knowledge/tools.md

Lab Tools & Systems

Illustrative sample content for the Lab Assistant RAG template (fictional systems).

LitMap — graph-based literature discovery

A system that helps researchers map a field rather than just rank hits. Instead of returning a ranked list like a standard search engine, LitMap is coverage-oriented: it builds a graph over the literature so a user can see the structure of a research area, discover related work, and navigate connections between papers and ideas. It reflects the lab's broader thesis that graph structure is a first-class signal for understanding and retrieval.

ThriftRoute — cutting LLM API cost through routing + distillation

A method/toolkit that reduces the number of expensive large-LLM API calls through selective invocation (only calling the large model when it is actually needed) combined with online distillation (learning from the large model's outputs so a cheaper model can handle more cases over time). This is the research basis for the cost-aware model routing used inside this very assistant: cheap questions are answered by a small/fast model and only substantive research questions escalate to a stronger model.

How this assistant embodies the lab's research

This lab assistant is itself a small demonstration of several of the lab's lines of work:

• Retrieval-augmented generation (RAG) over the lab's publications — see GraphWeave.
• Citation-/graph-aware retrieval: after vector search, results are expanded one hop along a paper graph (shared authors and topics), echoing GraphWeave and CiteTrace.
• Trustworthy generation: answers are grounded in retrieved sources with inline citations, and the assistant declines to answer when the corpus does not support a claim — reflecting the lab's emphasis on robustness and uncertainty.
• Cost-aware routing: a direct nod to ThriftRoute.