Lab Log

Login
Dashboard
New entry (main feature)
Voice Activated Mode
Example Protocol
Example Dynamic Analysis Tool
Shared with me (sharing between accounts is functional)
Safety Logging
Meeting Notes

By Jethro, Matt, Sidd, and Rishi

Inspiration

Modern lab work is still heavily burdened by inefficient, fragmented documentation—paper notebooks, scattered notes, and inconsistent formatting slow down research and collaboration. We were inspired to reduce this friction by creating a smarter, faster way for scientists to capture, structure, and interact with their lab work in real time, especially at the bench where typing is inconvenient.

What it does

LabLog is an AI-powered digital laboratory notebook that helps researchers quickly log, organize, and analyze their work. Users can create entries via typing, voice commands, or PDF uploads. The system automatically cleans and structures notes, categorizes them, extracts key data, and flags important issues like errors or safety concerns. It also supports reusable calculation tools and allows users to query their own lab history through an intelligent voice assistant.

How we built it

We built LabLog using Next.js for the frontend and integrated Google’s Gemini 2.5 Flash model for AI processing. Voice functionality is powered by a custom assistant (“Jethro”) that handles dictation and queries. The backend processes entries through AI pipelines for cleaning, categorization, markdown formatting, and structured data extraction. Claude code, Antigravity, and Codex is used to develop this. We also implemented sharing features, filtering systems, and a responsive UI optimized for lab environments.

Challenges we ran into

One major challenge was ensuring that AI “cleaning” preserved critical scientific details like measurements, units, and reagent names without distortion. Building reliable voice interactions in a noisy lab setting was also difficult. Additionally, structuring unorganized lab notes into meaningful categories and formats while maintaining flexibility required careful prompt design and iteration.

Accomplishments that we're proud of

We’re proud of creating a seamless multi-modal input system (text, voice, PDF) that reduces friction for researchers. The automatic structuring into clean, readable markdown and the extraction of reusable computational tools are standout features. We also successfully built a context-aware voice assistant that can answer questions based on a user’s own lab data for real-time data analysis.

What we learned

We learned how to design AI systems that augment technical workflows without compromising accuracy. Prompt engineering for scientific contexts requires precision and careful constraints. We also gained experience building real-time, user-friendly interfaces that integrate complex AI pipelines behind the scenes.

What's next for Lab Log

We plan to expand LabLog with native figure integration and sketch-based AI interaction, allowing users to draw experimental setups, workflows, or data trends directly in the app and embed them into their notes or protocols. Building on this, a “sketch-to-AI” feature will let users treat drawings as prompts—enabling the AI to interpret the scenario, answer questions, suggest improvements, or convert rough sketches into clean, labeled figures. This will make visual elements first-class components of the notebook, enabling researchers to think and communicate not just in text, but through diagrams that are searchable, interactive, and deeply integrated into their experimental workflow.