How it works

You talk. It asks the questions a careful new analyst would ask, and a few they would not dare to.

It draws out what is obvious to you and invisible to everyone else, and it never fills in a value you did not give.

One question at a time

No forms. It follows your workflow where it goes and checks back with you in your words.

It chases the tacit

'I drop the odd one' becomes: what are you looking at, and what would change your mind?

It never fills a gap

Anything you did not state is marked Not stated. Every line traces to something you said.

It asks what you would never mention

The file you start from, the copy-paste between tools, the check you do without thinking.

You sign off

Nothing leaves until you approve it. The tool drafts; you stay accountable.

"I have worked alongside scientists running their own analyses who did not know that an automated tool already existed for the step they were doing by eye. Nobody had ever asked them to describe the step."Tricia Estrada, founder

METHODSSynthetic sample session for illustration.

Methods · turn 6

After the run finishes, what happens to the Ct values? Walk me through it from the moment you export them.

You · turn 7

I export to Excel, drop the reference gene next to each target, and work out the ΔΔCt myself. Usually takes an afternoon.

Methods · turn 8

Before the ΔΔCt, do any replicates get excluded? If so, what is the rule?

You · turn 9

If the triplicates disagree I drop the odd one. I go by eye, honestly. There is no number.

Methods · turn 10

Recorded as a judgment call with no stated threshold. Which reference gene, and did you check it is stable across your conditions?

You · turn 11

GAPDH. Everyone in the lab uses it. I have never checked.

Methods · turn 12

Noted as unverified. One more on this step: is the amplification efficiency assumed at 100%, or measured?

You · turn 13

Typing

See this session in the sample app (opens in a new tab)

Writing it down as you go

What Methods is recording

Judgment call

Replicate QC: triplicates that disagree are dropped by eye. Threshold: Not stated.

turn 9

Not stated

Reference gene GAPDH; stability across conditions not checked.

turn 11

Idea to automate

A written replicate-QC rule (Ct standard deviation cut-off) in a script, plus an existing reference-gene stability check.

turns 9 to 11

Walkthrough

Your session. Your input. Your output, customized to your lab.

Screens from the Methods sample app. Click through, or let it play.

Frame 1 of 7. The Methods sample app's Home screen, marked SAMPLE DEMO: example data for illustration, nothing is saved. A banner greets with Good morning and reads Let's dive into the science, three things in the order that unblocks your team fastest; its Start a new session button is muted, with a Live in the full app chip. Three to-do rows follow: sign off the record for qPCR relative quantification, send 3 questions to your team, and confirm 2 new rules.

Frame 2 of 7. Session replay of the synthetic qPCR session, part 1, The journey. In the chat, Methods asks for the very first thing the scientist does, a marker shows turns omitted, the scientist says they loaded the qpcr-protocols folder, and Methods replies. Beside it, What Methods is recording lists the session type, 6 files read and a question asked once.

Frame 3 of 7. Session replay, part 3, The contradiction. Methods points out that the README runs the melt curve only for new primer pairs, the scientist says it runs on every plate, and Methods keeps both accounts. The recording panel collapses parts 1 and 2 and shows a two-accounts card: Said in the session, a melt curve on every plate, beside In the lab's README, only when a primer pair is new.

Frame 4 of 7. On the record, the procedure. Step 05, choose the reference gene, is marked judgment call and threshold not stated. Step 06, check the melt curve, carries a violet marker with the two accounts side by side. Step 07, drop outlier replicates, is marked not stated. Beside them, Questions for your team routes three questions by role with a Send questions button, and the sign-off bar offers Approve, Correct and Hold.

Frame 5 of 7. Templates, with Who does what open. On the right, a sample responsibility matrix (RACI) filled from the qPCR record; a line under the subtitle says names are removed and roles are shown instead, and R, A, C and I chips mark each section and role, with primer efficiency marked Not stated. On the left, what Methods fills in, a link to the approved document already created, and Add to my workspace.

Frame 6 of 7. What your agent knows, with tabs for science, design and procedure. A New since your last session box holds two rules, each with the exact quote and the turn it came from, and Confirm, Edit and Retire buttons. Below it, a confirmed rule from the plate layout file.

Frame 7 of 7. On the record after Approve. Among the documents made from this record, the record itself now reads Approved with a green status dot. The color legend shows written down, judgment call or not stated, question for the team, and planned, and the sign-off bar has Approve selected: approved as version 1. In this demo, nothing is saved.

Try the sample app Example data. Nothing is saved.

A session, start to finish

You bring the knowledge and the analysis.

BEFORE

Review your documents, or talk it through

It starts by asking how you would like to begin: it can review your documentation first and then ask only about what it leaves out, or you can talk through your process from the beginning.
Then map the whole process, or go straight to one step.
No notes needed. It asks about procedure, never results or hypotheses.

DURING

Talk it through

One question at a time, in your words.
Stop whenever you need to and pick it up later.

AFTER

Review and sign off

Your record, with every unstated value marked, and what could be automated.
Exported only when you approve.

BackThe problem NextWhat you get →