TrialSight

Clinical intelligence for pre-commercial biotech — enterprise insight without the enterprise contract.

The Problem

A 30-person biotech with two assets in Phase II needs to know the same things a 3,000-person pharma needs to know: who the key opinion leaders are in their indication, which competing trials are about to read out, whether their endpoint design has been tried and failed before, and what the realistic probability of success is.

The tools that answer those questions — Citeline, Evaluate, Informa — cost $150K–$250K per year. That is a full headcount for a company that has three. So the work gets done the other way: an associate runs manual PubMed searches, maintains a competitive-landscape spreadsheet that is stale the week it is built, and the company makes eight-figure development decisions on nine-figure assets using a document nobody trusts.

The data is not the bottleneck. Nearly everything those platforms sell is downstream of public APIs that anyone can query for free. What they sell is aggregation, cross-referencing, and analyst time — and that is exactly the layer AI collapses.

The Solution

TrialSight aggregates 10 public biomedical data sources into a single dashboard and runs AI analysis across them: KOL identification, competitive landscape monitoring, and calibrated trial outcome forecasting. No proprietary data licenses. No scraping. Every source is a documented public API.

The result is 80% of the insight of a $200K platform at 10–20× lower cost, aimed at the segment those platforms have priced out entirely.

The 10 Data Sources

Every record ingested from every source is stamped into an append-only provenance chain (see *Stack*), so any number on the dashboard can be traced back to the source response that produced it, at the time it was fetched.

KOL Intelligence

For any therapeutic area, TrialSight builds a ranked list of key opinion leaders by cross-referencing five independent signals rather than counting publications:

• >Publication volume weighted by citation impact (PubMed + OpenAlex)
• >Trial involvement as PI or co-investigator (ClinicalTrials.gov)
• >Grant funding history and award size (NIH RePORTER)
• >Institutional affiliation and co-authorship networks (OpenAlex)
• >Conference and editorial activity (Europe PMC)

Output is a ranked profile set with engagement scores, collaboration maps, and contact pathways — the deliverable medical affairs teams currently commission from consultants at $50K+ per engagement, regenerated on demand.

Ensemble Trial Forecasting

Every monitored trial carries a **TrialSight Score (0–100)** — a calibrated probability of clinical success produced by a 17-specialist AI ensemble. Each specialist reasons over a different axis:

• >Historical phase-transition rates for the indication
• >Endpoint design quality and statistical power
• >Competitive landscape and differentiation
• >Sponsor track record and runway signals
• >KOL involvement as a correlate of trial execution quality

Because the score is an aggregate of independent specialists, no single model's bias dominates, and each specialist's reasoning is exposed alongside the number. A forecast a scientist cannot interrogate is a forecast a scientist will not use.

Competitive Landscape Monitoring

Continuous alerting on the events that change a development plan:

• >New trial registrations in your indication or mechanism
• >Competitor status changes — enrollment opened, results posted, terminated
• >New publications from competing programs
• >FDA decisions on related assets
• >Grant awards to competing research programs

Alerts are scoped by therapeutic area, mechanism of action, indication, or named competitor.

Market Timing

Three things are true at once for the first time:

1. **The public data is complete enough.** ClinicalTrials.gov, OpenAlex, and Open Targets have all matured into full-coverage APIs within the last few years. The proprietary premium is now mostly aggregation, not access.

2. **The aggregation layer is automatable.** Cross-referencing ten schemas and reading thousands of abstracts was the analyst headcount that justified the price. It no longer is.

3. **The customer segment exists and is growing.** Post-Series A biotechs with active clinical programs and 10–50 employees are the fastest-growing slice of the industry and the one incumbents have priced out by design.

Competitive Position

We do not win by being cheaper at the same thing. We win by serving the companies the category has decided are not worth serving, and by being auditable in a way an analyst-curated database structurally cannot be.

Revenue Model

Marginal cost per tenant is API calls and inference, not data licenses — the reason the pricing gap is structural rather than promotional.

Stack

Built on four Rising Sun platform components rather than assembled from scratch:

• >AAASP — multi-tenant agent deployment (Elixir/OTP) runs collection, transformation, and alert pipelines per tenant
• >OpenMed — unified biomedical API integration layer across all 10 sources
• >Ensemble forecasting — the 17-specialist system that produces TrialSight Scores
• >Etch — namespace-isolated system-of-records provenance; every ingested record and every forecast carries a Merkle receipt

Python, FastAPI, PostgreSQL, deployed as an AIOS service.

**Rising Sun** · [risingsun.name](https://risingsun.name) · [trialsight.online](https://trialsight.online) · August 2026

*Every number traceable to the API response that produced it.*