Drop a discovery-call transcript in. Get a scope, risks, and a defensible estimate range — in under three minutes.
We're building Scope with a small group of software agencies. The chain improves weekly, seats are limited, and your feedback shapes what ships next.
Most agencies write scopes by re-reading a transcript at 2am. The asks are buried, the constraints are implied, the risk is in the aside. Scope reads it the way a senior delivery lead would, and gives you a draft you can defend.
Average time from discovery call to landed quote at a 12-person agency.
End up scope-creeped past statement of work — billed back as “in-scope” by the client.
Margin lost per agency per quarter to under-estimated engagements.
A single API call won't ever read a discovery call well. Five chained calls, each with a job, can. Here's the chain — same as runs in production.
Verbatim only. No inference.
Pulls explicit asks, constraints, decisions, stakeholders, and dates out of the transcript. If it wasn't said, it isn't here.
explicit_asks: [ "iOS + Android patient app", "Magic-link auth", "Spanish localization" ]
Must, should, implied — with sources.
Re-reads with a delivery lead's brain. Every implied requirement carries a confidence score and the quote it was inferred from.
{ item: "Audit logging of PHI access",
confidence: "high",
source_quote: "HIPAA-ready backend" }Open questions and defaults.
Every gap becomes a question for the client and a default assumption if they don't answer. The quote in your inbox is defensible either way.
open_questions: [ "Spanish content — translated by whom?\n Medical-grade?" ]
Components, phasing, team.
Breaks the build into components with low/high day ranges, dependencies, complexity, confidence, and a recommended team composition.
components: [
{ name: "Clinician dashboard",
complexity: "L",
days: [20, 30] }
]Ruthless. Catches what earlier passes missed.
Reviews the whole chain as if it were the agency CEO. Padding, gaps, scope-creep traps. The pass that pays for itself.
padding_or_overestimates: [ "Discovery 14–20d is high for\n a call this specified.\n 8–12d defensible." ]
The full product at a beta price while we build it with you.
Beta pricing — it goes up when we leave research beta; early users keep their price. No free tier, no free trial: we pay for every API call your scope generates, and one landed quote pays for years of it.
Anything else, the team replies within a working day.
It's sent to Anthropic's API for the chain to run and stored encrypted at rest in our Supabase. We don't train on your data and we don't share it with anyone.
Sign up, drop a call, send the scope before your competitor finishes lunch.