RevoraX
A go/no-go verdict on a startup idea in under two minutes, where every number links to its source.
<2 min
from idea to scored verdict
3
parallel research streams
/100
score, with the working shown
The problem
Validating an idea means market sizing, competitor research and financial benchmarking — days of work, usually done badly under time pressure, or skipped.
The tools that promise to automate it produce confident prose with invented numbers. For a decision this consequential, an unsourced figure is worse than no figure, because it gets repeated in a pitch deck.
How it works
Three streams, synthesised rather than concatenated
Market sizing, competitive landscape and financial benchmarking run in parallel and are combined into one scored report — TAM, first-year revenue projection, risk scores ordered by priority, comparison against real companies with similar models, and a 30/60/90-day roadmap tied to what was actually found.
Live data, per run
Research runs against the live web each time rather than a snapshot dataset, because a competitive landscape from six months ago is a different landscape.
Every data point cited
Each number links to the source it came from. That is the whole trust model: the report is checkable, so a founder can disagree with a specific figure instead of trusting or discarding the entire thing.
Embeddable
The agent can be embedded on third-party sites, so accelerators and venture platforms can offer the assessment inside their own flow rather than sending founders elsewhere.
In the product
The scored report — a Go/No-Go verdict out of 100, with the roadmap and every figure traceable to the source it came from.
Decisions worth defending
Anti-hallucination is positioned as the feature, not a disclaimer in the footer. It is what the product is for.
A numeric score out of 100 with the components visible, rather than a verdict paragraph — a score you can interrogate beats prose you can only agree with.
Four pricing tiers from free upward, so the output can be evaluated before it is paid for.
Outcome
Shipped on the Fleapo agent marketplace. The design problem was making a two-minute automated verdict feel accountable rather than glib — which came down to citation density and showing the working.
Built with
More work
Start a conversation