Phases

What do they actually do

Phases builds an AI agent called Polly that automates patient recruitment tasks for clinical trials at research sites. When deployed, Polly connects to a site’s patient records and tools, scans for potential matches to active studies, runs outreach via phone, SMS, chat, or email, conducts automated voice/chat prescreening to collect eligibility details, and can schedule screening visits and arrange transportation. The company emphasizes that Polly plugs into the site’s existing workflow (no new logins) and operates continuously so coordinators don’t have to do every call themselves Phases site.

The team says they’re already live with customers and focused on helping sites “fill trials faster.” The product page calls out HIPAA-aware handling and integrations with existing systems; public materials position the current product narrowly around recruitment, prescreening, scheduling, and pushing data back to site systems YC company page, Phases site.

Who are their target customer(s)

• Clinical research site study coordinators / recruiters: They spend hours searching records and making long screening calls, and they miss eligible patients outside business hours. They want 24/7 outreach and automated prescreens so they’re not tied up on the phone.
• Site operations managers / site directors: They’re accountable for enrollment and site revenue but face unpredictable timelines and manual logistics (scheduling visits, arranging transport). They need automation that fits into existing tools and surfaces scheduled, prescreened patients.
• Central recruitment teams at CROs or multi‑site networks: They manage high-volume prescreening and coordination across many studies and sites, creating operational overhead. They want continuous EHR scanning and scalable prescreening without retraining local staff.
• Sponsor clinical operations / trial managers (pharma/biotech): Underperforming sites and slow enrollment delay studies and raise costs. They want predictable enrollment pipelines and fewer bottlenecks at the site level.
• Patient outreach / call‑center and engagement teams: They handle repetitive screening and scheduling with limited coverage on nights/weekends and struggle with no-shows and coordination. They want automated voice/SMS/chat outreach that can schedule visits and reduce manual work.

How would they acquire their first 10, 50, and 100 customers

• First 10: Run founder-led pilots at high-enrolling research sites via warm intros, handling setup and initial outreach end-to-end so the site quickly sees scheduled, prescreened patients without changing tools.
• First 50: Convert pilots into repeatable deals using anonymized before/after results and a referral program for site directors; sign regional CRO and site-network rollouts with a one-page integration playbook and dedicated onboarding.
• First 100: Add channel and enterprise deals: reseller/volume agreements with national CROs, site networks, and EHR vendors; build self-serve onboarding, customer success, and compliance packages to support multi-site contracts.

What is the rough total addressable market

Top-down context:

The global clinical trial patient recruitment services market is estimated at about $11.0B in 2024, with North America the largest region and continued growth expected Grand View Research. Recruitment and retention consistently rank among the top challenges for sites WCG 2024 Site Challenges.

Bottom-up calculation:

ClinicalTrials.gov lists ~66,600 recruiting studies globally as of Oct 2025. Assuming ~75% are interventional and average 8 sites per study, that’s ~400,000 site–study instances; if sites spend ~$10–15k per study on outreach/prescreening tools and services, the annualized opportunity is ~$4–6B, aligning with a portion of the broader recruitment/retention market when excluding centralized marketing and retention services ClinicalTrials.gov.

Assumptions:

• Use recruiting studies as a proxy for annual demand; ~75% are interventional and addressable by site-level recruitment automation.
• Average of ~8–10 sites per interventional study globally.
• Average site-level spend of ~$10–15k per study on outreach, prescreening, and scheduling tools/services.

Who are some of their notable competitors

• Deep 6 AI: EHR mining and NLP to find patients matching protocol criteria for feasibility and cohort identification; overlaps on patient matching but is less focused on site-embedded conversational outreach and scheduling Deep 6.
• Antidote: Consumer-facing matching and recruitment network driving patient referrals to sites via search tools and partner embeds; competes for enrollment budgets via patient marketing rather than in-place EHR scanning at sites Antidote.
• Science 37: Decentralized/virtual site operator that recruits, screens, and conducts remote visits with logistics; competes on end-to-end enrollment and patient engagement rather than embedding an agent in a site’s existing stack Science 37.
• Trialbee: Recruitment platform + agency (Honey) offering targeted digital outreach, prescreeners, and live review to deliver referrals; more of a centralized recruitment/marketing service than a site-integrated agent Trialbee.
• Elligo Health Research: Uses identified EHR data and provider networks to find ‘known’ patients and deliver vetted candidates; focuses on clinician-validated referrals versus autonomous, 24/7 patient conversations and scheduling Elligo.