Tools

AI tools and scientific systems

Alongside its experimental biology work, the lab builds internal AI-enabled systems for literature synthesis, research memory, and scientific operations. This page is a public overview of that work; interactive prototypes remain private while they are still evolving.

These systems are designed to reduce administrative drag, preserve project continuity, improve evidence traceability, and make scientific work more inspectable and reusable.

What this is

A compact view of the systems layer behind the lab: workflow-shaped systems built around real research questions, source traceability, and human review.

In practice, this includes tools that compare methods papers across bounded corpora, link claims to evidence with explicit review metadata, and support drafting and planning across ongoing projects.

Built as internal AI-assisted systems with self-hosted agent workflows and structured records.

Prototype shapeShared structured core with separate review and methods surfaces
Current wedgeMethods intelligence and review support for infection-associated chronic conditions
Operating modeInternal systems under active development, with selected public framing here
Onyx: self-hosted AI collaborator

A lab AI system used for drafting, planning, documentation, search, project continuity, and operational support. The emphasis is not novelty theater; it is practical reliability inside real scientific work.

Methods intelligence workspace

A structured literature environment for comparing what papers actually did, not just what they claimed. It connects review writing, evidence tracking, and method characterization through one shared record.

Research memory and workflow infrastructure

Internal systems for preserving context across projects, turning recurring tasks into reusable workflows, and keeping scientific operations less fragile and more inspectable.

Design principles

How these systems are built
  • Source-linked records rather than free-floating summaries
  • Separation between public framing, internal prototypes, and private lab operations
  • Human oversight by default for scientific judgment and external-facing use

Why this matters

Evidence over performance

The aim is to make scientific work more structured and reusable: clearer synthesis, better continuity, lower administrative drag, and systems that can be inspected rather than merely trusted.

Current status

Public overview, private prototype

The underlying tools are active internal builds. This page stays public, while interactive prototype surfaces can remain gated during iteration.

Roadmap

What is being built next

Review companion

A claim- and paper-centric surface for synthesis, comparison, and evidence review within a bounded corpus.

Methods characterization

A parallel surface for normalizing what methods papers are actually using so comparable approaches become easier to inspect and reuse.

Broader workflow memory

Extending the same structured substrate into wider research memory and scientific-operations workflows without collapsing everything into one vague tool.

Private by default where it matters

Experimental interactive workspaces may stay unlisted or access-controlled while they are changing quickly. The point of this page is to make the underlying direction visible without pretending unfinished internal tools are polished products.