Your eDNA, answered.
PHYLO is the AI interpretation layer for environmental DNA: cited, human-approved answers from any lab's data.
eDNA can detect every species in a river. Then it hands you a spreadsheet.
The land manager asks
“Is this place getting better or worse, and what do we do?”
Gets: 4,000 rows of Latin names.
The scientist asks
“How do we know it's really there?”
Gets: a cell that says “detected.”
The program director asks
“Can I defend this decision in council, or in court?”
Gets: a week of R and ArcGIS, and no reference to compare against.
Detection is solved. Interpretation is wide open.
Make every eDNA result something a land manager can act on and a scientist can defend, whichever lab produced it.
Code computes. Agents explain. A person signs.
The numbers are computed, not generated. No fact originates in a model — every sentence in a PHYLO report cites a computed result.
-
Level 1 · code
Code computes
(deterministic)Deterministic bioinformatics and statistics, not a model.
Before any model sees your data, code resolves every name, checks every sample, flags every protected species, and tests every comparison.
Automated checks- Every name resolved against WoRMS and GBIF
- Every sample checked for depth and stray human or livestock DNA
- Every protected species flagged from IUCN, CITES and ESA lists
- Every comparison tested with repeatable statistics
-
Level 2 · agents
Agents explain
They turn the computed evidence into language and charts, and stress-test it — they never add a fact of their own.
Every arrow is a citation. Critics challenge it until it holds.
-
Then · a person
A person signs
Sees only the candidate that comes out the other side, and approves it before release.
Nothing ships on a model’s say-so.
Final human approval
Watch PHYLO read a real survey.
A live Salish Sea report, end to end: the science, then cited and confidence-rated answers. About half a minute.
More demos are on the way. Tell us what you'd like to see.
Data serves the community or organization it comes from, under its authority.
Sovereignty by design
Built for any organization whose data matters, including sovereign nations, where data sovereignty is a design requirement. Guided by the CARE principles.
Segregated by default
Sensitive data stays segregated from everything else, and PHYLO can be licensed into your own cloud.
Controlled by the owner
The data owner decides who sees the data and what it is used for.
Never used for training
No models or model providers train on or retain your data.
A founder who has built both sides of this.
Fifteen years building distributed platforms and AI-powered consumer products at global scale, with a biophysics background in structural and molecular biology from UCSD. Designed, built and shipped PHYLO to its first users.
We work with the research community, not against it. Our sister nonprofit, the Molecular Intelligence Institute, publishes the field's open standards, models and interpretation. Open substrate, proprietary platform.
Your first study is on us.
Bring your monitoring question. We build with you, and the data stays yours.
- 01
A call, so we understand what you're sampling and who needs the answer. If the study isn't designed yet, we help with that too.
- 02
On that call we settle how your data is handled, in writing, before any of it moves.
- 03
You sample, your lab sequences, we interpret. You get the answer in your own language.