← all environments

Competitive Trial-Landscape Analysis

Clinical / CICompetitive-intelligence analyst

Given an indication and optionally a target or modality, survey the interventional-trial landscape from an AACT-shaped database and produce a competitive read: who is running what, how designs differ (phase, line, control arm, endpoint), which sponsors lead, and where the modality x line-of-therapy whitespace is. The agent has read-only trial-query tools and must pull and reason itself.

Why this is fundable

Scarce expert who grades this
Oncology competitive-intelligence analyst / MD or PharmD with trial-design fluency (~$200–350/hr loaded)
What one decision is worth
Steers $50M–$500M development bets: which line of therapy, control arm, and endpoint to run — a mis-read landscape funds a trial that's already obsolete or duplicative.
Real-world data sources
ClinicalTrials.gov / AACT (studies, interventions, conditions, sponsors). Curated snapshot here; refreshable live via the CT.gov v2 API (see refresh_live.py).

Agent tools

list_indicationssearch_trialsget_trialsponsor_pipeline

Expert grading rubric

Dimension5 (excellent)1 (poor)
Landscape scoping & retrievalPulls the right trials for the indication/target, uses the search tools systematically, and captures the relevant competitors rather than a partial slice.Misses major programs, queries haphazardly, or reasons from memory instead of the tool data.
Design-delta analysisCompares trials on the dimensions that matter — phase, line of therapy, control arm, endpoint, enrollment — and surfaces meaningful differences.Lists trials without comparing them, or compares on irrelevant attributes.
Competitive read & leadersCorrectly identifies who leads (most advanced / largest programs) and characterizes the threat level per competitor.Mis-ranks maturity (e.g. calls a Phase 1 the leader), or ignores the most advanced asset.
Whitespace / differentiationIdentifies credible open positions (modality x line combos, endpoints, populations) that follow from the data and would actually differentiate a new entrant.Invents whitespace contradicted by the trials shown, or gives generic advice not grounded in gaps.
Evidence faithfulnessEvery claim (NCTs, sponsors, phases, enrollment) traces to tool outputs; no fabricated trials.Hallucinates trials, sponsors, or numbers, or contradicts the returned data.

Example queries

Trajectories

model panel (compare side by side)

ModelProviderTierJudge 1–5Verdict
Claude Opus 4.8anthropicfrontier4.0acceptable
GPT (frontier)openaifrontier3.2flawed
Claude Haiku 4.5anthropicsmall3.0flawed
GPT-4o miniopenaismall1.0unusable

batch 20260606_020437

QueryModelTool callsTimeStatus
Analyze the competitive trial landscape for DLL3-targeted therapies in extensive-stage sclaude-opus-4-8732.6sok
Survey the Claudin18.2 (CLDN18.2) competitive landscape in gastric/GEJ cancer. Compare mclaude-opus-4-8634.6sok
Map the HER2-directed competitive landscape in breast cancer across ADCs, bispecifics, aclaude-opus-4-8730.9sok
We are a biotech with a novel DLL3 ADC. Given the current ES-SCLC trial landscape, what claude-opus-4-8836.9sok