LabLog: A Structured Assistant for Experimental Notes

Many engineering and research teams rely on manual note taking while running hardware experiments. These notes capture important context about test conditions, setup changes, unexpected behaviours, and small adjustments that rarely make it into formal reports. However, this information is often scattered across notebooks, voice memos, screenshots, and emails. As a result, the details behind how a dataset was produced can be difficult to reconstruct later.

As part of work on structured measurement workflows, we built LabLog, a virtual assistant that supports disciplined experiment documentation. The goal is straightforward: help engineers record what happened during a test, organise it into a coherent structure, and produce an evidence-ready record that sits alongside the captured data.

What LabLog Does

LabLog takes unstructured notes generated during an experiment and converts them into a structured record with:

• a clear goal statement

• the experimental setup and configuration

• the step-by-step procedure

• observations and intermediate results

• issues encountered and their likely causes

• next steps or follow-up actions

Rather than simply formatting the notes, LabLog also reviews the completeness of the record. It identifies missing context, asks clarifying questions, and loops through revisions until the documentation is comprehensive. The result is a consistent, machine-generated report that captures both the intended procedure and the practical realities of the test.

Once the record is complete, LabLog prepares a PDF that can be stored alongside the raw data files. This allows future analysis to be anchored to a clear understanding of how the data was produced and what conditions were present at the time. For sensing and measurement work, this is particularly important because small, undocumented details often explain noise sources, drift behaviour, or unexpected coupling effects.

Why This Matters for Measurement Work

Early-stage sensing projects depend heavily on understanding the context in which data was generated. Subtle issues such as thermal drift, mounting changes, intermittent contacts, or blocked fluid pathways often only appear in informal notes. When these details are lost, teams may spend significant time reinterpreting data or repeating tests.

A tool like LabLog supports:

• reproducibility of experiments

• traceability of measurement context

• clearer communication between team members

• higher-quality evidence packs

• better linkage between raw data and field conditions

For Senstrali’s work, where structured testing and measurement integrity are central, maintaining a disciplined record of experimental behaviour is essential. LabLog helps capture the real progression of a test, including the false starts and unexpected behaviours that often matter more than the final clean result.

Demonstration Video

The video below shows LabLog in use: generating structured documentation from unstructured experimental notes, identifying missing context, revising the record, and exporting a final PDF suitable for archiving with measurement data.

Next Steps

Future improvements will focus on expanding the types of inputs the system can handle, including:

• speech-to-text for hands-free note taking during experiments

• image input for capturing equipment configurations or oscilloscope traces

• running the tool locally rather than through cloud notebooks for better integration with lab workflows

These capabilities will further support disciplined measurement practice and create clearer pathways between experiment setup, recorded behaviour, and downstream analysis.