makemode
for research · the science notebook

modern ai, without giving up your data.

Keep up with the AI everyone's using — without sending a single byte of your data somewhere your privacy rules won't let it go. Ask a plain-language question, get real analysis back, and your data never leaves your laptop.

The notebook is a friendly research tool: drop in a dataset, ask your question in plain words, and get a real, reproducible analysis back — no coding, and nothing uploaded.

runs in your browser · the eu ai only ever sees your question — never your data itself
the bind you're in

keep up, or keep custody. today you pick one.

Everyone around you is adopting AI — colleagues paste data straight into ChatGPT, Colab, Claude. But your data is about real people: personal, confidential, sensitive. Your privacy rules, your ethics board and the GDPR all say the same thing: not where those tools live. Every product on the market solves one half of the problem.

the whole market in one look — only makemode ticks every box
what you need chatgpt / colab claude science institutional jupyter makemode
runs in the eu
no code needed
self-service, no it ticket
provably compliant, your data stays put
sensitive / clinical data allowed

MakeMode is the only tool that ticks every box. Here's what that means for your institution:

performance

the top-tier ai model

A top-tier AI model for your whole institution — and it's GDPR-approved.

tools

the right tools to build

The right tools to turn any idea into a real concept or prototype — for colleagues, students and teaching.

service

we run it for you

MakeMode provides and maintains the servers, handles onboarding for your institution, and helps where you need it.

01 · potential

start making with ai within your institution.

You don't have to choose between modern and compliant. Drop in a spreadsheet, ask in plain language, and get real analysis back — no programmer, no IT ticket, no cloud you can't defend. From data to finding to a shareable link, in an afternoon.

  1. 1

    a closed door, opened

    AI-assisted analysis used to be off-limits on your data. Now it runs on infrastructure your institution already approves.

  2. 2

    no coding needed

    Ask "did the training raise satisfaction, and is it significant?" and get a real test, a chart and a plain-language answer back.

  3. 3

    your data stays yours

    Keep doing your work at the pace the field now moves — without ever handing over custody of your data.

  4. 4

    even your most sensitive data

    Interviews, patient records — the data no other tool can touch without an upload. That's exactly what we're opening up.

02 · tools

the compliance story isn't a promise — it's a test you can run right now.

The "fits policy" half. The analysis runs right in your browser — zero bytes of your data leave the device. The EU AI, when it's used at all, only ever sees your question — never a single row of your data. We don't ask for trust; we hand you the test.

in-browser notebook

Real analysis running on your own machine. Nothing to install, nothing uploaded.

prove it yourself

Watch the data-out counter sit at 0. Then unplug from the internet and keep working — a test no US tool can pass.

a desktop version for heavy work

A full analysis engine on your own machine — all the heavy tools, your graphics card, big files — and your data still never leaves.

own your work, no lock-in

Export a standard notebook file that runs anywhere. Publish a link only when you choose to. Open standards throughout.

the security level, in your institution's own words.

You set how much may leave — before you drop a file. The classes are TU Delft's StorageFinder terms, not ours.

critical

0 bytes leave

Personal, confidential, human-subjects data. Runs fully on your device — you can prove it in airplane mode.

sensitive

derived facts only

The model may see schema and summaries — never raw identifiable rows.

standard

public / non-identifiable

Public data can use the EU model freely. You choose, up front.

03 · skills

the model doesn't guess the method — it's handed one, and you can read it.

The real question is "but does a non-coder actually get the statistics right?" Curated skills — small, readable method guides, hosted in the EU — load into your build the moment your question needs them. Ask "is the difference significant?" and the right effect-size and survey methods load automatically. The AI follows a vetted method, not a guess.

survey & likert effect sizes + apa reliability / psychometrics a/b & doe time-series / fft curve-fitting correlation / regression honest eda

keyed to your question

You describe what you're working on; the right skills load into the build. No menu of statistical tests to get wrong.

readable, not a black box

Open and read the exact skill the AI was given — and the code it wrote. The method is visible, every time.

portable, no lock-in

They're an open, standard format — the same one professional coding tools use. Curate and import your own; never locked in.

more coming per field

Each new method is one readable card, not a rebuild — the shelf grows toward the workload you bring.

power → policy → rigor

miss any one, and it stays out of reach.

Potential gives you the reason to want the AI at all. Tools removes the thing that actually keeps you out today: your privacy rules. Skills removes the doubt that would make you distrust the result: getting the method right.

Claude Science is powerful — on US servers, for people who already code, with clinical use prohibited. MakeMode is the EU-sovereign version for researchers, students and teachers who don't code and can't upload.

not only for scientists

one tool, everyone in your institution.

The same safe, plain-language tool works across teaching and research — not just in the lab.

students

your thesis data

Analyse your own data without learning to code, and share the result with your supervisor.

teachers

a live data lab in class

Run an exercise where every student works in the same EU-safe tool — nothing to install.

researchers

data to finding, fast

Go from a fresh dataset to a reproducible finding in an afternoon, with your data staying yours.

your data never has to leave your laptop.

Ask a real question of your real data — and watch it stay right where it is.