Mantis Biotech - Databricks for Biomedical and Clinical Data
Mantis encodes biological and clinical meaning directly into reusable datasets
Hey Guys,
I’m Georgia, an extreme athlete turned software engineer and repeat biomedical founder, building Mantis Biotech.
TL;DR
80% of clinical trials are delayed by data inaccuracies like manual entry, inconsistent practices, and lack of validation systems. We’re the first company that can access all data in a clinical trial through a full-stack search. Each clinical trial loses on average $15M per trial because of data quality issues - our software prevents these from appearing.
Watch our Launch Video
The Problem
Earlier this year, an Alzheimer’s drug called Simufilam was discontinued in late-stage clinical trials due to scrutiny over data traceability. By the time the program was halted, hundreds of millions of dollars had been spent, years of potential time-to-market were lost, and 7.2 million Americans continued to suffer. This was not a standalone issue.
When trial and R&D data are fragmented across systems, every new question turns into a bespoke data project with real financial consequences. A single cross-system analysis can require 4–12 weeks of work from data engineers, biostatisticians, and clinical operations staff, often involving 3–6 people and costing $100,000–$500,000 in internal labor and vendor support before a result is produced. Because pipelines and mappings are built for one-off questions, much of that work is discarded afterward, forcing teams to repeatedly pay to reconstruct the same logic for future analyses. Over the life of a Phase II or III program, organizations may run dozens of these projects, quietly spending millions of dollars on redundant data wrangling alone.
Our Solution
We provide a domain-aware data platform for life sciences that encodes biological, clinical, and human-performance meaning directly into datasets. Instead of raw tables that must be reinterpreted for each use case, we deliver canonical, reusable, continuously maintained datasets that reflect how life sciences work in practice.
When questions arose in the Simufilam program about whether changes in a specific biomarker (e.g., filamin A–related signaling) actually correlated with cognitive scores at certain sites, timepoints, or assay runs, our platform would have let a scientist type a query like “show all patients with ≥X biomarker change, grouped by site and assay batch, alongside ADAS-Cog outcomes” and get an answer immediately, instead of coordinating data pulls from EDC, lab systems, and analysis code for weeks.
By integrating with existing life sciences systems and preserving the full lineage of source data, our platform enables teams to reuse trusted datasets across analytics, operations, AI, and compliance**.** We allow scientists to answer questions with a single query, rather than a lengthy data project, which shortens clinical trial lengths, amplifies early warning signals, and prevents late-stage dismissal.
What We offer:
- Integration across core life sciences systems (EDC, CTMS, labs, omics)
- Canonical, versioned datasets with stable definitions
- Traceability from datasets back to source systems
- Reuse across analytics, ML, and operational workflows
Demo
https://www.youtube.com/watch?v=MT3sqNnltz4\
**The Team
**I’m Georgia, a former elite ice climber who helped build the U.S. team into the best in the world. I studied CS and psychology at Harvey Mudd, computational math at Johns Hopkins, and Biomedical engineering at the University of Washington. Before founding Mantis, I built Louiza Labs, a physics engine that powers digital twins for autonomous robotic surgery and simulated FDA trials.
Our Ask
If you know any of the following
Senior data leaders in life sciences who are tired of rebuilding clinical and biomedical data pipelines every time the business asks a new question.”