The product
Intake AI is an agentic operating system for clinical research sites. It brings your site's documents and systems into one secure platform, answers questions with citations to the source, and automates operational work like building source documents.
Clinical research sites and the people who run studies there: coordinators, investigators, regulatory and start-up staff, and site leadership.
Intake has three parts:
- Chat. Upload or connect your protocols, SOPs, ICFs, budgets and rosters, then ask the questions no single file can answer.
- Connectors. Link the systems your team already uses, such as email, Google Drive, ClinicalTrials.gov and your CTMS, so Intake can search them together.
- Workflows. Hand off repetitive work. For example, upload a study packet and get your source documents back for review.
Your team reviews and approves the work before anything is final.
From start to finish:
- Upload the study packet: protocol, CRFs and lab manuals.
- Intake builds the source documents and the visit schedule.
- Anything unclear is flagged for a person to review.
- Your coordinator reviews and signs off.
- Approved documents are written into your eSource system.
Connecting your tools
It depends on the platform. Where a system supports it, you connect with OAuth: you sign in to that system, approve access, and can revoke it at any time. For legacy platforms without that option, our browser agent works inside your existing session.
Tell us which systems your site uses and we will confirm the best way to connect each one.
The browser agent is a Chrome extension that works on top of your active session in legacy platforms. It handles tasks like moving data between systems, reading and entering data, reviewing information, and analysis. It can only see and do what your account can.
It is designed to follow 21 CFR Part 11, with firm limits:
- It never submits, signs or finalizes work. You do.
- It never logs in or out for you.
- It never takes destructive actions, like deleting records.
See what it collects in our Chrome Extension Privacy Policy.
Trust and accuracy
Every answer is cited to the exact document or evidence in your site's data, so you can check the source yourself. When there is not enough to answer, Intake says so.
Every product and workflow also keeps a human in the loop. Your team reviews and approves the work before it is final. We believe human review is essential for AI agents in production, especially in clinical research.
Intake flags ambiguous fields for human review. It doesn't guess, and it doesn't fill the gap.
Every extracted item carries a field-level confidence score and a link back to the exact source page.
Security and compliance
Yes. Our security and compliance program is continuously monitored by Vanta, an independent third-party compliance platform.
For a security questionnaire or a BAA, write to hello@useintake.ai.
Yes to the BAA. Intake runs on HIPAA-eligible AWS infrastructure. We will sign a Business Associate Agreement (BAA) with you, and we have BAAs with our own vendors.
All PHI is de-identified using Microsoft Presidio's HIPAA de-identification toolkit, which strips the 18 HIPAA identifiers.
We also have zero data retention (ZDR) agreements with all of our software partners.
Read our Privacy Policy for the full details.
We do not train models on your data. We may use data on how the product is used, such as workflow steps, corrections and timing, to improve the product and our models.
Any PHI is de-identified to the HIPAA 18-identifier standard, and we have BAAs with all of our model providers.
Yes. Each customer's data lives in its own dedicated database, encrypted in transit and at rest. Nothing is pooled or shared across customers.
Report security issues through our Bug Bounty Program. Live service status and any incidents are on our status page. For anything else, email support@useintake.ai.
Getting started
We are rolling Intake out to early partner sites now. If you are interested, book a demo. It takes 30 minutes and there is nothing to prepare.