Quarry

An AI go-to-market builder that turns what you sell and where into a ranked list of accounts worth pursuing, with the evidence for every one. Three fields go in: the offer, the market and the buyer. Quarry turns the offer into an ideal customer profile of industries, use cases, buyers and signals, plans a search matrix for the market and runs it against Google Maps, then removes duplicates and noise and checks every account against the profile. Scoring happens in code you can audit, with the reason for every point, so a model never writes the score or invents a fact. Each claim on an account is labelled: verified facts link back to their source, judgements are marked inferred, and gaps are listed as unknown so the first call starts with the right questions. The output is a ranked list with buyer roles and a CRM-ready export, built for regulated markets: the whole stack runs in your own Docker, personal data is treated as such, and it works with Anthropic, OpenAI, Gemini, a local model or no model at all in demo mode.

• Five steps from a sentence to a target account list: understand, discover, qualify, explain, act
• Scores computed in code with the reason for every point, never written by a model
• Every claim labelled verified, inferred or unknown, with source links on verified facts
• Turns 2,000 scraped listings into the 50 accounts a team can actually work
• One-command Docker stack: app, worker, Postgres, Redis and the discovery engine
• Any model or none: Anthropic, OpenAI, Gemini, local, plus a keyless demo mode
• Personal data handled as personal data, with the data staying on your infrastructure