← Back to Insights

Everything You Learned About Product Development is Backwards for AI

The conventional wisdom in product development is straightforward: identify a problem, then build a solution. Business schools, startup accelerators, and innovation consultants all champion this methodology. However, this approach fundamentally breaks down when applied to artificial intelligence.

OpenAI's release of ChatGPT in late 2022 demonstrated this principle spectacularly. Rather than solving an identified market problem, they released a technology and invited people to experiment with it. The response was overwhelming - not because users needed AI-generated stories, but because the implications were staggering.

Solution-First Innovation

The ChatGPT phenomenon inverted traditional innovation logic. Instead of problem-first methodology, a new pattern emerged: solution first, problems discovered through experimentation.

This approach - called Solution-First Innovation - follows five stages:

  1. Encounter: Meet technology with curiosity rather than predetermined requirements
  2. Experiment: Play without pressure; combine tools creatively
  3. Emerge: Problems reveal themselves through actual use
  4. Extract: Identify real value created during experimentation
  5. Execute: Build proper solutions based on discovered insights

This completely reverses the traditional sequence of research, problem identification, solution design, and execution.

Real-World Example: The Language Test Tutor

A compelling case study illustrates this principle. A woman preparing for a language diploma exam ran out of practice materials. She fed existing tests into an LLM and requested new practice test generation - a simple problem with a straightforward solution.

However, experimentation revealed something unexpected. When she used the AI to grade her answers, she discovered it matched human examiner accuracy precisely. She had accidentally created a comprehensive language tutor capable of both testing and grading.

She discovered the actual problem - comprehensive test preparation and grading - through experimentation rather than market research. While she moved on after passing her exam, she could have immediately productized this solution.

What Companies Should Do Differently

Organizations beginning their AI journey frequently express uncertainty about direction. The counterintuitive response is simple: stop seeking the perfect use case and begin experimenting.

Companies succeeding with AI aren't those with superior strategy documents - they're organizations where employees have genuine permission to experiment. Someone should be able to spend an afternoon combining tools without justifying their efforts through mandatory business cases.

The most valuable AI applications discovered weren't identified through analysis phases. Rather, they emerged when curious employees worked somewhere that permitted genuine exploration.

The Real Obstacle: Culture, Not Technology

Technical AI adoption has become surprisingly straightforward. Qualified professionals can move organizations from zero to functional pilots in approximately five days. Tools exist. Capabilities are available.

The genuine challenge is cultural. Leaders must become comfortable with "experimentation and observation" as legitimate strategy. Organizations must create space for exploration without demanding advance justification. Companies need to accept that their best AI applications likely haven't been discovered yet - and won't be through consultant-led workshops, but through employee-driven exploration.

The language test story didn't originate in strategy sessions. It emerged when someone lacking materials thought, "What if...?"

The Critical Question

Are organizations waiting for perfect problems before permitting AI exploration? Are business cases required before experimentation? Are analysis phases running when experiments should?

The companies winning at AI aren't those with excellent strategies - they're experimenting fastest. They're organizations where discovering unexpected applications happens naturally.

Traditional development says: find problems, build solutions.

AI adoption says: find curiosity, let problems find you.