# Phases

> Phases deploys AI agents to oversee clinical trials, improving operational performance, data quality, and audit readiness by detecting risks and resolving issues autonomously.

Phases is a Y Combinator-backed company. Its AI agents quality-check and file trial master file (TMF) documents, review monitoring visit reports, and run automated data listing reviews.

## Pages
- [Phases home](https://phases.ai/): AI agents for clinical trial oversight and execution.
- [Blog](https://phases.ai/blog): Writing on AI in clinical trials, clinical operations, TMF, and monitoring.

## Blog posts
- [The thing no one wants to admit about AI drug discovery](https://phases.ai/blog/ai-drug-discovery-clinical-trials): AI may help produce more drug candidates. But unless how we run trials gets faster and higher quality, many of those candidates will still spend years moving through the same slow and expensive process. The industry needs more than just molecules, it needs a better way to run the trials that decide whether those molecules become medicines.
- [The Copilot Ceiling in Clinical Trials](https://phases.ai/blog/copilot-ceiling-in-clinical-trials): Clinical trials are stuck at the AI copilot stage. To run trials faster, the industry needs purpose-built agents that integrate with EDC, TMF, and CTMS systems and abide by regulations like 21 CFR Part 11.
- [Jevons paradox: Doing more with less in clinical trials](https://phases.ai/blog/jevons-paradox-doing-more-with-less-in-clinical-trials): AI will lower the marginal cost of clinical trial execution work, making continuous review across monitoring, data management, and the TMF practical instead of aspirational.
- [21 CFR Part 11 wasn't written for AI agents](https://phases.ai/blog/part-11-never-written-for-ai-agents): FDA's own guidance shows what an inspection-ready audit trail actually requires, and why that gets harder, not easier, once AI agents start making the changes.
- [You can't trust ChatGPT or Claude to run your clinical trials](https://phases.ai/blog/you-cant-trust-chatgpt-or-claude-to-run-your-clinical-trials): Even frontier AI models still guess confidently when they should admit uncertainty. What actually makes AI trustworthy for clinical trial operations is the system built around the model, and the evals that test it, not the name on the model.

## Company
- [LinkedIn](https://www.linkedin.com/company/phases-ai)
- [Y Combinator](https://www.ycombinator.com/companies/phases)
- [Trust center](https://trust.phases.ai)

## Legal
- [Privacy policy](https://phases.ai/privacy-policy)
- [Terms of service](https://phases.ai/terms-of-service)