Decide before you build
Kill it before you build it.
60 seconds. Live market evidence. A verdict you can't argue with — build it, sharpen the wedge, or kill it before you spend a weekend.
Trigvale won’t write your code, won’t generate your idea, won’t predict success — it tells you which of your ideas to kill before you waste a weekend.
Try it now · no sign-up
Paste an idea. See the verdict.
One sentence to one paragraph. We'll run it against the rubric and tell you what it actually looks like — kill, pivot, test, or build — with the load-bearing weakness called out. 3 free runs per day per IP, no email.
Or try one of these
Built for
Three founders who shouldn't waste another weekend.
Indie hacker
You ship 3 ideas a year and 2 of them shouldn't have happened.
Trigvale tells you which ones to kill before launch — with reasons you can argue with, not vibes.
Cursor / Claude Code user
You want your build agent to validate before scaffolding.
MCP server slots into your stack. The build agent calls Trigvale before it writes the first scaffolded line.
Solo consultant
You sell early-stage validation as a paid step.
Each Trigvale brief becomes a defensible deliverable: rubric, named risks, kill / pivot / test / build verdict.
How it works
From raw idea to build decision in four steps.
Idea
01Paste a raw idea — mess is fine. Founder assumptions, URLs, competitors, notes.
Validate
02Trigvale normalizes into a structured object and separates evidence by source.
Score
03Ten weighted rubric dimensions, each scored against calibrated anchors.
Verdict
04Kill, pivot, test, or build — computed in code, with reasons you can argue with.
What you actually get
Not advice. A structured brief you can argue with.
- VRS — the 0–100 score, computed in code from 10 weighted dimensions, not emitted by the model
- Per-dimension breakdown with anchor descriptions: see why each dimension scored what it scored
- Evidence rows tagged by source (observed / inferred / AI / user-claim) — never mixed
- Live signals from Reddit, GitHub, Hacker News, Stack Overflow, dev.to, refreshed every 6h
- Verdict: kill / pivot / test / build — computed in code, with the gates that decided it
- A sharpen path: which assumption to test, which dimension to strengthen, what to drop
Idea
AI-powered Slack workflow automation
VRS
31
Breakdown (excerpt)
- Pain intensity78
- Buyer urgency64
- Differentiation42
Evidence
- observed47 well-funded teams shipped this category in the last 90 days (GitHub).
- inferredBuyer is the IT admin, not the end user — sales motion is heavy.
Pure AI chat gives advice. Trigvale brings what no chat can.
A frontier model can score your idea against a rubric — that part is becoming a commodity. What it can't do is pull live market signals per idea, calibrate against your specific skill graph, see patterns across thousands of other founders' ideas, or be the validation step inside your build agent.
Pure AI chat
- — No live evidence — frozen to training cutoff
- — No persistent founder profile across providers
- — Cannot cluster ideas across users it doesn’t see
- — Cannot be the validation step inside your build flow
- — Inconsistent scoring across chats; flatters weak ideas
Trigvale
- Live web evidence pulled per idea — Reddit + GitHub + Hacker News + Stack Overflow + dev.to, refreshed every 6h
- Calibrated to your skill + distribution graph; verified GitHub repos, stars, languages, and 30-day commit cadence outrank declared signal
- Archetype clusters with corpus-derived stats and semantic neighbours (embedding cosine within the cluster)
- Callable from Cursor, Claude Code, and any MCP-aware build agent
- Opinionated rubric + deterministic kill / pivot / test / build verdict (table stakes)
About the rubric: — a 0–100 rating across 10 weighted dimensions, computed in code from per-dimension model scores against fixed anchors. Stable across runs. See all 10 dimensions →
Pricing
One unbuilt weekend pays for a year.
Same rubric on every tier. Higher tiers unlock volume, sharpen / sprint depth, and the agent-callable surface.
Or just $9 for fifteen evaluations — no subscription.
Explore
Free
3 evaluations, no card. One-time grant — try the rubric on your idea.
Organize
$25/mo
50 ideas a month, full evidence brief. About $0.50 per scored idea.
Validate
$79/mo
200 ideas a month with live evidence + sparkline trends. ~$0.40 per idea.
Integrate
$299/mo
Unlimited + MCP endpoint for your build agent. The validation step in your stack.
Three ideas on us · Your data is yours · Cancel from settings, no email required.
FAQ
The objections worth answering.
Make your next idea defend itself.
Free to try. No credit card. Three evaluations on us.