← All Dossiers
Keiji AI – Seed
In Conversation Round: Seed  ·  Sector: AI · Clinical Trials  ·  Vehicle: Fund II
Dossier generated 2026-07-09 by /deal-dossier  ·  Deal record: rec0PDYz3KqHVodbD  ·  Source: Airtable appjxAR3LPe3fkHOp

One-Liner & Thesis

TrialMind: an agentic AI platform that runs the full clinical-trial lifecycle — literature synthesis, protocol design, eligibility criteria and cohort extraction, data-science co-pilot for biostatisticians, and regulatory drafting Web. Spun out of the University of Illinois ML-for-Healthcare lab of Jimeng Sun, ex-Global Head of AI Research at IQVIA Web.

Keiji is a rare revenue-funded generative-AI company in clinical trials — on pace for $5–7M ARR this year with paying pharma and biotech customers (Regeneron, Structure Therapeutics, Loomis, Garden Health, AbbVie) and no institutional capital raised — sitting on top of a proprietary trial-data pipeline (850K protocols from 14 countries linked to 1.3M publications) built inside a top academic AI-for-health lab. The wedge is bespoke clinical-data access plus a domain-tuned agent stack rather than a generic LLM wrapper: contracts run $200K–$1M, expansion is happening inside customers (Garden went from 1 to 3–4 SOWs), and the platform ships an MCP server so it can co-exist with corporate Claude Code deployments Internal. The drag is founder posture — Jimeng, the CS-professor CEO, is not urgently seeking capital and has been slow to engage on fundraising process; StoryHouse’s posture is to keep the option warm at $800K–$1M in a $2–3M-sized syndicate rather than push for a full round.

Investment Score & Recommendation

63/ 100
HOLD

Biggest driver: real, growing enterprise ARR ($2M last year to $5–7M target this year) inside a category (gen-AI for clinical trials) growing ~25% CAGR to $8B+ by 2031 — anchored by a top-tier academic AI-for-health team. Biggest drag: no active round, no defined lead, and a CEO who by his own team’s admission is not engaged on fundraising — the deal is a warm option, not a live financing.

Momentum: Steady Red Flags: 4 / 9 Confidence: MEDIUM
Market & TAM9/10
25% weight
Team & Founder7/10
25% weight
Product & Traction7/10
20% weight
Deal Terms & Return3/10
20% weight
VC Syndicate3/10
10% weight

Deal Box

Funding Round
Seed (contemplated, not active)
Round Size
~$3.5M current (Airtable); founders have discussed up to $8M
Structure
Prior conversations referenced small SAFE; nothing termed
Lead Investor
TBD — not yet identified; SH could co-lead a $2–3M sub-round
SH Check
Contemplating $800K–$1M (co-lead scenario)
Fund
StoryHouse Fund II
Deal Terms
Not discussed in detail. UIUC has a license agreement with Keiji covering the underlying research.
Timing
Team retreat mid-October 2025 to decide; potential raise Nov-Jan window; as of Jun 2026 still not actively fundraising

Company Snapshot

Sector
AI · Clinical Trials · Data · Pharma / Life Science
Founded
2023 (research origins in Jimeng Sun’s UIUC lab; formally spun out 2025)
Location
Distributed (Jimeng in Las Vegas / UIUC; team clustered around UIUC)
Headcount
~4 full-time + Jimeng part-time + contractors (Sep 2025); grown since
Est LTM Revenue
$2.0M ARR (Sep 2025); targeting $5–7M this year
Total Raised
None — revenue-funded to date
Website
keiji.ai
License
University of Illinois license/royalty agreement on core IP

Market Size (TAM)

$2.68B
AI in Clinical Trials 2026 Web
$8.24B by 2031 · 25.2% CAGR (Mordor)
$685M
GenAI in Clinical Trials 2026 Web
$4.28B by 2033 · 29.9% CAGR (Coherent)
$80–92B
Global CRO Market 2025 Web
$100B+ by 2027 — the buyer pool
52.1%
Pharma/Biotech AI Spend Share Web
CRO segment growing 26% CAGR (Mordor)
Early
Agentic AI Adoption Curve Internal
Most enterprise pilots still <24 months old

Clinical-trial gen-AI is one of the fastest-growing enterprise AI verticals: 25–30% CAGR across three independent research houses, anchored by the R&D-cost pressure inside pharma (average trial costs $50M+ per Phase 3) and the multi-year push toward decentralized and adaptive trials. Keiji is entering while enterprise buyers are still calibrating agentic-AI budgets, and the moat that matters — access to structured trial protocols and outcome data — is being built in a category where scraped-text LLMs have limited leverage.

Competition

PlayerPositioningFunding / StageEdge vs. them
Keiji AI (TrialMind)Agentic platform across the full trial lifecycle; proprietary 850K-protocol + 1.3M-publication data pipeline; MCP-first architecture InternalBootstrapped · $2M ARR → $5–7M target
Tempus AIPublic precision-medicine / trial-matching platform; acquired Deep 6 AI to add 30M patients across 750+ sites WebPublic (NASDAQ: TEM)Tempus is oncology-first with clinico-genomic data; Keiji targets the operational trial layer regardless of TA
Unlearn.aiDigital twins for control-arm augmentation in trials WebSeries C, $15M+ ARR rangeDifferent wedge: Unlearn is a statistical trial-design product; Keiji spans the full lifecycle including exec/data-science co-pilot
OwkinMultimodal patient-data federation + AI biotech; Paris/NY, ~$300M raised WebSeries B+Owkin runs its own biotech pipeline and hospital federation; Keiji is a pure enterprise SaaS/agent play for sponsors and CROs
Anthropic Claude / OpenAI (in-house)Enterprise LLMs used directly by pharma dev teams (“Claude Code”) InternalN/A (platform)Real threat — but Keiji ships an MCP server so it composes with, rather than competes head-on against, corporate Claude deployments
Trials.aiNamed direct competitor per Tracxn; earlier-stage WebSeedNarrower feature scope; less academic-lab data-moat depth

Moat: a proprietary trial-data pipeline (850K protocols, 14 countries, linked to 1.3M publications, structured to ICD/MeSH) built inside the UIUC lab plus a domain-specific agent stack. The MCP-server strategy positions Keiji as a data-and-workflow layer that composes with enterprise LLMs rather than competes with them — a defensible posture against Claude/OpenAI DIY.

Traction

$5–7M
ARR Target 2026 Internal
$2M ARR in 2025 · ~3–5x YoY
8+
Enterprise Customers Internal
Regeneron, Structure Therapeutics, Loomis, Garden, AbbVie
$200K–$1M
Contract Size Internal
Upfront integration + license/usage SaaS
100–150
DSP Active Users Internal
Data science platform — main commercial product
50
Garden Seats @ $7,500 Internal
Renewed and expanded from initial pilot
Self-Funded
Capital Base Internal
Team paid from revenue; no institutional capital

Exit Potential

Strategic M&A
Likely Path
4–7 yr
Time to Liquidity
5–12x
Return Scenario Internal

The category is consolidating around strategic buyers: Thermo Fisher closed the $8.875B Clario acquisition in March 2026, IQVIA acquired AI analytics firm Whiz, and Veeva acquired the Ostro patient-engagement platform for $100M in March 2026 Web. Tempus AI absorbed clinical-trial-matching Deep 6 AI to bolt patients-and-sites onto its precision-medicine offering. The plausible acquirors for Keiji are IQVIA (natural Jimeng-alumni buyer), Veeva, Medidata/Dassault, Tempus, Thermo Fisher, and Icon; a strategic pharma buy (Regeneron, AbbVie) is possible but less common in this workflow-tooling category. A revenue-multiple exit at 8–12x forward ARR is the base case if ARR compounds through $25–50M, and the return math works cleanly from a sub-$40M entry.

Founders

Jimeng Sun — Co-Founder / CEO
Health Innovation Professor of CS, UIUC · ex-Global Head of AI Research, IQVIA · ex-IBM Research Web
Top-tier AI-for-healthcare researcher — 300+ publications, h-index 81, one of the Top 100 AI Leaders in Drug Discovery per industry rankings. PhD Carnegie Mellon; BS/MPhil HKUST; ran AI research at IQVIA (the buyer for products like Keiji’s), previously at IBM T.J. Watson. Runs the Sunlab at UIUC where the core Keiji research was developed. Not a professional operator — the founder posture is that of an academic still deciding how commercial he wants Keiji to become.
Brandon Theodorou — Co-Founder / Head of Product Engineering & Customer Success
HMC ’21 BS Math + CS · UIUC PhD CS (ML for Healthcare) · ex-NIH, ex-IQVIA, ex-ServiceNow, ex-Microsoft Internal
HMC alum (Claremont connection to StoryHouse) with a UIUC PhD from Sun’s Sunlab. Runs product engineering and customer success day-to-day; recently expanded team by hiring a Head of Business Development and a developer with domain expertise. Deferred fundraising decisions to Jimeng in the June 2026 email exchange.

Open Questions & Risks

Next Steps

Latest Meeting Notes

2026-06-22Email UpdatePost-call materials delivered; Jimeng to own next fundraise signal

Brandon apologized for a delayed follow-up and sent the deck plus demo access via keiji.ai. Confirmed Keiji is not attending BIO San Diego and explicitly deferred all fundraising discussion to Jimeng — a soft signal that Brandon is not the decision-maker on capital.

Source: Meeting Notes recDDDQobgnmNk4WO
2026-05-21ZoomModel shift to renewable licensing; StoryHouse could co-lead $2–3M

Keiji is transitioning from co-development pilots to renewable annual licensing. Data Science Platform (main product) has 100–150 active users, Garden renewed 50 seats at $7,500 each, Structure Therapeutics and Loomis at 10 seats each on-prem. Serious fundraising exploration but no urgent timeline. StoryHouse indicated a possible $800K–$1M check, willing to co-lead a $2–3M round; would not lead a $5M+ raise alone.

Source: Meeting Notes recOtHJFKA2Yi5MBU
2026-05-12Reconnect CallInteresting business, disengaged CEO

Reconnect after several months of limited engagement. Confirmed $5–7M ARR target unchanged; three products: data-science platform (main), literature review, eligibility criteria generation. Verdict: interesting business, but CEO is not engaged on capital and not worth pushing hard.

Source: Meeting Notes recr3pFjSsqFRPmIg

Deal Timeline