How to find AI startup ideas using domain knowledge and market gaps

Executive overview

Most founders hunt for ideas in a vacuum. The best ideas come from direct exposure to broken workflows, vulnerable incumbents, or proven markets.

Three principles cut through the noise: exploit your own domain knowledge, target companies that cannot adapt to AI, and enter markets where product-market fit is already established.

Build where you have unfair knowledge, against competitors who cannot move fast.

Get a job or mine your past work experience

• Domain knowledge from any job or internship is a direct pipeline to startup ideas.
• Look for workflows that are manual, repetitive, or broken — these are AI automation candidates.
• Even job listings reveal what tasks AI agents could replace or augment.
• Interview processes generate market data before you write a single line of code.
• If you have no relevant experience, take a remote or entry-level knowledge-worker role and map it for automation opportunities.

Build an AI-first version of a PE-backed SaaS

• PE-backed companies follow a predictable playbook: acquire captive customer bases, raise prices, offshore dev, cut marketing.
• R&D atrophies; AI adoption stalls; customer experience degrades.
• Customers cannot easily leave — but they are unhappy and unserved by new features.
• AI-first means: fewer engineers to build an equivalent product, fewer clicks via AI agents or co-pilots, and AI-assisted go-to-market.
• These incumbents will either try to acquire you or get displaced — both outcomes are acceptable.

Build a competitor to a series B company

• Series B signals proven product-market fit at scale — the idea risk is gone.
• Series B companies carry execution risk: new executives, scaled go-to-market, tech debt, possible lack of AI-native architecture.
• They bolted on AI features rather than rethinking the product from first principles.
• You build from scratch, AI-first, against a target whose customers already exist and whose playbook is public.
• Use Crunchbase or PitchBook to find recent series B raises as a sourcing list.

Build momentum and pivot where the heat is

• No founder's first idea is the one they succeed with — pivoting is the rule, not the exception.
• A zoom-in pivot narrows a broad idea to the one feature customers actually want (e.g., Instagram dropping everything except photo filters).
• Time spent on idea selection without market contact is wasted; early customer conversations surface the real opportunity.
• Start with your best current idea, go through the motions of taking it to market, and let data dictate the pivot.