The operating system for medical & academic research
Every decision here has a name on it.
Dr Costa is an operating system for medical and academic research: one governed path from a plain-language idea to a drafted protocol, an ethics pack, participant consent and data collection, and a locked, exportable dataset.
For the universities, hospitals, research groups and funders that answer for research; and for the investigators who run it. AI drafts; a qualified person signs; an append-only ledger keeps the record. And because no one can quietly rewrite it, the name is protection: the record of what you decided — and what you didn’t.
Free for students · Researcher Pro £39 a month · every price published.
Starting a study starts the paperwork, not the research: recruitment stays closed until the study’s approvals and its named safety contact are signed into the record.
Researchers this way. Participants this way.
Who this is for
Built for the people who answer for research.
Different desks, same problem: research scattered across tools nobody chose is hard to oversee and expensive to defend. Dr Costa gives every study the same signed path, and the people responsible clear sight of it.
Universities & departments
A department’s research lives in fifty spreadsheets and six survey tools. The audit trail lives nowhere. The research office answers for all of it.
One governed path — so when the question comes, the version, the signature and the history are already there.
Hospitals & NHS R&D offices
Clinician research runs on the edges of clinic time. Every workaround an investigator invents becomes your data-protection problem.
A governed route built to fit inside clinic time. Your office gets what it is asked for: versions, signatures, history.
Research groups & networks
Across a programme, instruments fork and consent versions blur. You find out at pooling time — when it is too late to fix.
Locked protocols and versioned instruments keep sites comparable. Every dataset leaves with its data dictionary and audit history. The version is the version.
Sponsors & funders
Funding investigator-led research means carrying its risk without holding its systems, and hearing about drift when the report is late.
Clear sight of how each study is actually run, while the study team keeps ownership of the work. Sight, not control.
The shape of a study
One study, in five signed steps.
Dr Costa is a single chain. The protocol your ethics pack cites, the consent your participants sign and your team countersigns, and the dataset you lock are always the same study. Between systems is where versions slip; here there is no between.
Describe
Start with the question
Say what you want to learn, from whom, and what you will measure, in plain language. That is enough to open a study.
Draft
Shape the protocol
An AI assistant drafts protocol sections; you review, edit and sign each one in. A draft stays labelled a draft until a person accepts it.
Prepare
Assemble the ethics pack
Participant information, consent documents and supporting papers are drafted from the locked protocol — the pack and the protocol cannot quietly disagree. Committees rarely refuse ideas; they query inconsistencies.
Consent & collect
Bring participants in
Consent is a signing ceremony, not a checkbox: the participant signs, the study team countersigns, and every version is kept. Participants record what the protocol asks, in an app designed around the person holding the phone.
Lock & export
End with a dataset that holds
When collection ends, you lock the dataset. It leaves with its data dictionary and audit history; the analysis happens in the tools you already trust.
Architecture, not badges
Built to be questioned.
Trust in research software should be mechanisms you can inspect and refusals that hold.
AI drafts. People sign.
AI drafts. A qualified person reviews, edits and signs — nothing written by AI becomes part of a study without a named human signature.
Every AI draft is stored with the model that produced it, and stays labelled as AI-suggested until a person accepts it.
A ledger, not a logfile
Every state change writes an audit event to an append-only, hash-chained ledger, in the same transaction as the change itself.
Append-only by construction: the database role that writes the ledger cannot update or delete.
Identity, kept apart
Names and contact details belong in a separate identity vault with its own keys. Participants appear in the research store only under a code — it has no participant name, email or date-of-birth column.
Data is encrypted in transit and at rest, with identifying data under its own keys.
UK-based, said precisely
Runs on UK-based infrastructure: the deployment region is AWS London.
AI inference runs through EU-resident endpoints — never a global route.
Approval is a click. A signature is a person.
What Dr Costa will not do.
Design decisions — and they hold even where the feature would be easy to build.
- No silent AI
- It will not publish, send or file anything a person has not reviewed and signed. There is no path around the signature.
- No AI near symptom reports
- No AI reads, scores or sorts what a participant reports about their health. That path has no AI in it at all.
- No verdicts about your data
- In-product analytics are descriptive only. Statistical inference stays in your own tools, working from a clean export. An AI model never computes a number: where the platform calculates, versioned deterministic libraries do the arithmetic.
- No quiet reopening
- Locked protocols, signed consents and locked datasets never silently reopen. Unlocking is scoped, dual-authorised and attributed.
Why this exists
Dr Costa began in clinic, with families managing food allergy at home — and with research that still ran on paper forms, spreadsheets and goodwill.
Dr José Maia e Costa
Founder · Consultant Paediatrician in Allergy
Read the founder’s storyThe other half of every study
The same care for the people in your study.
The design floor is a cautious 67-year-old at her kitchen table: WCAG 2.2 AA minimum, legible at 200% text on a small screen, keyboard-operable throughout — no dark patterns, no streaks, no gamification.
Complete, truthful datasets begin with participants who are treated well. No streaks, no badges, no countdown timers — people stay because the study deserves it.
Pricing
A price one person can authorise.
Free for students. £39 a month for an investigator. £29 a seat for teams of five or more. Institutional licences priced with you — and every number is public.
Fair questions.
- My institution already gives me a system — why pay?
- Keep it; see above. Dr Costa is the whole signed journey — drafting, ethics pack, consent, collection, locked export — and it works without your institution’s queue. The free tier exists for exactly the studies institutional systems leave out.
- Does this replace ethics approval?
- No. Nothing replaces your ethics committee. Dr Costa drafts and assembles the pack you take to it, and keeps every document consistent with the signed protocol. The decision is theirs, as it should be.
- Who owns the data?
- The study does — which means you and your institution, under your approvals. Every tier can export the full dataset, its data dictionary and its audit history, at any time.
- What happens if I cancel?
- You leave with everything: dataset, dictionary, history. Locked things stay locked — cancelling your subscription never unlocks or deletes a study record.
- What does the AI never touch?
- What participants report about their health. That path has no AI in it at all — and no AI on this platform computes a number or files anything unsigned.
- Can my R&D office review this before I start?
- Please. The governance page is written to be forwarded, and a demonstration with your reviewers in the room is the meeting we most enjoy.
- Is VAT included?
- Prices exclude VAT. Individuals pay by card; institutions can pay by invoice or purchase order.
Next step
Start your first study.
Sign up, describe the question, and the signed path is yours — protocol to locked dataset, with your name where it matters. If you would rather see it first, bring your hardest reviewer to a demonstration.
Free tier for students — academic email required. No card needed to begin.