Each agent has one job. They run sequentially, passing context to the next, until the full picture is assembled.
“Make sure we're solving the right problem before spending a cent on research.”
Rephrases your raw idea into a clear problem + solution statement, names the product, identifies target user personas, and extracts the core value proposition.
“Know exactly who you're up against — and how dangerous they really are.”
Identifies 4–6 direct and indirect competitors, scores each on threat level, and produces a competitive intensity score from 1–10.
“Understand the forces working for and against you before you build.”
Estimates market size, identifies bullish signals and headwinds, and scores the overall market opportunity from 1–10.
“Even if the verdict is tough, there's always a better angle.”
Reasons over all prior research and surfaces 2–3 concrete differentiation strategies, each grounded in the actual findings with a real-world example.
“One honest score. No hedging. No sugarcoating.”
Synthesises everything into a 1–10 rating, strengths, risks, and a brutal truth paragraph. Every field is available in Layman's terms for non-founder readers.
The pipeline is built to grow. Each new agent plugs into the same sequential flow — receives prior context, adds its layer, passes it on.
Failure Pattern Agent
v3Stress-tests your idea against historical startup failures in the same space.
Regulatory Risk Agent
v3Surfaces compliance obligations and licensing requirements before you build the wrong product.
Team Fit Agent
v4Assesses whether your background is a genuine fit for executing this specific idea.
Comparison Agent
v4Runs two ideas head-to-head through the full pipeline and delivers a comparative verdict.
Re-Analysis Agent
v4Re-runs specific agents after you tweak the idea, without restarting the full pipeline.
Agent contract
Every agent has one job, returns predictable output, and receives all prior context. If it touches external data, it handles both free and premium paths. If it fails, the pipeline continues and flags the gap.
Submit an idea and watch all 5 agents run.