Product

Pipeline intelligence, agents and connectors, all working from the same database. The database itself is open to browse: drugs, companies, targets and US protection by year.

Agentic pharma pipeline intelligence

Every company’s pipeline, and who needs a new molecule.

Probate holds every company’s pipeline, past and present, and shows where a company needs something new. Agents (in early access) take your task over that base and answer with a report in which every claim cites the fact behind it.

Request access Read the worked example

No free tier. The worked example shows what every claim carries. Edition of 20 October 2026.

Figure 1. An agent answerdone · 4 steps · 6 sources cited

TaskWhich companies would want a new molecule for CTSK?

What the agent did

1Resolved CTSK to the protein cathepsin K and listed every drug tested at it.

2Found each company that ran one, and what became of its drug.

3Kept the companies still in the field with no active drug here, after checking the closed list of removing facts.

4Ranked them on three questions and attached the deciding fact to each line.

Answer: five companies, all route A, ranked

Merck & Co.Route A Took odanacatib to phase 3; its trial NCT00529373 ended early. The registry records: “The study was terminated early due to an observed increase in risk of stroke”. No active drug at the target. NCT00529373 · stopped 2017Approval registers · not marketed

NovartisRoute A Took balicatib to phase 2 in three trials; it went quiet in 2005. Balicatib · phase 2 · 3 trials

GSKRoute A Took relacatib to phase 1 in one trial; it went quiet in 2006. The company’s interest in this target looks lower. Relacatib · phase 1 · 1 trial

MedivirRoute A Took MIV-711 to phase 2 in three trials; it went quiet in 2018. The company’s interest in this target looks lower. MIV-711 · phase 2 · 3 trials

Ono PharmaceuticalRoute A Took ONO-5334 to phase 2 in two trials; it went quiet in 2012. The company’s interest in this target looks lower. ONO-5334 · phase 2 · 2 trials

Not yet knownWhy Merck & Co. ended its work on the drug as a whole (counted as neutral in the ranking). Drugs that no public register lists are not seen.

Fig. 1. An agent’s answer to a task, as the workspace shows it. Companies are ranked on three questions: is the biology proven in humans, how much does the company care, and is there room to act. Route A: the company ran a drug at this target, it failed or went quiet, and the company is still in the field. Grey chips name the fact each line rests on. Edition of 20 October 2026.

§ 1

One base of decided facts, three ways to work with it.

Pipelines are the base. Agents and connectors work on top of it.

Agents recommend; the facts decide. An agent never changes a route, an exclusion or a rank: those follow stated rules from sourced facts, and an agent can only cite them.

Layer 1 · Pipeline intelligence

Every company’s pipeline, past and present

By target, therapeutic area and phase, including what was stopped, sold, approved or withdrawn. Each company page shows where the company needs a new molecule, of two kinds:

Route A Failed or went quiet: the drug stopped, and the company is still in the field.
Route B Marketed, protection ending, no follow-on: a marketed drug here, no successful follow-on of its own, and the year its US protection ends.

Target, drug, trial and area pages are other views of the same evidence.

Figure 2. Merck & Co., pipeline Past and presentStoppedApprovedNeeds a new molecule

Target · drugPh 1Ph 2Ph 3ApprovedStatus

CTSKodanacatib

Stopped 2017route A

PTGS2rofecoxib

Withdrawn 2004approved 1999

Also a route A company atCALCRLPSEN1CETP

Fig. 2. A company pipeline page, shown at two targets. Bars show the highest phase reached; the end colour shows how it ended (amber stopped, red withdrawn). Source: ClinicalTrials.gov and the approval registers, edition of 20 October 2026.

Layer 1 · The clock

When a marketed drug’s US protection ends

Of the 2,016 route B rows, 768 still have US protection running. 118 lose it within two years and 256 within four. Each shows US protection ends and a year.

Figure 3. US protection ends, by year route B · nine of 256

Company at target · drug

20262027202820292030

Ends

Boehringer Ingelheim DPP4 Linagliptin

2026

Takeda PTH1R Parathyroid hormone

2027

Novartis MME Sacubitril

2027

Novartis IL1B Canakinumab

2027

Pfizer F10 Apixaban

2028

Amgen PSMB5 Carfilzomib

2028

Merck & Co. HCRTR1 Suvorexant

2029

Novartis CALCRL Erenumab

2030

AstraZeneca EEF2 Moxetumomab pasudotox

2030

Fig. 3. Route B rows placed by the year their US protection ends, from 2026 to 2030. US protection ends = the latest counted date: the last listed US drug-substance patent or exclusivity in the FDA Orange Book; for a biologic, its reference-product or orphan exclusivity in the Purple Book; where neither is listed, 12 years from the first US licence, shown as computed. A hollow dot is a computed year. Not the date generics or biosimilars can enter.

Layer 2 · Agents

Tasks you give it, reports that show their sources

An agent takes a task over the pipeline base and returns a report. Every claim in the report cites the decided fact it rests on, and says what is not yet known. Agents are in early access, in the Institutional plan.

Find the companies that would want a given molecule
Pipeline due diligence on a company
A target landscape, and where there is room
Watchlists with explained alerts
Outreach briefs
Which companies lose US protection on a marketed drug within four years with nothing to follow it?

Figure 4. An agent task Done3 claims · 2 cited · 1 open

Companies for a moleculeDue diligenceTarget landscapeWatchlistOutreach brief Brief for a first call with Novartis about FDPS.

Cited

Novartis holds a marketed drug at FDPS: pamidronic acid, approved in 1991.Approval register · 1991

Cited

No later approved drug of its own at FDPS is on record for Novartis.Approval registers · 20 Oct 2026

Open

Zoledronic acid anhydrous is a later drug at the same target. Whether Novartis originated it is not yet established, so the follow-on is marked unclear.Open · originator of zoledronic acid anhydrous

Fig. 4. A task card from the workspace: the type of task, the claims in the brief, and the fact behind each. A claim that rests on something not yet established is shown as open, not smoothed over.

Layer 3 · Connectors

For your own agents

An MCP connector and an API let your own agents and systems call the same base. The connector reads the published editions, so an answer names the edition it came from.

Included in the Institutional plan.

Figure 5. A tool call and its cited answerMCP · edition 2026-10-20

Call

tool: find_companies

{

"target": "CTSK",

"route": "A",

"limit": 2,

"edition": "2026-10-20"

}

Answer

{ "target": "CTSK",

"results": [{

"company": "Merck & Co.",

"route": "A", "rank": 1,

"claims": [{

"text": "Phase 3 trial of

odanacatib stopped 2017",

"cites": ["NCT00529373"]

}],

"not_known": ["why the

company ended its work

at the target"] }, … ] }

Fig. 5. One tool call from a customer’s agent and the shape of the reply: the ranked company, the claim, the identifier of the fact it cites, and what is not yet known. The tool name and fields are illustrative.

§ 2

A worked example, read from the first trial to the last line.

From 604,331 trials to 4,815 company-level needs.

The case study follows the registered trials down to the drugs, the targets and the companies, and shows where each of 12,650 company-target rows ends. 4,815 of them are needs: 2,799 on route A and 2,016 on route B. Each of the other 7,835 rows ends some other way, with the fact behind it.

It then traces one row end to end, Merck & Co. at cathepsin K, with the sources, the dates and what is still open, and shows an agent using it.

Read the case study

Figure 6. Where 12,650 company-target rows end

Route A2,799

Route B2,016

All other rows7,835

Fig. 6. Route A and route B rows are ranked. Nothing is removed except by one sourced fact from a short, fixed list. Sources: ClinicalTrials.gov (monthly), FDA Orange Book and Purple Book, Drugs@FDA, European and national approval registers, ChEMBL 37. Edition of 20 October 2026.

What we do not know Some rows are undetermined: a fact that would decide their route is not yet established, and each row names it. Why a drug stopped is not yet judged by hand, so it counts as neutral. The method and its limits.

§ 3

Who gives the agents their tasks.

Who it is for

Pharma business development

Find which companies hold a stopped drug at a target and still have a reason to return.
Ask an agent for the companies that would want a molecule you hold.
Open any claim to the trial, the date and the source behind it.

Search and evaluation

Check a target you were handed against every company that has already tried it.
Get a due diligence brief on a company, with what is open marked as open.
See how far each earlier drug went and why the registry says it stopped.

Biotech founders

See who tried your target, who is still active, and who has a reason to come back.
Build the outreach list from the companies that need a new molecule.
Take an outreach brief into the first call.

Investors

Trace a company’s stopped, sold and withdrawn drugs across everything it has run.
Spot marketed drugs whose owner has no successful follow-on.
Keep a watchlist and read why each alert fired.

Research strategy

Find targets where the biology is proven in humans and few other companies are active.
Compare therapeutic areas by how many of their targets have a company with a reason to return.
See the questions a ranking weighs, so it can be argued with.

Consultants

Answer a client’s question about a target or a company in an afternoon, with the sources attached.
Run the same task across a list of targets.
Export the evidence behind every line.

The same evidence for all

Whoever asks, the answer carries the same trail: the route, the facts that decided it, the source and its date, and what is still open.

§ 4

Priced per organisation, per year. There is no free tier.

Every plan shows the full evidence behind every claim.

Explorer

$2,500per year

3 seats · a first look

Every company pipeline, with the evidence behind each row
Export up to 100 rows a month

Request access

Professional

Recommended

$9,500per year

5 seats · a team working every week

Everything in Explorer
Saved lists
Unlimited export: CSV, Excel, XML and JSON

Request access

Institutional

$19,500per year

15 seats · a group that builds on it

Everything in Professional
Agents in early access: questions, tasks, watchlists with explained alerts and briefs
The MCP connector and the API
Your own ranking weights, and edition briefings

Request access

Need more? Talk to us about a private deployment, or custom agents that run on your own target list. Read the worked example.

Bring a task. We will run it with you.

In 30 minutes we give an agent the target, company or area you are looking at, and read the answer together: who has tried it, what stopped them, and what is still open. We reply within one working day.

Request access Worked example