HomeCompaniesPhases

Phases

Automating clinical trials, starting with patient recruitment.

Phases uses AI to automate clinical trial operations, starting with patient recruitment. Our agent scans medical records and conducts human-like voice interviews to assess patient eligibility for clinical trials. Our vision is to run the entire clinical trial back-office with agents, getting drugs to market at a fraction of the time and cost it takes today.
Active Founders
James Wall
James Wall
Co-founder & CEO
Accelerating & automating clinical trials with AI Agents. - Designed and deployed AI data science agents to Fortune 100 companies and pharma companies such as J&J, Syneos Health. - Built the business case that led to Unilever’s €8bn ice cream division spin-off.
Anton Zaliznyi
Anton Zaliznyi
Co-founder & Chief AI Officer
Accelerating & automating clinical trials with AI Agents. Ex SWE at Palantir, AI Researcher at University of Toronto. - Streamlined developer and deployment experiences at Palantir. - Co-authored peer-reviewed deep learning research on real-time surgical microrobot pose estimation that outperforms SOTA (RA-L / ICRA under review). - Built autonomous reconnaissance drones at a defense startup. - Developed advanced distributed processing systems at Avid.
Jonathan van Wersch
Jonathan van Wersch
Co-founder & CTO
Accelerating & automating clinical trials with AI agents. - Previously a software engineer at Meta, where I worked on integrating machine learning into the mobile ads manager app. - Before Meta, I was at Improbable, a UK unicorn, where I helped scale the web platform team from 3 to 20 engineers. - I hold a bachelors and masters in mechanical engineering from the University of Bath, where I was project manager and engineering lead for the university's rocket team.
Company Launches
Phases: Automating clinical trials with AI agents, starting in patient recruitment
See original launch post

Hi YC! We’re Jonathan, Anton, and James - and we’re building Phases.

We’re already live with customers, helping them fill trials faster, hit recruitment targets, and prevent costly delays, without adding any additional work to research teams.

https://www.youtube.com/watch?v=10N_b2s4128

❌ The Problem

75% of people die from diseases we still can't cure (WHO).

Clinical trials are how we bring new treatments to the world, but today it takes over a decade and more than $1 billion to bring a new drug to market. Nowhere is the problem more obvious than in patient recruitment.

Recruitment is the biggest bottleneck:

  • Research sites read 1000s of medical records and run 30+ min screening calls to enroll just 1 in 10 patients, using phones, spreadsheets, and disconnected systems.
  • Patients: Wait days or weeks to hear back, face slow and frustrating screening processes, and often drop out due to poor communication and lost trust.
  • Pharma companies: Lose up to $312M per delayed trial (12.2-month average delay), with 80% of studies missing recruitment targets despite spending 30% of their budgets on it.

👷 The Solution

Meet Polly - the AI recruiter for clinical trials.

Polly is an AI agent that helps research sites find, interview, and schedule patients for clinical trials. She integrates with the tools the team already uses, so there’s nothing new to learn.

Polly handles recruitment end-to-end, with potentially eligible patients automatically landing in research site calendars for their first visit.

How Polly helps:

  • Speed: Saves team's hours per week by automating medical records reviews and turning 30+ minute screening conversations into 1-minute decisions.
  • Scale & Flexibility: Engages every patient the moment they express interest, 24/7, and scales on-demand, handling thousands of conversations at once without burdening research teams.
  • Cost: Automates the most labor-intensive parts of patient recruitment, filling trials faster at a fraction of the cost.

🚀 The Vision

Polly is just the start. We’re building the digital workforce for clinical research that will run all back-office trial operations so research teams can focus on patients, not paperwork.

For anyone living with or caring for someone with a chronic condition, one thing is clear: Developing life-saving drugs shouldn't take a decade.

👋 Our Backstory

Our founding team worked on integrating machine learning into the mobile ads app at Meta, developer and deployment experiences at Palantir, and published deep learning research in microsurgical robotics at the University of Toronto. Our CEO, who grew up around clinical trials with a mother overseeing dozens of studies as medical director at J&J, Janssen, and Schering Plough, has designed data science AI agents adopted by leading companies, including J&J and Syneos Health.

🙏 Call to Action

If you know anyone running or working in clinical trials at research sites, CROs, or pharma companies, we’d love intros!
📩 Reach out to us at founders@phases.ai

YC Photos
Phases
Founded:2025
Batch:Summer 2025
Team Size:3
Status:
Active
Location:San Francisco
Primary Partner:Gustaf Alstromer