AI search for clinical trial documents and study data — every answer with its source
Ask any question about a study in plain words and get the answer in seconds, with the source shown.
See one case, screen by screen ↓“Which sites still need re-consent for Amendment 3?” sounds like one question. To answer it, a study manager reads the amendment for the rule, pulls ethics approval status from CTMS, pulls the informed consent page and visit dates from EDC, and checks all of it against the re-consent tracker the CRAs keep in a spreadsheet — which is already a few days old.
The answer sits across documents and data in different systems, and they do not always agree. So the number reported to the team is either late or unchecked, the gap behind it goes unexplained, and the actions for each site’s CRA are written up by hand afterwards.
Our estimate, not a measurement: one study-wide re-consent check across 88 sites, by hand and with the solution, including the person’s own checks.
Six specialist agents route, read, query, cross-check, answer and plan follow-ups; the study manager checks what matters.
VLT-302 is a Phase 3 study in plaque psoriasis: 88 sites in 11 countries, Protocol v4.0, Amendment 3 just approved. Here is one morning’s question, screen by screen, in the working solution.
Lena opens Home. Her standing questions were re-answered after the CTMS and EDC extracts arrived: 67 participants still to re-consent for Amendment 3, 12 fewer than yesterday; 74 open queries older than 30 days; 512 of 540 randomised against plan. “Needs you” holds four items, one marked “Sources disagree”.
“Blinded data only · your site access”
She types “Which sites still need re-consent for Amendment 3?” and presses Ask. Five specialists take it in turn, each step shown as it runs: the question router picks documents and data, the document reader reads 12 passages, the data analyst runs one query over 3 extracts and 167 rows, and the cross-checker compares the result with a second source.
67 participants at 7 sites still need to sign consent form v5.0. 81 of 88 sites have ethics approval, and at those sites 371 of 438 participants on study have signed. 14 participants are overdue — they had a visit after approval without signing — at Site 214 Warsaw, 131 Leeds and 305 Valencia. Seven sites are still waiting for their ethics committee.
Confidence 0.94 · 6 sources · answered from documents and data
Citation 1 opens Amendment 3’s summary of changes at page 4, section 3, Informed consent, with the sentence highlighted: re-consent at the participant’s next scheduled visit following IRB/IEC approval, and before any procedure added by the amendment. The data citations open the CTMS and EDC rows they rest on.
The CRAs’ re-consent tracker, updated 3 Oct, shows 64 outstanding; EDC shows 67. Three participants at Site 214 signed v5.0 on paper on 6 Oct and are not yet entered in EDC. The answer uses EDC, says so, and lists the three as a follow-up.
The cross-checker never hides a disagreement; an unexplained gap goes to a person.
Six plain-language steps — find the rule, sites with approval, who is on study, which consent they signed, who is overdue, cross-check — each with its citation. Below them, the read-only query that produced the numbers, run on this morning’s blinded extracts with Lena’s site access.
“Nothing in CTMS or EDC can be changed from here.”
Re-consent 6 overdue participants at Site 214 before any further study procedure, 5 at Site 131 and 3 at Site 305; enter the 3 paper signatures in EDC; chase ethics approval at 7 sites. Owners are the site CRAs and due dates the next visit; the ethics chase stays with Lena. Lena ticks them and presses Create — Priya Raman, Sam Patel, Elena Vidal and she are notified.
The planner only proposes. A person creates the follow-ups.
Lena marks the answer as checked, then files it to the TMF at 01.01 Trial oversight as a study note, with the data timestamp — CTMS 06:00, EDC 05:30. She signs with her password and the meaning of the signature: “I confirm this answer is accurate for the data stated”.
Answers are working notes until a person files them to the TMF.
Priya’s new actions sit in To do — re-consent 6 participants at Site 214 by Oct 9, enter the 3 paper signatures by Oct 8 — each marked “from Q-1042”, next to the work already in progress and done.
The same study, seen by the people who carry it day to day — what their week looked like, and what it looks like now.
Six specialist agents route, read, query, cross-check, answer and plan follow-ups; the study manager checks what matters.
Reads the question, decides whether it needs documents, data or both, and picks the smallest set of study sources that can answer it.
Searches the protocol, amendments, plans, manuals, minutes and decisions, and returns the exact passages with document, version, page and section.
Writes one read-only query per question over this morning’s CTMS, EDC and blinded IRT extracts, explains each step in plain words and returns the rows used.
Compares every number with a second source where one exists — tracker against EDC, monitoring reports against logs — and explains any difference.
Writes the plain-words answer: the number first, one short paragraph with numbered citations, and a table or chart when the answer has more than three rows.
Turns an answer into proposed actions where it shows a gap — the site’s CRA as owner, the due date from the next visit or the plan’s deadline.
Checks and files. The agents propose; a named person decides.
Anyone on the study team can ask about VLT-302, or tell it what to change. Every answer shows where it came from.
Which sites still need re-consent for Amendment 3?
67 participants at 7 sites still need to sign ICF v5.0; 14 are overdue at three sites — 214 Warsaw (6), 131 Leeds (5) and 305 Valencia (3). Seven sites are still waiting for ethics approval.
Pin late SAE reports as a standing question
Done — “Serious adverse events reported later than 24 hours, this quarter” is now a standing question on Home, re-answered every morning at 06:30 from the protocol and the EDC SAE form. Today’s answer: 2 of 41, both at Site 121.
Show me the answers that need my check
Three answers are waiting for you: the consent-deviation count where sources disagree (Q-1043), the Week 16 window (Q-1041) and the temperature excursions (Q-1036).
Create a follow-up for the paper signatures at Site 214
Created: “Enter the 3 paper signatures from 6 Oct in EDC” for Priya Raman, Site 214, due Oct 8 — linked to Q-1042. Priya has been told by email.
13 screens from the working solution, on its sample data. Pick one to see it large.
What the assistant may read, how it answers and who can see what are settings.
| Setting | Default | Choose from |
|---|---|---|
| Use superseded document versions | Only to compare | Only to compare · Never |
| Answer below this confidence goes to a person | 0.90 | 0.80 · 0.85 · 0.90 |
| Never read unblinded data | On | Locked on |
| Show the steps and query behind data answers | On | On · Off |
| Flag when sources disagree | On | On · Off |
| CRAs see only their own sites’ data | On | On · Off |
| Filing an answer to the TMF | Study manager | Study manager · Study manager or medical monitor |
| Re-answer standing questions every morning at | 06:30 | 06:30 · 07:00 · 08:00 |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample study data · “target” = the design goal, measured in the live solution · “estimated” = an estimate, before measurement; the split of days is ours · time by hand = study-team estimate · context: ICH E6(R3). People, companies and products named on this page are fictional — characters and sample data in the working solution.
A search and analytics assistant for a clinical study. It answers questions in plain words from the study’s documents — protocol, amendments, plans, manuals, minutes — and from daily CTMS, EDC and blinded IRT extracts and team trackers, with a citation or the query behind every number. It can also propose follow-up tasks for the site CRAs.
A router decides whether a question needs documents, data or both. The document reader finds the rule — for example the re-consent requirement in Amendment 3 — and the data analyst runs a read-only query over the extracts to apply it. A cross-checker then compares the result with a second source, such as the CRAs’ tracker, before the answer is written.
Yes. Treatment arm, kit-to-arm and closed-session sources are never read, IRT is read through blinded fields only, and a blinding guard strips unblinded fields from every extract before an agent sees it. The setting is locked on.
Not when “CRAs see only their own sites’ data” is on: data answers are filtered to the asker’s site access in CTMS. Every answer states whose site access it was answered with.
The answer gives both numbers and the reason they differ — in the example, 3 paper signatures not yet entered in EDC — and proposes a follow-up to close the gap. If the difference cannot be explained, or confidence is below the threshold, the answer goes to “Needs your check”.
Yes. Answers are working notes until a person files them to the TMF with an e-signature. The follow-up planner only proposes actions; a person creates them. Every question, check, filing and agent step is on the audit trail.
Yes. In the working solution three studies are connected — VLT-302, NXL-115 and CRV-210 — each with its own documents and sources, and each can be switched on or off in Settings.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your protocol, plans, TMF index and a sample of your study extracts — and it goes live once every quality gate has passed. We will walk you through it on your own study first.
A trial synopsis becomes a checked first-draft protocol in days, in your own approved language.
1-min filmTMF CheckEvery trial document is read, filed in the right place and checked, so the file is inspection-ready every day.
1-min filmCSR WriterResults sections and patient narratives are drafted within days of database lock, every number linked to its table.
1-min filmWe’ll run Ask the Study on your own protocol, plans and a sample of your study extracts.