Introduction
BenchSci is a Toronto-based biotechnology software company founded in 2015 that applies artificial intelligence to preclinical research and drug discovery. Liran Belenzon serves as co-founder and chief executive officer per BenchSci’s company page — he joined after cold-messaging the founding team while working at the Creative Destruction Lab accelerator. BenchSci has raised over $200 million from investors including Generation Investment Management, iNovia Capital, TCV, and Gradient Ventures.
March 2026 marked the launch of EMET — Evidence Mapping and Exploratory Tool — which BenchSci describes as the agentic research environment it has been building toward since a 2015 founding moment when co-founder Tom’s failed experiment crystallized the conviction that discovery bottlenecks are infrastructural, not scientific.
What the company does
BenchSci builds AI that decodes biomedical literature and experimental evidence through a proprietary knowledge graph and machine learning models. ASCEND acts as an AI assistant for preclinical organizations — increasing productivity across therapeutic areas by surfacing biological insights, strengthening hypotheses, and supporting evidence-driven experimental design. Merck renewed a two-year ASCEND contract in December 2025, with Belenzon citing Merck teams using the platform to unravel disease biology.
EMET extends ASCEND into an agentic research environment — orchestrating AI models, software tools, and scientific workflows so researchers explore complex biological questions in one environment rather than fragmented point tools. BenchSci frames EMET as the product it always intended to build after ten years of knowledge graph and model development.
Who it serves
BenchSci serves top-20 pharmaceutical companies, over 4,500 research centers worldwide, and strategic partners including Thermo Fisher Scientific — a partnership BenchSci announced to deliver AI-powered research tools integrated with lab equipment and consumables ecosystems. Merck’s renewed ASCEND agreement validates enterprise renewal revenue in preclinical disease biology, a buying center distinct from clinical-stage AI vendors.
Research scientists evaluate BenchSci on evidence quality behind recommendations, integration with existing ELN and procurement workflows, and time saved designing experiments that avoid redundant or contradictory literature paths.
Company background
Four co-founders started BenchSci in 2015 after recognizing that failed experiments often stem from inaccessible or misinterpreted prior evidence rather than bench skill deficits. Belenzon brought MBA training and Creative Destruction Lab network access; the company grew through Deloitte Technology Fast 50 recognition, Great Place to Work certification, and Canada’s Top Growing Companies list — growth Belenzon tied to tackling drug discovery’s most pressing infrastructure challenge in September 2025 company news.
Product and AI capabilities
BenchSci’s knowledge graph encodes biomedical relationships extracted from literature, patents, and experimental databases — powering ML models that answer researcher questions with traceable evidence rather than opaque LLM outputs alone. EMET adds agentic orchestration: AI agents that plan exploratory research paths, map evidence landscapes, and support iterative hypothesis refinement within a governed preclinical environment.
Agentic AI in drug discovery raises validation requirements BenchSci addresses through pharma-grade deployments — Merck renewal and Thermo Fisher partnership signal enterprise confidence in ASCEND evidence quality before EMET agent expansion.
Key developments
2015: BenchSci founded in Toronto; Belenzon joined as co-founder and CEO. December 9, 2025: Merck ASCEND agreement renewed two years. 2026: EMET Agentic Research Environment launched (benchsci.com news, official source). Platform used across top-20 pharma and 4,500+ research centers per company disclosures.
Why it matters
BenchSci matters because 2026 pharma R&D budgets increasingly fund AI infrastructure that compresses preclinical decision cycles rather than adding another literature search tab. EMET’s agentic framing tests whether discovery teams adopt orchestrated research environments over disconnected Copilot plugins — a strategic question for R&D leaders comparing BenchSci to internal LLM builds. BenchSci complements wet-lab AI biotech like LabGenius by owning evidence synthesis upstream of antibody engineering.
Sector context
BenchSci belongs in Brel’s AI in healthcare coverage under biotechnology and AI drug discovery infrastructure, distinct from clinical AI companies like Signal 1 but sharing Toronto’s AI research hub geography with healthcare analytics vendors.
Sources and references
BenchSci EMET launch news article and company page.
- BenchSci — EMET launch (benchsci.com, 2026)
- BenchSci — Company (benchsci.com)
Reviewed July 10, 2026. Liran Belenzon is CEO and co-founder.