Skip to content

Research Notebook

A block editor with inline deterministic calculations.

Last updated July 2026

The Research Notebook is a block-based electronic lab notebook (ELN). Each entry is composed of blocks you stack and rearrange, mixing prose, live computation, and references in a single timestamped record. Entries belong to notebooks, and notebooks can be kept private or shared within your team. Click New notebook on the Notebook page to start one.

Building an Entry from Blocks

Compose an entry from typed blocks: rich-text paragraphs, headings, checklists and lists, quotes and callouts, tables, figures, and dividers. Add link / embed blocks that reference your protocols, papers, and findings, and create backlinks so an entry shows what else points to it. Everything stays in one continuous, reorderable document.

Python Cells

Insert a code block to run Python directly in the entry. The code executes in your browser via Pyodide, in a Web Worker, so your data never leaves your machine to compute a result. Output and figures are captured inline alongside the rest of the entry.

Calc Blocks

Insert a deterministic calc block: pick any calculation from the same verified registry used across the app (including the Engineering Workspace tools), enter your inputs, and the result recomputes live from those inputs via a pure engine. There is no AI in the numbers — the block stores only the tool id and your inputs, so it is always re-derivable and auditable.

Journal, Compute & Timeline Views

Work the notebook three ways: the Journal view for reading and writing entries, the Compute view focused on code and calc blocks, and the Timeline view that lays out your entries chronologically so you can trace how an investigation unfolded.

Signing & Witnessing

When an entry is final, sign it to seal a tamper-evident record (a Part 11-inspired workflow — not a certified or validated Part 11 system), and have a colleague apply a witness signature. Each signature is bound to a SHA-256 content hash of the exact entry it was applied to, so any later edit is detectable: the recomputed hash no longer matches, and the entry is flagged as changed since signing.

Your data stays local where it can
Python cells run in your browser via Pyodide, and calc blocks compute from a pure, tested engine — no AI touches either result. Signed entries are sealed with a content hash so the record is auditable and any tampering is detectable.
Related
Was this page helpful?Send feedback