Primary buyer
Market Access, HEOR, and RegulatoryTeams working against dates that do not move — a coverage review, an HTA window, a label decision — for whom an institution changing its position is a surprise with a cost.
Flagship project
NextConsensus
Claims becoming harder to ignore.
You already know when a new paper publishes. What standard alerts miss is the gradual movement underneath: a claim surviving successive revisions, shedding its initial qualifiers, and appearing across sources that used to disagree. By the time an official guideline changes, the window to prepare has already closed.
The gap the work lives in
A formal label or guideline change is just the final snapshot of a shift that has been developing for years. Reconstructing that evolution by hand takes days per claim, leaving most teams to react after the fact.
A safety signal that steadily persists across revisions carries very different weight from a transient spike, yet standard alerts treat them identically. Trajectory is what reveals the difference.
What a program delivers
Available now Claim Intelligence
Claim-trajectory reconstruction against a scope agreed in writing, delivered with the revision history behind each ranking. Rebuilt from dated public revisions, each carrying a hash you can recompute against the source records used to build it.
- Trajectory reconstruction From a claim's first appearance through its wording changes, evidence, persistence, and spread.
- Historical provenance When a claim emerged, whether it was contested, and how long it held.
- Source-by-source comparison How one claim reads across drug, condition, and class pages.
- Cohort analysis Which claims are shared, which are contested, and which are unique to one drug.
- Ongoing monitoring Claims you name, with alerts when their signals change.
What a program needs from you
- One claim, drug, or class where knowing what moved would change your priorities. Starting with one is cheaper, and tells you whether portfolio-wide tracking is worth it.
- The decision it feeds — a CI brief, a safety review, an evidence strategy, a market access plan.
- A willingness to act on the answer, including when the answer is “not yet consequential.”
How it works
- 01 Extract Refract replays public revision history into dated, verifiable events. NextConsensus classifies those events into claim trajectories: wording changes, citation movement, certainty shifts, and propagation.
- 02 Classify Each event is labeled: evidence that actually moved, the public catching up, an authority's wording imported, editorial maintenance, or noise.
- 03 Prioritize Direction and strength of movement rank what deserves attention now — gaining support, shedding qualifiers, spreading across influential sources, appearing in regulator or guideline language.
- 04 Deliver Prioritized claim intelligence arrives with the trajectory evidence attached, early enough to investigate, brief, or escalate.
The classify step carries the weight. Without it, a burst of editorial housekeeping ranks identically to a claim genuinely gaining ground.
What it is built on
Trajectories are reconstructed by Refract, the open-source engine underneath (AGPL-3.0). No model sits in the path, so the same source produces the same events on every run. Its behaviour is published rather than asserted: a hash-pinned ground-truth corpus ships as a release asset, reproducible byte-for-byte from the manifest bounds.
The same properties are what let something other than a person use it. An agent reasoning about a claim needs the dated events and their sources, not a summary of them — and a result it can recompute is one it can check rather than trust.
Select a layer for detail.
Without this layer
The live public record at nextconsensus.com is the authoritative account of every forecast filed.
Current status
Three rungs, in order: what can be bought today, what is being checked against its sources, and what is built but has not yet produced a resolved result.
- Claim Intelligence Available now
Available today against an agreed scope, with the revision history behind each ranking and a recomputable hash on each source.
- Discovery In validation
Reads accumulating evidence and surfaces the authority decisions a team needs to prepare for. Eighteen entries are currently published, and a person checks each one against its source. It reports how far evidence has moved, never a probability.
- Authority-transition forecasting Research program
Registers a probability and deadline against a defined transition, freezes the evidence state, and scores the outcome against public resolution rules. The machinery and rules are built; no registered forecast has resolved yet, so public examples are illustrative.
What it does not do
- It does not tell you what is medically true. It reports what the public record did: which claims moved, in which direction, and how far.
- Discovery reports how far evidence has moved, never a probability. Probabilities belong to the forecasting research, and no registered forecast has resolved yet.
- The record is the thing that is built. A forecast is a query against it, which is why there is no forecast where the record does not yet reach.
- It does not recommend an action. The decision it feeds — brief, escalate, wait — stays with the team that owns it.
Who it is for
Mid-transaction, where the diligence question is whether a claim's institutional status will move before close.
Where the evidence record behind a claim has to be ready when an institution acts on it.
Founder and primary builder. I built the registration system, deterministic claim-scoring engine, trajectory-reconstruction pipelines, and public resolution rules on top of Refract, the open-source change-detection engine underneath it.
Continued The rest of what I'm building Ambit, Refract, and the evaluation work around them