# Mendel.ai

> Mendel.ai sells clinical AI software that turns fragmented patient records into searchable data for trial screening, cohort research and chart review. Its Hypercube platform combines language models with a clinical hypergraph; a randomized oncology study found that AI-assisted reviewers were more accurate than reviewers working alone, though no faster in the final analysis.

- **Founded:** 2017
- **Headquarters:** San Jose, United States
- **Founders:** Karim Galil (Co-founder and CEO), Wael Salloum (Co-founder and Chief Science Officer; previously CTO)
- **Products:** Hypercube, Hypercube Cohorts, Hypercube Charts, Hypercube Analyst, Hypercube Redact
- **Notable:** A 2025 randomized trial final analysis of 356 charts reported 76.1% chart-level accuracy for human-plus-AI review versus 71.5% for human-alone and 59.9% for AI-alone; review time was similar., A 2020 retrospective study at Comprehensive Blood and Cancer Center found 24% to 50% more potentially eligible patients in two enrolling trials, while neither method found candidates in a third nonenrolling trial., Hypercube became available on Google Cloud Marketplace and natively on Snowflake in 2024.

## Products & services

- **Hypercube** — Clinical AI platform combining large language models with a proprietary clinical hypergraph for patient-record search and reasoning.
- **Hypercube Cohorts** — Defines and refines patient groups across structured and unstructured records.
- **Hypercube Charts** — Natural-language chart review and medical-record exploration.
- **Hypercube Analyst** — Low-code querying of clinical data for research and analytics.
- **Hypercube Redact** — De-identifies and tokenizes patient records; also outputs redacted PDFs, text and coordinates.
- **Hypercube Build-Your-Own** — Custom clinical AI copilots for organization-specific workflows.
- **Resolve** — Consolidates clinical information into a longitudinal patient journey; announced alongside the 2022 Series B.

## Achievements

- A 2025 randomized trial final analysis of 356 charts reported 76.1% chart-level accuracy for human-plus-AI review versus 71.5% for human-alone and 59.9% for AI-alone; review time was similar.
- A 2020 retrospective study at Comprehensive Blood and Cancer Center found 24% to 50% more potentially eligible patients in two enrolling trials, while neither method found candidates in a third nonenrolling trial.
- Hypercube became available on Google Cloud Marketplace and natively on Snowflake in 2024.

## Latest updates

- **2026-02** — ASCO AI in Oncology covered the published randomized study: AI-assisted prescreening improved accuracy while average chart-review time was similar.
- **2025-04** — AWS published a founder profile describing Mendel’s cloud migration and clinical AI copilot approach.
- **2025-02** — Meta described Mendel’s use of fine-tuned Llama models within Hypercube.
- **2024-09** — Hypercube became generally available as a Snowflake Native App.

## Links

- Website: https://www.mendel.ai/
- LinkedIn: https://www.linkedin.com/company/mendel-ai
- Twitter/X: https://twitter.com/mendel_ai_
- Facebook: https://www.facebook.com/mymendel/

---

Profile page: https://yespress.io/mendel-ai
Published by YesPress — https://yespress.io
Last updated: 2026-09-28