UPDATE – April 2026

We take data privacy and patient protection seriously. Following feedback from compliance-focused readers in the clinical research community, we’ve updated this post to more precisely reflect AI vendor data training and retention policies, particularly distinctions between consumer and enterprise plan tiers. If you are using or evaluating AI tools for site operations, please review the updated Data Governance section below before proceeding.

Clinical research sites are moving fast on AI. Across the industry, teams are looking for ways to use large language models to forecast enrollment, analyze visit trends, and surface operational insights from their own data.

The challenge? Most site technology platforms don’t give you a way to connect your data to AI tools. Your data is locked inside proprietary systems with no clean path out.

CRIO is different. Every CRIO customer already has access to a modern data infrastructure, powered by Google BigQuery and Looker, that makes your site data AI-ready today. No custom integrations. No data engineering project. No waiting for a vendor roadmap.

This post walks through exactly how to connect your CRIO data to an AI assistant like Claude, and why CRIO’s architecture gives you a head start that other platforms simply can’t match.

Why This Matters for Research Sites

AI tools are only as useful as the data you feed them. If your operational data lives in a system that doesn’t expose it cleanly, you’re stuck. You end up manually exporting spreadsheets, copying and pasting, or waiting months for your vendor to build an integration.

CRIO’s data platform eliminates that bottleneck. Your visit data, SiteApp activity, enrollment metrics, and operational data all flow into BigQuery, and Looker gives you a governed, user-friendly way to access and schedule reports from that data.

That means you can connect your data to AI tools right now, on your terms, while maintaining full control over what gets shared and how.

Two Paths to Connect CRIO Data to AI

There are two approaches, depending on your team’s technical comfort level and what you’re trying to accomplish. Most sites will find the Looker path faster and easier, but BigQuery offers more flexibility for advanced use cases.

Comparison at a Glance

Looker (Recommended)BigQuery (Advanced)
Best forBusiness users, operations leads, site managersData analysts, technical teams, custom reporting
Technical skill requiredNone. Point-and-click setup.Moderate. Requires BigQuery access and familiarity with SQL or API connectors.
Setup timeMinutesHours to days, depending on approach
Data controlYou choose exactly which Looks (reports) to share. Nothing else is exposed.Full dataset access. Potentially requires BAA with your AI provider, and/or careful scoping to avoid exposing sensitive data.
PHI considerationsStrong. You control the data at the report level. Only share de-identified or aggregated Looks.Requires more care. BigQuery contains all data available in your account, so you will need to take appropriate precautions such as having your AI LLM provider sign a BAA, managing access controls, and ensuring queries exclude PHI.
AI connection methodSchedule Looker exports to Google Drive, then connect your AI tool to that folder.Use the AI tool’s BigQuery connector (if available) or export query results to a shared location.
AutomationBuilt-in. Looker schedules keep your data fresh automatically.Possible, but requires scheduled queries or external orchestration.
FlexibilityLimited to existing Looks and reports.High. You can write any query against your full dataset.

Our recommendation: Start with Looker. It’s faster to set up, easier to manage, and gives you strong data governance out of the box. If you outgrow it or have advanced use cases, BigQuery is always there.

How to Set It Up: The Looker Approach

Here’s a step-by-step walkthrough using Claude as your AI assistant. The same general approach works with other AI tools that support Google Drive or file-based input.

  1. Identify the Looker Look you want to analyze. This could be a SiteApp visit summary, enrollment tracker, revenue forecast, or any existing report. If you need a new Look, your CRIO account team can help.
  2. Schedule the Look to export to Google Drive. Open the Look, click the gear icon, and set up a scheduled delivery to a dedicated Google Drive folder. Choose a refresh frequency that matches your needs (daily, weekly, etc.).
  3. Connect your AI tool to that Google Drive folder. In Claude, for example, you can create a Project or use Cowork and point it to the Drive folder. Instruct it to always use the most recent file.
  4. Ask questions. Once connected, you can prompt your AI assistant to summarize trends, forecast metrics, flag anomalies, or generate shareable dashboards from your data.
  5. Iterate and expand. Start with one Look, prove the value, then add more. Each additional data source you connect makes the AI more useful.

The entire setup takes minutes, and because Looker handles the scheduling, your AI assistant always has access to up-to-date data without any manual effort.

What Can You Do With It?

Once your CRIO data is connected to an AI assistant, the use cases are wide open. Here are a few examples sites are already exploring:

  • Revenue forecasting from SiteApp visit data
  • Enrollment trend analysis across studies and sites
  • Operational dashboards generated and shared as interactive artifacts
  • Data quality checks that flag missing or unusual values in your trial data
  • Ad hoc analysis in plain English, without writing SQL or building custom reports

A Note on Data Governance and PHI

Any time you connect data to external tools, governance matters. The Looker approach gives you strong, practical controls:

  • You choose exactly which reports to export. No bulk access, no open pipelines.
  • Use aggregated or de-identified Looks to avoid exposing PHI.
  • Google Drive access controls let you restrict who can see the exported files.
  • AI tool data policies should be reviewed carefully to confirm how data is handled — and the details matter more than you might expect.

Not all plans from the same vendor offer the same protections. This is a critical point that many sites overlook.

Take Anthropic’s Claude as an example. On commercial and enterprise tiers, Claude for Work and API access, Anthropic does not train on customer data. Those tiers come with contractual commitments that make them appropriate for sensitive operational use. However, consumer plans (Free, Pro, and Max) may use your inputs for model training unless you explicitly opt out. The same tool, very different data handling depending on how you’re accessing it.

This distinction isn’t unique to Anthropic. It’s the norm across the industry. Before connecting any clinical research data, including operational metrics, enrollment data, or anything that could intersect with PHI, confirm the following:

  1. Plan tier and training policy. Are you on an enterprise or commercial plan with explicit contractual protections against model training on your data? Do not assume a paid subscription is sufficient — verify the specific tier and its data use terms.
  2. Data retention policies. How long are your inputs stored? Is data deleted after a session or retained? Where does it reside, and under what jurisdiction?
  3. BAA availability. For anything touching PHI, a Business Associate Agreement may be required. Confirm whether your AI provider will execute one, and under which plan tiers.

The bottom line: plan tier is not a minor configuration detail, it is a compliance decision. If your team is accessing an AI tool through a consumer account, even a paid one, you may not have the protections your organization requires. Use enterprise-grade plans with clear, written contractual commitments before connecting any site data to an AI assistant.

Why CRIO Customers Have a Head Start

This isn’t a feature on a future roadmap. It’s available to CRIO customers right now, using infrastructure you already have.

Most competing platforms don’t offer anything comparable. They lack the modern cloud data layer that makes this kind of integration possible. Their data is locked inside closed systems, and connecting it to AI tools would require significant custom development, if it’s possible at all.

With CRIO, your data is already structured, governed, and accessible through BigQuery and Looker. That foundation makes AI adoption a practical, low-risk next step rather than a major IT project.

Ready to Get Started?

If you’re a CRIO customer interested in connecting your data to AI, we’re here to help. Reach out to your Customer Success Manager or [email protected] and we’ll walk you through the setup. It’s faster than you think.

Want to see this in action? Join us at CRIO Connect Live in Raleigh, North Carolina, where we’ll be demoing how sites are using AI today in their practices. We’d love to see you there.

Your site’s AI strategy starts with the data you already have.

Blog

Sponsor-Provided Source Templates: Your Questions Answered

January 14, 2026

Give sites what they need and watch your study succeed.

Schedule a Demo