An invitation, not a launch.
Evidence, knotted in.
Khipu Research Labs is an automated causal inference platform for policy evaluation. Difference‑in‑differences, synthetic control, regression discontinuity, ITS, IV, matching — in audit‑grade workflows. Cohort 01 is opening to a small group of analysts, researchers, and evaluation officers.
ContinueMost policy tools answer the wrong question — confidently.
Input‑output models tell you where money flowed. Dashboards tell you what patterns exist. Neither answers the question your legislature, your board, or your funder is actually asking: did the program cause the outcome?
The Foundations for Evidence-Based Policymaking Act of 2018 directed agencies to build the capacity to evaluate what works. OMB M-20-12 set the standards for that work — relevance, rigor, independence, transparency. Multiplier estimates and visualization layers cannot meet those standards. Not as a feature gap. As an architectural one. KRL was built to close it.
A record, by method.
Each knot corresponds to a quasi‑experimental method packaged inside the platform — with the identification strategy, assumptions, and audit trail that policy work demands. Knot height encodes the method's evidence‑tier rating.
A small group, over the next quarter.
Cohort 01 is not a public beta. It is a closed pre‑release evaluation. Wave 1 admits eight to twelve design partners; steady state across waves is twenty‑four to thirty‑two. The exchange is honest: cohort members get early access and shaping influence; the team gets rigorous use, candid feedback, and the discipline of being held to your standards.
- Who it’s for
- State and local agency analysts. University‑based evaluators. Foundation data leads. Independent consultants doing quasi‑experimental work.
- What you get
- Access to the platform, our open‑source utilities, and the connector library. A direct line to the team that built it — no tier wall. Cohort access is granted under a brief mutual NDA. We send it after we read your application; signing takes about three minutes.
- What we ask
- Real use against real questions. Notes when something breaks. Notes when something doesn’t. The willingness to be specific.
- Cost during cohort
- Zero. At general availability, cohort members in good standing receive cohort‑recognition pricing. Specific terms will be confirmed in writing before any conversion.
A short note on fit.
If you are looking for a generalist analytics dashboard, an LLM that summarizes documents, or a tool that produces a chart for a slide deck — this is not that. KRL is built for people who care whether an estimate is identified, whether parallel trends hold, and whether the result will survive a WWC or peer review.
The Knot Ceremony
Tell us what you’re trying to prove.
The application is four short sections — identity, where you sit, the methods you already use, and the policy question you’re trying to answer. Each section ties a knot in the cord. Tied cords get read.
- IIdentity
- IIWhere you sit
- IIIMethods
- IVWhat you’re proving
Approx. 4–6 minutes. Cohort 01 admits 24–32 across waves; submissions are evaluated for fit. A reply follows from a real address — not a no‑reply.