Qor
Qor is a decision intelligence product from BenevolentAI Technology, the London company behind the BenevolentAI knowledge graph, sold to biotech executives, life sciences consultancies and investors. It keeps a position on each drug asset current. When a rival reads out, a program changes phase or is discontinued, or a comparable deal or financing lands, Qor works out what the event does to the asset and carries the consequence through clinical design, business development, the investor narrative, intellectual property and cash runway. Every factual claim passes a source check before delivery, and the reasoning behind each conclusion is shown.
Biotech teams get board briefings and a high impact digest, with an alert threshold they set themselves. Consultancies apply their own method across client markets, learn which client a market move hits and draft client notes and proposal outlines, while investors see which thesis assumptions an event puts under pressure across a portfolio. Invios uses Qor to plan cell therapy business development, and one case study traces Lilly's Phase 3 orforglipron readout through a competing weight maintenance asset.
A Starter plan bought online covers three asset briefings a year. Larger subscriptions run on order forms, and a free trial, a demo and an MCP connector into other tools round out the offer.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Interpretation is the product. Qor reads developments in an asset's scientific and commercial landscape, decides whether a rival's readout, a phase change, a discontinuation or a comparable deal helps or hurts the asset, and rewrites the position and its consequences for clinical design, partnering, financing, intellectual property and runway.
Briefings, digests and alerts are all outputs of that reasoning, which runs on BenevolentAI's knowledge graph and ontologies, retrieval and ranking, and language models reached through OpenRouter. Take the models away and what remains is a list of events, the thing the database tools Qor positions itself against already sell. Consultancies and investors buy the same engine pointed at client markets and portfolios.
Every output is advisory. The Responsible AI Statement of February 12, 2026, which the MSA incorporates, says all outputs require human interpretation and validation and lists what Qor does not do: execute transactions, make binding decisions, act as an investment adviser, broker or fiduciary, or replace professional judgment.
Customers set how much change in an asset's landscape triggers an alert, and each briefing shows its sources and the reasoning behind a conclusion, so an executive can check a rival's readout before changing a trial plan. Verification and the decision stay with the customer under the MSA. How the system treats a conclusion it holds with low confidence, and any point at which it holds one back for a person to check, is not described.
BenevolentAI's own models work alongside outside language models. The Responsible AI Statement says Qor may use proprietary BenevolentAI models and third party model providers engaged under contractual safeguards, and the subprocessor list routes model access through OpenRouter. Around them sit the group's scientific knowledge graph and ontologies, embeddings, retrieval and ranking, and orchestration that checks each factual assertion against its source before a briefing goes out.
BenevolentAI commits to version management of model deployments and regular review of model performance, and the MSA lets it update the platform during a subscription. Which outside models a given workflow calls, how they are chosen and what the proprietary models were trained on are not named, and customers are not given model versions.
Four subprocessors are listed by name and location on the trust portal: Amazon Web Services in the EU for storage, compute and application hosting, OpenRouter in the US for access to selected third party language models, Mixpanel in the US for product analytics and HubSpot in the US for contact data. Section 16.4 of the MSA makes BenevolentAI answerable for its subprocessors and promises advance notice of any new or replacement one through that list.
Model providers reached through the router may not use submitted data to train general purpose models unless that use is disclosed and agreed. The language model providers behind OpenRouter, and the retention terms each applies to prompts, are not named.
Invios, a cell therapy company, is the named customer: Dr. Mario Kuttke, its head of cell therapy, says Qor has been helping the company plan its next business development moves. The other testimonials are unnamed, from a single asset spinout, a metabolic disease biotech, a competitive intelligence lead at a top five pharma company and partners at two consultancies.
The three case studies, a consultancy reopening a client after an eight week engagement, a biotech preparing a financing and partnering push, and Lilly's orforglipron readout traced through a competing asset, report qualitative outcomes only. The AstraZeneca, Novartis and Merck logos on the site sit under BenevolentAI's foundations rather than Qor deployments. No accuracy figure or measured result is reported for the briefings.
Customer data and outputs stay out of training. Section 7.4 of the September 21, 2026 MSA says BenevolentAI will not use them to train or fine tune models without a written agreement signed by both parties, and outside model providers may not train general purpose models on submitted data unless that use is disclosed and agreed.
The acceptable use policy bars users from uploading sensitive personal data, and factual assertions are checked against their sources before a briefing is delivered. At the end of a subscription the customer chooses deletion or return of personal data, subject to backup retention and legal holds. No retention period is stated for prompts, outputs or backups, and there is no procedure for a wrong or unsafe model output.
Patient data is outside the product. Qor works on drug assets, markets and deals, and section 2.6 of the acceptable use policy, incorporated into the MSA, forbids users to upload sensitive personal data, the UK GDPR class that includes health data.
For the personal data it does handle, such as account details and user submissions, section 16 of the MSA sets processor terms under UK and EU data protection law: processing on the customer's instructions, subprocessor controls, security measures, help with data subject requests, breach notification and impact assessments, and deletion or return at the end of the term. A privacy notice dated February 12, 2026 covers the service itself. No business associate agreement is offered.
Certification is still ahead. The SafeBase trust portal says BenevolentAI follows industry practice and is working toward compliance certifications, and it lists no SOC 2 report, ISO 27001 certificate or penetration test summary, only the MSA, policies, a status page and the subprocessor list, all open without a request.
The MSA commits to technical and organizational measures including pseudonymization and encryption of personal data and the ongoing confidentiality, integrity and resilience of the service, and the privacy terms add encryption in transit, authentication, role based access, monitoring and logging. Staff reach customer content only for support requests, breach investigations and legal requirements. No outside party has tested or attested these controls.
Use is confined to business decisions. Section 3.2 of the MSA licenses the platform only for internal strategic, analytical and operational decision making, and an asset is defined as a drug, therapeutic candidate or medicinal product at any stage from discovery to market.
The clinical design consequences Qor draws, such as what a rival's Phase 3 readout changes for a trial plan, go to the executives who own the program, and the Responsible AI Statement calls the product a decision support tool that does not replace professional judgment. Nothing in it touches an individual patient's diagnosis, treatment or monitoring, and BenevolentAI makes no device claim for it.
A Responsible AI Statement, dated February 12, 2026 and incorporated into the MSA, sets out the governance BenevolentAI applies to Qor: access controls and authentication, monitoring of system performance, logging and audit, internal review and validation, version management of model deployments, and regular review of model performance and controls as regulation and standards change.
It warns that outputs may be probabilistic, inaccurate or incomplete and are not guaranteed to be comprehensive or error free. The commitments are policy language. No evaluation of how Qor's conclusions hold up across therapeutic areas, asset stages or thin evidence is reported, and no outside audit or AI management certification covers the product.
Each briefing traces its facts to sources and shows its reasoning, so a customer can challenge a conclusion before acting on it. Section 14.2 of the MSA warrants that the platform will operate in all material respects according to its published documentation and that services are performed with reasonable skill and care, while section 14.4 disclaims error free operation or any particular outcome and the Responsible AI Statement tells users to verify material information.
BenevolentAI's indemnity covers intellectual property claims only, and each side's liability is capped at the fees paid in the 12 months before the claim, with indirect losses excluded. A wrong recommendation about an asset or a trial design carries no remedy of its own.
The Qor MCP connector carries briefings and evidence into the tools a biotech or consultancy team already runs, and the site says the product plugs into those tools without naming any. The data Qor reasons over is market and scientific: trial readouts, program changes, deals and financings, rather than patient or study records. Outputs can be saved and shared inside the customer's organization and, under the MSA, with investors, professional advisers and transaction counterparties. No connection to a clinical trial management system, an electronic data capture system or a patient record system is offered.
One hosted option, with its location stated. The subprocessor list puts storage, compute and application hosting on Amazon Web Services in the EU, and section 16.4 of the MSA makes that list the contractual record, with advance notice of any new or replacement subprocessor. Model calls leave the region, going through OpenRouter in the US, and product analytics and contact data sit with Mixpanel and HubSpot in the US.
The privacy terms allow transfers beyond the UK and the European Economic Area under safeguards such as standard contractual clauses. No choice of region, private deployment or tenant isolation is offered.
Two routes are open. The Starter plan is bought through the online sign up, where the plan and fees chosen become the order form; it covers up to three asset briefings a year, unused briefings lapse, and it carries a flexible cancellation right.
Larger subscriptions run on order forms that set scope, fees, authorized users and term, invoiced annually in advance and renewing for 12 months unless either side gives 30 days' notice, with any increase tied to the UK Consumer Prices Index capped at 5% a year on 60 days' notice. A free trial and a 30 minute demo built on one of the buyer's assets are offered. Neither route shows a price before sign up or contact.
Three buyer groups share one interface. Biotech teams range from preclinical founders to executives with clinical stage assets and portfolios of programs; consultancies keep client markets under watch after an engagement ends and look for reasons to reopen past clients; investors monitor portfolio assets against their thesis. Assets can sit at any stage, from discovery through marketed products.
Named work spans cell therapy business development at Invios, weight maintenance after GLP 1 treatment and autoimmune landscape analysis for a strategy consultancy. No test of how well the analysis holds across therapeutic areas or development stages is reported.
Alternatives to Qor
Other products that do the same kind of work as Qor, and what each one does.
DrugBank sells Biopharma OS, an intelligence layer for pharmaceutical research, portfolio strategy, business development and competitive intelligence teams, built on the curated drug knowledgebase that began as a University of Alberta research project 20 years ago.
QuantHealth is a clinical trial simulation company with offices in Tel Aviv and Worcester, Massachusetts.
Biorce sells Aika 2.0, an AI operating system for clinical development, to pharmaceutical and biotechnology sponsors and contract research organizations.
IQVIA, based in Research Triangle Park, North Carolina, sells clinical research services, health data and technology to life sciences companies.
ConcertAI's Accelerated Clinical Trials platform helps pharmaceutical sponsors and contract research organizations design, plan and run oncology studies.
Zephyr AI develops software that helps biopharma and diagnostics teams investigate how people with cancer respond to therapies.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Starter plan online, larger subscriptions by quote
|
Qor is licensed per organization. The Starter plan is bought through the online sign up and covers up to three asset briefings a year, with unused briefings lapsing at the end of each year and a flexible cancellation right. Larger subscriptions are set on an order form covering scope, fees, authorized users and term, paid annually in advance. A free trial and a 30 minute demonstration built on one of the buyer's assets are available. | No business associate agreement. The acceptable use policy bars uploading sensitive personal data, and personal data is handled under UK and EU processor terms in the MSA. | The standard subscription terms contain no separate charge for implementation. | Vendor Published |
Standard subscriptions renew for 12 months unless canceled with 30 days of written notice. An annual increase linked to the UK Consumer Prices Index is capped at 5% with 60 days of notice unless agreed otherwise. Paid rates and subscription allowances are agreed on the order form.