Ground
The knowledge network
MicroMap by default, other knowledge graphs alongside it, and your own Company Graph. Every association links to its sources.
Learn more →For R&D teams in biotech, pharma, CROs and research cores
General models invent biology. We build the layer they check against.
Graphomics builds the verified data layer for biological AI. Operon gives your team a private research platform grounded in curated knowledge graphs. MapForge builds those graphs, including your own. Every result carries a record of the data, method and version that produced it.
or email info@graphomics.com
OperonMapForgeMicroMap + your Company Graph
Where AI for biology falls short
What AI research tools offer, and where they break down.
On offer
- Gene–disease links, proposed on demand
- Mechanistic hypotheses built from omics data
- Investigation cycles run end to end
- Publication-ready reports drafted in minutes
In practice
- Species, diseases, and pathways get hallucinated
- No source of truth linking gene → disease → target
- Bioinformatics tools that don't connect to each other
- No data isolation between one lab and the next
Why grounding matters
AI got good at code because code can be checked.
An agent writing software has somewhere to stand: it can run the code, read the trace, watch the test fail, and correct itself. Biology gives an agent none of that. There is no executable, inspectable record to check against, so it guesses.
01
Executable
An answer becomes an operation that can be run and re-run.
02
Inspectable
Every intermediate step is exposed as structured, queryable evidence you and the agent can audit.
03
Stateful
Progress persists across steps, so what was decided is still there at the next one.
How it works
How a question moves through Operon
Ask a question grounded in curated biology, run the analysis that answers it, then write the result back with its provenance, so the next question starts from it.
Execute
Operon
A research agent that cites what it used, analysis steps and validated Nextflow pipelines, approved before costly runs.
Learn more →Contribute
Your Company Graph
Validated results are written into your own private graph with their provenance and a trust label, so the next question builds on them.
Learn more →↻ MapForge builds and extends every graph in the loop. Nothing is written until a reviewer approves it.
MCP
Built on open standards. Your own agents (Claude Code, Cursor, or something you built) query the knowledge graphs over MCP or a REST API, alongside Operon's own agent.
At the centre of every engagement
Your Company Graph.
Every organization gets its own knowledge graph inside its Operon deployment: your data, assays, experiments and validated findings, linked to the curated biology in MicroMap. It is private to you, and it grows with every result.
Built from what you have
MapForge turns datasets, instrument outputs and internal databases into reviewed graph facts.
Linked to the reference
Shared identifiers mean one question draws on your results and the published evidence together.
Compounds with use
Validated results are written back with provenance, so the next question starts further ahead.
Yours alone
Visible only to your organization. Never pooled with other clients.
MicroMap is connected by default. Reference and domain graphs are designed to federate alongside.
Products
Operon and MapForge
Operon is where your team works. MapForge builds the graphs it works on.
Research platform
Operon
Your team's private research workspace, deployed for your organization alone.
- A research agent that grounds answers in the knowledge graphs it can reach, and cites what it used
- Analysis steps and validated Nextflow pipelines, approved before costly runs
- Campaigns: plan and run batches of experiments
- Every result recorded with its inputs, method and version
- Partner methods run through a registry, with the run recorded as provenance
Graph engine
MapForge
Builds knowledge graphs, including your Company Graph, from your data and public sources.
- Reads VCF, GFF/GTF, BIOM, AnnData, mzTab, spreadsheets and tables
- Matches your entities to standard identifiers and existing knowledge
- Scores confidence on every proposed link
- Nothing is written until a reviewer approves it
- Licensable on its own, for teams building graphs in their own domain
The default knowledge graph
MicroMap: curated microbiome biology, built with MapForge.
Every Operon deployment is connected to it. Every association is linked to its sources, with a declared confidence calibration.
63,136
source-linked microbe–disease associations, spanning 2,981 microbes and 239 diseases
10,000+
linked papers
2,800
metabolites
2,028
proteins
1.15M+
taxa (NCBI taxonomy)
Counts as of October 2026.
How we work with companies
Three ways to work with us
Pick the relationship that fits how your team works today. Each one runs on the same graphs and the same record.
Operon for your organization
Your own Operon deployment: your Company Graph, a grounded research agent and validated pipelines, for your whole team.
About Operon →A graph for your domain
MapForge builds and maintains a knowledge graph from your data and the public sources for your field. Or license the engine and run it yourself.
About MapForge →Knowledge for your agents
Your engineers query MicroMap from their own tools, Claude Code, Cursor or their own agents, over MCP or a REST API.
About the graphs →How it runs
01 · Scope
The questions that matter to your team, and the data you already hold.
02 · Deploy
Hosted by us or in your own cloud, connected to your data.
03 · Build
Your Company Graph, built with MapForge from your data and the public sources for your domain.
04 · Run
Answers, analyses and pipelines, with the evidence and lineage behind every result.
YOURS
Your data, your Company Graph and your results stay yours. They live in your deployment and are never pooled with other clients.
Why this is possible now
Three things changed.
01 · Capability
Reasoning got good, and cheap. Frontier models can plan an experiment and read an analysis.
02 · Standards
MCP made pipelines and data sources something any agent can call.
03 · Stakes
A hallucinated pathway can cost a lab months at the bench.
Frontier models can plan an analysis. They can't tell you which microbe–disease link has several studies behind it and which has none. MicroMap can.
A fair question
Why not point a general-purpose agent at public databases?
You can, and for a first look it helps. What it lacks is somewhere to check its answers. Public sources disagree, use different names for the same microbe, and rarely say how strong a link is. An agent reading them cold has to guess which to believe.
MicroMap has done that work already: 63,136 microbe–disease associations, with names reconciled to standard identifiers and each association linked to the studies behind it. Your Company Graph adds what only your team knows. Operon keeps the record of what was asked, what was used and what was decided.
Bring a question from your own work. We'll walk you through how Operon would answer it.
or email info@graphomics.com