Skip to main content
Shipped 2026-04-26 · AI The single biggest complaint about lab notebooks is that they’re a write-only medium: you capture everything, then can’t find anything. Bower’s hybrid search helped. This goes further.

​
What’s new

Bird now reads every note you write — quietly, in the background — and infers which other artifacts in your workspace it’s related to:
  • Protocols — “this experiment looks like it’s running the Western blot protocol you saved in March.”
  • Prior experiments — “this is the third time you’ve measured this binding affinity; here are the previous two.”
  • Attachments — “this gel image looks like the one in note 2026-03-14 competition assay.”
  • Concepts — “you mention CRISPR-Cas9 here and in 12 other notes.”
These connections appear as inline link chips at the top of the note view. Click any chip to jump to the related artifact.

​
How this differs from manual linking

Linked entities are explicit, user-created connections — you decide what links to what. Inferred relationships are suggestions from Bird, made by reading your notes and finding artifacts they relate to. The two coexist: manual links carry full weight, inferred links sit alongside them and are clearly labelled so you always know what’s authoritative.

​
Where they come from

Inferred relationships run on a nightly batch (no real-time API call per note, no foreground latency). The pipeline uses semantic similarity (embeddings) + concept extraction + a graph-aware ranking model. Per-artifact link audits in the graph view let you accept or reject suggestions; rejected suggestions won’t reappear.

​
Why this matters

The links you’d build manually if you had infinite time — Bird builds them for you, automatically, and updates them as your workspace grows. The result is a notebook that gets more useful over time, not less.