Frekil

Frekil is a real-world evidence platform that turns clinical data into study-ready outputs through an eight-stage AI agent pipeline. It addresses the data preparation and analysis work that sits between a clinical question and usable evidence, covering literature review, question clarification, cohort construction, causal DAG building, statistical analysis plan generation, data extraction, sandboxed execution, and report writing. Users can initiate studies in natural language and review or adjust each stage before execution.

The platform connects to EHR, claims, registry, and pharmacy data, including through Databricks, Snowflake, AWS, GCP, and Azure, and maps schemas to OMOP CDM and medical ontologies such as ICD-10, SNOMED, RxNorm, and ATC. It generates statistical code in R, Python, and SAS, executes that code in a sandboxed environment separate from patient data, and produces tables, figures, listings, and audit-tracked reports. Agents are described as self-improving, learning from each study run.

Frekil is aimed at RWE teams, biostatisticians, epidemiologists, clinical researchers, HEOR and market access functions, and CROs, supporting use cases such as comparative effectiveness, post-market safety, label expansion, trial feasibility, and external control arms. Pricing is offered on a flexible basis, including per-study and platform options, with no free plan or trial stated on the pages.