What's New in Pipette: From Plans You Approve to Figures You Edit

Pipette.bio Team

Jun. 5, 2026

We've been heads-down for the last couple of months, and Pipette looks and feels noticeably different than it did in early April. Some of these changes are big swings, others are quiet quality-of-life wins. Here's the tour of what landed and, more importantly, what it means for you.

You approve the plan before anything runs

This is the change we're most excited about. Pipette no longer jumps straight from your question to running code. Instead, it reads your request, thinks through the approach, and shows you a plan first: what it intends to do, which tools it will use, and how it will get to your answer. You look it over, and only when you give the nod does any code actually run.

Why does this matter? Because a few seconds of "yes, that's what I meant" up front saves you from a long run that quietly went the wrong direction. It also means there are no surprises. You always know what Pipette is about to do with your data before it does it. The plan shows up right in the chat, and it carries through into your exported PDF reports too.

A brand-new interface

We rebuilt the entire Pipette web app from the ground up. Everything is faster, smoother, and more consistent than the old version. If you've logged in recently, you're already using it. The old interface has been retired and everything now lives in one place.

Edit your figures right in the browser

Publication figures are never quite right on the first pass. A label overlaps, a color is off, the title needs a tweak. Until now, fixing that meant re-running the analysis. Not anymore.

Every figure Pipette produces now comes with an editable version you can open directly in your browser. Change the text, adjust the colors, nudge things around, and save. No re-running, no waiting, no asking the agent to try again. It's your figure, so you get to put the finishing touches on it yourself.

A lot more science under the hood

We added a wave of new analysis capabilities over the last two months. If your work touches any of these areas, Pipette can now help:

• Microbiome: differential abundance and association testing with the tools researchers actually trust (MaAsLin2, ANCOM-BC, ALDEx2), covariate testing, and machine-learning classification. QIIME2 workflows now run cleanly end to end.
• Comparative and evolutionary genomics: orthology with OrthoFinder, plant synteny with JCVI/MCscan, plus bacterial synteny and mobile-element detection.
• Protein work: fast local homology search against SwissProt with Diamond BLAST.
• Motif discovery: the full MEME Suite for finding sequence motifs.
• Human and population genetics: GWAS and polygenic risk scores, liftover between genome builds, fine-mapping, mutational signatures, cancer cohort analysis, methylation clocks, and sample-swap checks.

In short, more of the analyses you'd normally stitch together by hand are now a single conversation away.

Long jobs are sturdier

Real bioinformatics runs are long, and a lot of our work went into making sure they finish.

• Jobs can now run up to 20 hours, up from 15, so the heavy pipelines have room to complete.
• If a run hits the time limit, you can pick it right back up in the same session instead of starting over.
• A clear Stop button lets you cancel a run whenever you want, and it works even while a job is just starting up or moving between steps.
• You can watch your background jobs from a dedicated panel, complete with a live elapsed-time counter and copyable job IDs for your records.
• The chat now rides out brief server hiccups by quietly reconnecting instead of throwing an error in your face.
• Downloading a big bundle of results as a ZIP no longer chokes on large files. It streams everything through smoothly.

The theme here is simple: fewer ways for a long run to be wasted.

Smarter handling of your files

Getting data in and results out should be the easy part, so we tightened it up.

• A new "Select Inputs" picker lets you tell Pipette exactly which files to use, with no ambiguity, and that choice flows straight into the plan.
• Uploads now use a resilient engine that supports pausing and resuming, and automatically retries any piece that fails. Big files, flaky connections, no problem.
• The data view loads faster, lets you multi-select and bulk-delete, hide finished jobs you're done with, and rename folders so your workspace stays tidy.
• Internal scratch files and bulky intermediates are kept out of your way, and you can preview files without downloading them.

Better-looking, more honest results

A figure is only useful if you can read it and trust it.

• Pipette now actively guards against overlapping text and unreadable labels, and the built-in reviewer double-checks figure quality before handing results back.
• For pathway enrichment, we closed a subtle issue where the background gene set could be silently trimmed. Now nothing gets quietly dropped behind the scenes.
• The agent makes full use of the compute you're paying for, spreading work across all available CPUs instead of leaving cores idle.

That's the spring update

A new interface, a plan you approve before anything runs, an in-browser figure editor, a stack of new analyses, and a lot of work making long runs reliable. We build Pipette to be the bioinformatics collaborator you can actually hand real work to, and every one of these changes pushes in that direction.

As always, we'd love to hear what you think and what you'd like to see next. Happy analyzing.

Try it at pipette.bio. Your first 20 credits each month are free.