What we're building next

Tell us what to build next.

Every idea here comes from conversations with scientists, lab heads and founders in the last weeks. The ones you ask for get built first, and you hear when they ship.

Exploring

Why didn't this run work?

Theme from conversationsStep-by-step troubleshooting would help; that's where the time goes.

Pick a failed run and Gelly lines it up against the runs that worked: the deviations, the decisions, the lots and the timings that differ, and what to try first.

Tell us, and we'll prioritise it.
Exploring

How often does this protocol work?

Theme from conversationsIndustrial protocols have to work reliably for whoever runs them, not only for the expert who wrote them.

Every protocol shows its success rate across runs, the QC numbers you track, and which changes made it more reliable, so someone who isn't the expert can still run it.

Tell us, and we'll prioritise it.
Exploring

How sure is the AI?

Theme from conversationsTell me “use this” or “take care with that”. A score alone doesn't help a biologist.

Every AI answer, protocol and analysis says in plain words how confident it is and why: thin evidence, out-of-range data, a proxy instead of the real measurement.

Tell us, and we'll prioritise it.
Exploring

Watch new papers and patents in my field

Theme from conversationsWe have to stay permanently up to date on whether our method is still state of the art.

A standing search that reads new papers and patent filings in your area every week, and adds the ones that matter to the notebook with a short note on why.

Tell us, and we'll prioritise it.
Exploring

A slide for the lab meeting

Theme from conversationsHow do I give that person the slide that makes their boss happy?

One tap turns a result into a slide: the figure, the n, the controls and one sentence of conclusion, linked back to the run.

Tell us, and we'll prioritise it.
Exploring

Import our protocols and keep the trade secrets in-house

Theme from conversationsEstablished protocols don't need clever AI, just structure, and their details are trade secrets.

Turn your existing SOPs into Gelly protocols with steps, amounts and timers, using the model you choose: DeepSeek hosted in Paris, or one on your own computer if nothing may leave it.

Tell us, and we'll prioritise it.
Exploring

Find results that look like this one

Theme from conversationsShow me all the Western blots that look like this.

Search your notebooks by picture: find earlier gels, blots and micrographs similar to the one in front of you.

Tell us, and we'll prioritise it.
In development

How much DNA is in each band?

Theme from conversationsWe size the bands already; now I want to compare amounts between lanes.

Band intensity with the background removed, relative to the strongest band and as a share of the lane, with saturated bands flagged.

Tell us, and we'll prioritise it.

Something missing? Suggest a feature or tell us on a call.