Diamond is simple but powerful. It doesn't tell you what to build — it forces you to confirm you're building the right thing. AI massively accelerates both diverge phases. In converge phases, AI helps structure thinking, but a human always makes the decision.

Start: draw a Double Diamond for one current project. Is there a Define phase before Develop, or have you skipped it?