Company profile · 3 min read
Astraea: Sponsor-run automation for clinical biometrics and submissions
Astraea automates SAPs, CDISC datasets, TLFs, QC and FDA submission documents inside a sponsor’s environment with traceable human review.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Astraea is sponsor-run clinical-trial automation for the post-protocol biometrics workflow. It fits Phase II/III sponsors that need SDTM, ADaM, TLFs, QC and submission documents produced faster while keeping data inside the sponsor’s environment and every output reviewable.
What it does
Astraea’s current site describes agents that automate SAPs, eCRFs, CDISC mapping, SDTM and ADaM datasets, tables/listings/figures, Define-XML, QC and submission documents. The software is installed inside the customer’s environment; Astraea says it does not see or hold the clinical data. The product is explicitly sponsor-run, not a CRO that takes over trial operations.
The useful buyer decision is whether the bottleneck is repeated standards and programming work after raw data collection. A study team still owns scientific judgment, validation and regulatory sign-off; Astraea is selling a traceable execution layer. The public platform page says every action is logged, versioned and reviewable, and that workflows are built around CDISC and 21 CFR Part 11 controls.
The company reports 30–50% faster biometrics cycles, 12–30 days removed from post-lock timelines and 100% CDISC-ready outputs. These are company claims, not independent trial-program results. The YC launch says work that normally takes around nine months with teams of five to ten can be compressed into days in its workflow. A sponsor should test one study and one deliverable rather than use the top-line number as a planning assumption.
Founder context and tradeoffs
The YC profile identifies Joshua Wang as CEO, with Stanford computer-science and mathematics training, and Sanmay Sarada as CTO, with clinical data engineering and fetal-medicine publication experience. The about page says the team includes biostatisticians, clinical programmers, AI engineers and former sponsor-side leaders.
Pricing is not public. The buyer should ask about validation packages, change control, audit trails, customer-environment deployment, model updates and exactly which human approvals are required before a submission artifact is used.
Editorial take
I would shortlist Astraea for a sponsor with a live study, a clear biometrics bottleneck and a team willing to own review. The right proof is one post-lock workflow with traceability from raw input to approved output. If the team is looking for a generic AI assistant rather than regulated execution, this is the wrong category.
Quick facts
| Field | Sourced detail |
|---|---|
| Product | Sponsor-run clinical biometrics and submission automation |
| Buyers | Pharma, biotech and clinical-trial sponsors |
| Workflow | SAP, SDTM, ADaM, TLFs, Define-XML, QC and submission documents |
| Deployment | Installed inside the sponsor environment; data-retention claims attributed |
| Pricing | Not publicly listed |
Sources checked
| Source | Checked |
|---|---|
| YC profile | 2026-09-19 |
| Astraea homepage | 2026-09-19 |
| Astraea about | 2026-09-19 |
| Astraea launch | 2026-09-19 |
Cohort context
Astraea is listed in Spring 2026. In our 2026-09-18 directory snapshot, 112 of 193 listed companies in that cohort have YC’s primary industry label B2B (58.0%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:16:06.847Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Observed |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗