Why “tacit”.
Expert judgment, made into a system.
Tacit knowledge is what an expert knows but cannot easily write down: the cementer who always runs a retarder below a certain depth, the estimator who knows which jobs need a second truck, the planner who can tell from a form which crew to send. It lives in people, so it leaves when they do, and it does not scale.
A Tacit Engine makes that knowledge explicit one correction at a time. We start with the tables and formulas the expert already uses and encode them as a tested engine. Then we put the expert in a review loop: the engine proposes, they correct, and each correction is captured as a rule with the conditions it applied under. After a few dozen reviews, the rules say what the expert would have said, and the expert can show you why.
We are deliberately plain about what the AI does. The arithmetic is deterministic and testable. Language models help with the parts that are about reading and writing: understanding a messy job description, drafting the explanation that goes with a proposal, or finding which past job this one most resembles. The expert's rules sit above both and win every time.
Who builds it
Loren Brown has built software for thirty years, from field-data systems for engineering firms to election platforms and federal grant systems. Tacit Engine is based in Columbia, MS, and works remotely nationwide.