Confidential — Physician Advisory Brief

AI-Enabled Clinical Trials Services

Portfolio recommendations and competitive landscape for an agentic-AI clinical research offering. Built for a clinical trial MD partner exploring how our 240-agent platform can deliver sponsor- and CRO-grade services.

Prepared April 15, 2026 Author Donatello / Starship Labs Platform 225 agents · 15 Clinical Pharmacology · 8 IBC Reasoning Patent Pending — US Prov. 63/986,270

1 · Executive Summary

The clinical trials market ($80B+ globally, growing 6% CAGR) is being reshaped by AI-enabled services. Traditional CROs (IQVIA, ICON, Parexel, Syneos) are bolting AI onto legacy operations; platform pure-plays (Medidata, Veeva, Saama) sell software; but no incumbent offers an end-to-end agentic service layer that spans protocol design, enrollment, monitoring, regulatory, and readout. That is the gap our platform can fill.

Our current assets: (1) Clinical trials analysis agent team, (2) Phase II→III transition prediction engine, (3) Enrollment prediction engine. These three are entry points, not the product. The full offering described below rides on the same agent infrastructure and is what turns a point tool into a recurring-revenue services business.

2 · Recommended Service Portfolio

30 services across 7 domains (covering pre-clinical through post-launch). Green chips = already built. Blue chips = recommended additions. Pricing reflects 2026 market benchmarks for sponsor-facing engagements.

DOMAIN 0 — PRE-CLINICAL / DISCOVERY (BEFORE PHASE 1)

0.1

Target Validation & Druggability Scoring NEW

Genetic evidence (OpenTargets, GWAS), pathway mapping, tractability scoring, competitive target landscape. Go/no-go before medicinal chemistry spend.

$80K–$250Kper target

0.2

Lead Optimization & DMPK Prediction NEW

ADMET, solubility, hERG, CYP inhibition, blood-brain barrier prediction. Narrows chemistry iterations pre-candidate-selection.

$100K–$300Kper program

0.3

In Silico Toxicology Prediction NEW

Hepatotox, cardiotox, genotox, off-target binding. Reduces animal study failures and IND-enabling tox surprises.

$80K–$200Kper candidate

0.4

FIH Dose Selection (NOAEL / MABEL / PAD) HAVE — CP-01

Maximum Recommended Starting Dose via triangulated NOAEL allometry, MABEL receptor-occupancy, and pharmacologically active dose. ICH M3(R2) / S9 aligned.

$60K–$180Kper candidate

0.5

IND-Enabling Study Design NEW

GLP tox study design (species selection, duration, endpoints), safety pharmacology battery, genetox package. Maps to Module 4 requirements.

$150K–$400Kper program

0.6

Translational Biomarker Strategy NEW

PD biomarker selection, target engagement assays, imaging strategy — anchors FIH design to later-phase success.

$120K–$350Kper program

0.7

CMC / Manufacturing Readiness NEW

API synthesis risk, formulation feasibility, stability prediction, scale-up red flags. Protects against Module 3 deficiencies.

$100K–$300Kper program

0.8

★ Candidate-to-Clinic Success Prediction NEW HIGH VALUE

Probability that a preclinical candidate survives IND and reaches Phase 1. Integrates 0.1–0.7 outputs into a single go/no-go score. First-to-industry.

$150K–$400Kper candidate

DOMAIN A — TRIAL DESIGN & STRATEGY

Protocol Optimization Agent NEW

Tune inclusion/exclusion against eligible population, screen-fail rate, endpoint feasibility, and competing trials. Recommends protocol amendments pre-submission.

$150K–$400Kper protocol

Adaptive Trial Design (Bayesian / Seamless) NEW

Sample-size re-estimation, arm dropping, seamless Phase I/II and II/III designs. Simulation-based operating characteristics.

$200K–$600Kper design

Synthetic Control Arm / External Control NEW HIGH VALUE

RWD/historical-trial matching to reduce placebo enrollment. FDA Project Facilitate-aligned. Propensity scoring, G-computation, tipping-point sensitivity.

$500K–$1.5Mper trial

Site Selection & Feasibility Scoring NEW

Investigator performance history, patient density, competing trial load, startup timelines. Replaces spreadsheet-based vendor selection.

$75K–$200Kper program

Dose-Finding / Project Optimus (Oncology) NEW

Randomized multi-dose Phase 2 designs per FDA Project Optimus 2023 guidance. BOIN, mTPI-2, MRSD via CP-01 agent.

$200K–$500Kper program

DOMAIN B — PATIENT & ENROLLMENT

Enrollment Prediction Engine HAVE

Already built. Timelines, site-level yield curves, shortfall alerts.

$100K–$300Kper trial

Patient Matching / Digital Pre-Screening NEW

EHR & claims-driven eligibility matching at point of care. NLP over unstructured notes for I/E criteria.

$250K–$800K+ per-match fee

Dropout & Retention Prediction NEW

Patient-level churn scoring, early intervention triggers (missed visits, AE clustering, ePRO compliance decay).

$100K–$250Kper trial

Diversity & Representativeness Analytics NEW REGULATORY

FDA Diversity Action Plan (DAP) compliance — submission-ready. Monitors enrollment demographics vs disease epidemiology in real time.

$150K–$400Kper program

DOMAIN C — OPERATIONS & MONITORING

Risk-Based Quality Management (RBQM) / Central Monitoring NEW HIGH VALUE

ICH E6(R3)-aligned. Replaces on-site monitoring for ~60% of visits. Central statistical monitoring, KRI/QTL dashboards.

$500K–$2Mper trial (multi-year)

Protocol Deviation Prediction & Early Warning NEW

Site-level deviation forecasting 30–60 days out. Targets training and corrective action.

$80K–$200Kper trial

Data Quality / Query-Burden Forecasting NEW

Predicts query storm weeks before DBL. Reduces data manager burn.

$60K–$150Kper trial

Investigator Performance Scoring NEW

Longitudinal performance tracking feeds back into Site Selection (A4). Creates compounding data moat.

Bundled w/ A4 + C1

DOMAIN D — REGULATORY & SAFETY

IND / CTA Package Drafting + Gap Analysis NEW HIGH VALUE

Module 2 & 5 authoring, briefing books, Type B/C meeting prep. Leverages our 15 CP agents.

$300K–$1.2Mper submission

Safety Signal Detection (AE / SAE Clustering) NEW

MedDRA-coded AE pattern mining, disproportionality analysis, SMQ-level signals. Near-real-time dashboards for medical monitor.

$200K–$600Kper program/yr

DSMB / DMC Analytics Support NEW

Open- and closed-session report packs, unblinded interim analyses, stopping-rule monitoring.

$75K–$250Kper meeting

DOMAIN E — EVIDENCE & READOUT

Phase 1→2 Transition Prediction NEW HIGH VALUE

Probability of Phase 2 entry + probability of positive Phase 2 readout. Inputs: MTD/RP2D achievement, PK exposure vs preclinical target, early PD/biomarker response, DLT profile, competitor Phase 1 benchmarks. Mirrors Phase II→III engine architecture.

$120K–$400Kper asset

Phase II→III Success Prediction HAVE

Already built. Probability of Phase III success, pivotal readout handicapping, rNPV.

$150K–$500Kper asset

Competitive Intelligence / Landscape Agent NEW

Continuous scan of ClinicalTrials.gov, conferences, SEC filings. Alerts on competitive read-outs, mechanism overlaps, enrollment pressure.

$60K–$180Kper asset/yr

Biomarker & Subgroup Finder NEW

Post-hoc subgroup analytics with multiplicity control, biomarker-responder enrichment for Phase III.

$150K–$400Kper study

SAP & Mock TLF Automation NEW

Draft Statistical Analysis Plans + mock tables/listings/figures. 70% cycle-time reduction vs traditional biostats.

$100K–$300Kper study

HEOR & Payer Evidence Dossier NEW

AMCP-format dossier, budget impact / cost-effectiveness models, ICER / NICE-ready analyses.

$250K–$800Kper launch

DOMAIN F — COMMERCIAL BRIDGE

Launch Readiness & rNPV Valuation NEW

Ties directly to the Alpha Engine. Risk-adjusted NPV, peak-sales ranges, probability-weighted pipeline valuation.

$100K–$250Kper asset

Label Optimization Simulator NEW

Model alternative label language vs commercial uptake and payer coverage.

$150K–$400Kper asset

Post-Marketing / Phase IV Surveillance NEW

RWE monitoring, PMR/PMC commitments, PBRER support.

$300K–$1Mper program/yr

3 · Competitive Analysis

Six competitor archetypes. No single vendor covers the full domain footprint we propose. AI-native point players (Unlearn, Owkin, Saama) have depth but narrow scope. Legacy CROs have scope but bolt-on AI without agentic reasoning.

Capability	IQVIA~86K employees $15.4B rev · NYSE: IQV	Medidata~3K employees $1.5B rev · Dassault-owned	Saama~500 employees Private · Series F	Unlearn.AI~100 employees Series C · $130M raised	Owkin~300 employees Series B · $300M raised	ConcertAI~500 employees Private · SymphonyAI-backed	Our PlatformStartup · Donatello Patent pending · Agentic
Protocol optimization (A1)	Partial	—	—	—	—	Partial	Full
Adaptive design (A2)	Yes	—	—	Yes	—	—	Full
Synthetic control arms (A3)	Partial	—	—	Yes — core	Yes	Partial	Full
Site selection (A4)	Yes	Partial	—	—	—	Yes	Full
Enrollment prediction (B1)	Yes	Yes	Partial	—	—	Yes	HAVE
Patient matching (B2)	Yes	—	—	—	Partial	Yes — core	Full
Diversity / DAP (B4)	Partial	Partial	—	—	—	Partial	Full
RBQM / Central monitoring (C1)	Yes	Yes	Yes	—	—	—	Full
IND / CTA drafting (D1)	Partial (consults)	—	—	—	—	—	Full (agentic)
Safety signal detection (D2)	Yes	—	Yes — core	—	—	Partial	Full
Candidate→Clinic / IND success (0.8)	—	—	—	—	Partial	—	Full
Phase 1→2 prediction (E0)	Partial	—	—	Partial	—	—	Full
Phase II→III prediction (E1)	Partial	—	—	Yes (TwinRCT)	Partial	—	HAVE
Competitive intel (E2)	Yes	—	—	—	—	Yes	Full
SAP / TLF automation (E4)	Partial	—	Partial	—	—	—	Full
HEOR dossier (E5)	Yes	—	—	—	—	Partial	Full
rNPV / launch valuation (F1)	Partial	—	—	—	—	—	Full (Alpha Engine)
Agentic reasoning layer	—	—	—	—	—	—	Unique
IBC audit trail (19-agent)	—	—	—	—	—	—	Unique

Competitor Profiles

IQVIA + NVIDIA LARGE-CAP

Size: ~86,000 employees · ~$15.4B revenue (2024) · NYSE: IQV · Market cap ~$35B

Biggest threat. Scale + data + AI compute. Offers decision-support across the lifecycle, but services are consulting-heavy, slow, and non-agentic. Weak on protocol authoring and IND generation.

Medidata (Dassault) LARGE-CAP

Size: ~3,000 employees · ~$1.5B revenue · Owned by Dassault Systèmes (EPA: DSY, ~$45B mkt cap)

Software-first (Rave EDC, Acorn AI). Strong EDC + central monitoring. Does not offer agentic services; no IND authoring, no rNPV modeling.

Saama MID-SIZE

Size: ~500 employees · Private · Series F ($430M raised) · Carlyle-backed

AI-native biometrics + safety focus. Strong signal detection. Narrow scope — no protocol design, no commercial-bridge services.

Unlearn.AI STARTUP

Size: ~100 employees · Series C · $130M+ raised · Insight Partners-led

Twin RCT / digital-twin control arms in neuro & MSK. Deep but narrow. FDA-engaged. Competes directly on A3 and adjacent to E1.

Owkin MID-STAGE

Size: ~300 employees · Series B · $300M+ raised · Sanofi strategic partner

Federated learning on hospital data. Biomarker discovery (E3) and diagnostic AI. Does not offer operational services.

ConcertAI MID-SIZE

Size: ~500 employees · Private · SymphonyAI Group-backed · Valuation ~$1.9B

RWD (oncology-heavy) + trial matching. Acquired CancerLinQ data assets. Strong B2, weak on design & regulatory.

Parexel / ICON / Syneos LARGE-CAP

Size: Combined ~120K employees · ICON (NASDAQ:ICLR ~$24B mkt cap, $8B rev) · Parexel (EQT-owned, ~21K emp) · Syneos (~29K emp, PE-owned)

Legacy full-service CROs. Acquiring AI via partnerships. Slow cultural shift; AI layered on top of FTE-driven billing.

Tempus AI LARGE-CAP

Size: ~2,300 employees · NASDAQ: TEM · ~$700M revenue · Mkt cap ~$10B

Sequencing + RWD + TIME trial platform. Oncology-centric. Strong biomarker discovery; operational services limited.

Deep 6 AI / Antidote / TriNetX STARTUP→MID

Size: Deep 6 (~100 emp, Series B) · Antidote (~50 emp, acquired 2024) · TriNetX (~700 emp, Carlyle-backed ~$1.3B valuation)

Patient matching / cohort discovery point tools. Competes narrowly on B2. No design/regulatory/safety.

4 · Our 6 Differentiators

1. Agentic Service Layer

225 specialized agents (15 Clinical Pharmacology, 8 IBC Reasoning, etc.) collaborate per task. No incumbent ships this — they ship SaaS or FTE-hours.

2. IBC Reasoning Audit Trail

Every recommendation traced through Causal → Contradiction → Confidence → Mechanistic → Analogical → Abductive → Temporal → Meta. Defensible for regulatory scrutiny.

3. Full-Stack Coverage

Protocol → Enrollment → Monitoring → Regulatory → Readout → Launch — one vendor, one data spine. Competitors force stitched workflows.

4. Commercial-Bridge Integration

Alpha Engine rNPV + signal-attribution models plug directly into F1/F2. Uniquely positions us for venture- and PE-funded biotechs.

5. Speed + Cost

SAP drafting 70% faster, IND authoring 60% faster, site selection days-not-weeks. Pricing 20–40% below IQVIA/Medidata equivalent scope.

6. Patent-Pending IP

US Provisional 63/986,270 covers the agentic reasoning architecture. Moat grows with training-data accumulation.

5 · Three-Tier Pricing / Packaging

Essentials

$1.2M / program / yr

Enrollment prediction (B1)
Phase II→III prediction (E1)
Competitive intelligence (E2)
Monthly reporting
1 asset covered

Operator Suite MOST POPULAR

$2.8M / program / yr

All Essentials services
Protocol optimization (A1)
Site selection (A4)
RBQM central monitoring (C1)
Safety signal detection (D2)
SAP / TLF automation (E4)
Up to 3 assets

Enterprise / Portfolio

$5.5M+ / program / yr

All Operator services
Synthetic control arm (A3)
IND/CTA authoring (D1)
HEOR dossier (E5)
rNPV / launch readiness (F1)
Dedicated physician partner + IBC audit
Unlimited assets in portfolio

6 · Recommended Build Roadmap

Q2 2026

Anchor Services

Ship A1 Protocol Optimization, A4 Site Selection, C1 RBQM. Leverage existing enrollment + Phase II→III engines as co-sell.

Q3 2026

Regulatory Wedge

Ship D1 IND/CTA Authoring, D2 Safety Signal. Highest willingness-to-pay; fastest to differentiate vs IQVIA consulting.

Q4 2026

Synthetic Control + HEOR

Ship A3 (go head-to-head with Unlearn), E5 HEOR dossier. Target oncology and rare-disease biotechs.

Q1 2027

Commercial Bridge

Ship F1 rNPV (tie to Alpha Engine), F2 label simulator, E3 biomarker finder. Close the loop from design → launch.

7 · "First-to-Industry" Picks

Given the 225-agent infrastructure, three services are most likely to be genuinely novel — not faster versions of what exists, but capabilities that no incumbent can replicate without the agentic architecture.

★ Agentic IND Authoring (D1)

A 15-agent Clinical Pharmacology team drafting CTD Module 2 & 5 with IBC reasoning audit. IQVIA does this with FTEs; no one does it agentically. $300K–$1.2M per submission, 60% time reduction.

★ RBQM with Reasoning Trail (C1)

Central monitoring with Causal/Abductive agents explaining *why* a site is at risk, not just that it is. Defensible in sponsor audits and regulatory inspections.

★ Closed-Loop Design Optimizer (A1+A3+A4)

Single-pass: I/E criteria → eligible population → site selection → synthetic control arm feasibility. Competitors sell these as separate tools; we run them as one reasoning chain.

8 · Partnership Ask from the MD

What we need from a Clinical Trial MD partner:

Validate the clinical realism of A1 (Protocol Optimization) and D1 (IND Authoring) outputs against current sponsor expectations.
Introduce 3–5 sponsor or CRO decision-makers for discovery conversations.
Serve as medical monitor-of-record on demo engagements (builds credibility for D2 / C1 safety services).
Co-author a clinical case study (anonymized) demonstrating one service end-to-end — target JAMA Open or Clin Trials journal.
Advise on pricing calibration — what sponsors will actually pay vs what we benchmark against IQVIA.

Key risks to surface in MD discussion: (1) Regulatory acceptance of agentic IND drafts — FDA will want human authorship attestation. (2) Liability framework for safety signal detection (D2) — who signs off if an agent misses an AE cluster? (3) Data access — sponsors are protective of trial databases; need clear BAA/DUA templates. (4) Pricing credibility — services priced 20–40% below IQVIA may be perceived as "too cheap to be real" without MD co-signature.