FAQ

That is the starting point, not an obstacle. The first day maps out exactly what you have: spreadsheets, an outdated CRM, loose exports. The build starts with what exists, with an honest read of what needs tidying up first.

Usually not. Where possible, everything runs on tools and systems you already have. Where an extra subscription genuinely helps, that is reviewed with you beforehand — the choice and the cost stay yours.

Working software that runs on your own data, explained to the people who will use it. Not a report with recommendations, but something you can open and use straight away.

No. Kufu is set up for small and mid-sized businesses: companies where one person handles IT alongside everything else, and where a months-long external implementation project isn’t an option. That’s why the packages are short and fixed-price.

Your team gets the keys: how it works, the code, and a short handover where needed. Want to keep building? That’s what Individual days is for, at a fixed day rate — with no new engagement to set up.

After the free intro call, a date follows by mutual agreement. That’s usually possible within a few weeks, depending on schedules on both sides. There’s no internal approval process or tender ahead of it — if it fits, the date gets set quickly.

The scan and build days happen on site, with eyes on how your processes actually run and conversations with the people who work with them every day. For Individual days, after the first five days, that can sometimes move to remote by agreement — settled case by case.

Signing a non-disclosure agreement is not a problem. Kufu works with the data and systems found on site, doesn’t take data away without discussing it first, and only asks for the access needed to carry out the engagement. Where the software processes personal data belonging to your own customers — for example lead or email follow-up — Kufu is your processor for that work and you’re the controller. Before that work starts, a data processing agreement gets put in place, alongside the NDA, not instead of it.

Yes. The scan looks at what’s already there, with an honest read on what’s worth keeping and what’s better rebuilt. The engagement doesn’t start from a blank page — it builds on what already works.

Yes, if you want that. Coordination happens with whoever manages your systems, without replacing anyone, and the handover is just as clear to them if maintenance is needed after the engagement ends.

Usually not. What tends to disappear is the retyping and the chasing up — the administrative part of a task, not the person doing it. If a role consists purely of that kind of repetitive work, that will change; that belongs on the table beforehand, not afterwards.

Not every piece of software needs an AI model. Where it does, business data never runs through a personal, consumer-grade account: that always goes through a provider under a data processing agreement. If you’re already working with an AI tool today where that isn’t in place, that is raised explicitly before working with it any further. Your data is not used to train models and stays inside the software built for you. Access stays minimal and time-limited, and for lead and email follow-up — where personal data is unavoidably involved — a data processing agreement gets put in place for that, alongside the non-disclosure agreement.

That happens sometimes, usually because something the software relies on changed — a supplier alters an invoice layout, a system it talks to gets updated. That’s why every failure sends a visible signal to a named person instead of quietly sitting broken, and why an internal owner is agreed who knows what to do. If it needs real work, that’s what Individual days is for: a day at the fixed day rate.

Usually your own people. The software runs on systems you already own — your mailbox, CRM, ERP — or, if it needs a place of its own, in a portable container on a small server you pay for and control directly. The documentation is written for someone without a technical background. Prefer to have Kufu take care of it? That’s available for a day through Individual days.

Safety doesn’t come from reading every line of code — for software of any real size that stopped being a reliable safety net long ago, AI-written or not. It comes from the design: permissions and the data model are settled early, every piece of access gets no more rights than the task needs, and whether something refuses what it shouldn’t allow is tested separately from whether it works. See the approach page for what that looks like in practice.

Yes. The source code, the hosting and the passwords are in your name, not Kufu’s — that’s the default starting point, not an add-on to negotiate separately. If Kufu were ever unavailable, you’d already have what another developer needs to take over: the code, the environment it runs in, and access to both.

That risk is real with any sole practitioner, and it’s limited by technical choices, not a promise: the source code is in your name, the software runs in a portable container on a server you control yourself, and the passwords are yours. Another developer can take over — slower than whoever built it, but not from zero.

Vibe coding usually means AI-generated code shipped on trust, with nobody fully understanding what it does. Kufu works the other way round: permissions and the data model are deliberately designed, access is tested separately from functionality, failures are visible instead of silent, and one person — who also operates the software — knows exactly what it does, how it runs, and what it touches. Building faster with AI isn’t a synonym for building with less control.