Diligence on drug assets you do not own. Every claim traces to the record it came from, and every report states what could not be established.
Two programmes in the same drug class stopped early in the same indication. One because it was failing. One because it was working too well to continue.
A screen reading trial statuses shows two failures. The distinction is in the record, and it is the difference between a dead mechanism and a live one.
This is the work. Not retrieval. Reading.
OCEANIC-AF randomised 14,830 patients and was stopped on the recommendation of the independent data monitoring committee. Stroke or systemic embolism occurred in 1.3 per cent on asundexian against 0.4 per cent on apixaban.
AZALEA-TIMI 71 also stopped early. The published account reports a significant reduction in bleeding against rivaroxaban, which is the opposite finding.
Two programmes stopped early in the same indication for entirely opposite reasons. The word terminated flattens that distinction.
Each of these came out of a report on a named public asset. Every one is checkable against the source it cites.
The question: does one Phase 3 failure kill the mechanism?
Two programmes in the class stopped early in the same indication. The registry recorded both as terminated. One stopped because the thrombosis endpoint failed. The other stopped because the bleeding reduction was large enough that continuing was not justified.
A status screen shows two dead programmes. The record shows one failed mechanism test and one that had not yet been run.
The question: what does manufacturing look like for this modality?
No release specification or product characterisation strategy exists in the public record for any programme in the class. The 2025 guidances address trial design and post-market study and are silent on release testing.
The thing a buyer would price is not merely undisclosed. It has not been established by anyone, which is a different risk and a different negotiation.
The question: will this endpoint support a label?
A comparator in the same indication moved proteinuria and missed on kidney function in the same trial, at nine times the sample size of the asset under review.
The endpoint is not unproven. It has been tested in this disease and did not convert, which changes what the readout is worth.
It sat in an appendix nobody reached, a comparator nobody checked, a claim in the deck the study report does not support.
Not concealed. Never surfaced, because the hours were not there. A typical evaluation runs a few weeks, on one or two people, against a data room nobody can read end to end. Something gets triaged, and what gets triaged is what turns up at the committee.
We produce the read your team does not have time for, and we are explicit about the edges. Where the record is thin, where a search may have missed something, and where an assumption you gave us did not survive contact with the data.
Every week spent establishing what is already in the public record is a week not spent on the questions only the sponsor can answer.
We take the public evidence base off your desk in days, so the diligence window goes on the parts that need a person in a room.
Four conventions, applied in every report without exception.
A numbered section stating what could not be established, carrying the same weight as the findings. Never a caveat at the back, never omitted. You price the gap rather than inherit it.
You tell us the assumption you are working from. We test it against the record, and where it fails, the failure is on the page rather than in a footnote.
Where a search may have missed something, the report says so and quantifies it. A count that could be higher is stated as a floor, with the reason. Nobody volunteers their own recall gap. We put it in writing.
The verdict, the risk ratings and the questions for the sponsor are yours. Not review after the fact. Those fields arrive blank, because you are the one who has to defend them.
Each claim is checked against the record it cites. What cannot be grounded is removed rather than softened, and the sentence that depended on it goes with it.
Most reports end where the evidence ends. Ours carries a numbered section for what could not be established, and leaves the judgement fields empty rather than filling them.
The class is administered as an infused product. Registered protocols in the public record do not specify a lot release step.
Manufacturing appears broadly consistent with comparable modalities, and no specific concerns were identified in the material reviewed.
No release specification or product characterisation strategy exists in the public record for any programme in this class. The current guidances address trial design and post-market study and are silent on release testing.
Nothing in the regulatory record resolves what replaces the manufactured lot. This is not undisclosed. It has not been established by anyone.
A Word document you can edit, put your letterhead on and take into the room.
Anything that hands you a conclusion has taken a position it cannot defend in the room. You are in the room. We assemble the evidence, hold the line on what it supports, and stop there.
Anyone assessing a drug asset they do not own.
Running diligence for pharma and investor clients. Capacity you bill onward, in a document that carries your letterhead.
Screening something to in-license, or preparing your own asset for a partnering conversation before someone else finds the gap.
Particularly generalist funds with biotech exposure and no scientist in the building.
Pharma and mid-cap teams screening a licensing funnel longer than the headcount allows.
Not built for clinical operations, regulatory affairs, quality or medical affairs. In those functions diligence means an audit against a standard, which is a different job and not this one.
We will tell you what nobody has proved about it. A name and an indication is enough to start.
You do not need to send us anything confidential. We work the public evidence base, so nothing of yours enters our system unless you choose to send it.