A free one-week proof of concept
Pick one process. We build a working version of it in a week, on your data, and you decide afterwards whether it is worth paying for.
What it actually is
Not a slide deck, not a feasibility study, and not a trial account. A working thing you can click, running against your own documents or your own data, doing one job you already do by hand.
One week is a real constraint, not a sales figure. It forces the scope down to something that can be proved, and it means you find out quickly if the answer is no.
Who it is for
We can only give this away if it has a fair chance of turning into work, so there are three conditions. They are about the shape of the problem, not the size of your company.
- A decision-maker is in the room β someone who could say yes to the next step takes part in the 20-minute call and sees the demo.
- The process is concrete β you can name it in one sentence β "reading supplier invoices into our ERP", not "using AI in finance".
- A priced next step is agreed β before we start, we agree what happens if it works: the subscription, or a fixed-price build at a number we have both seen.
If your need is broader than one process, the fixed-price diagnostic is the honest starting point instead.
What happens, in order
- 20-minute call β you describe the process, we say whether a week is enough to prove it. Sometimes the answer is no, and we say so on the call.
- Scoping β we write down exactly what the proof of concept will and will not do, and what we need from you β sample documents, access, a person to ask questions of.
- One-week build β we build it. You do not need to manage us during the week.
- Demo β we show it running on your material, and we are specific about what it gets wrong as well as what it gets right.
- Decision β you say go or no-go. No-go is a normal outcome and costs you nothing.
If it works, what then?
The proof of concept is free. Putting it into daily use is not, and there are two ways to pay for that.
- Automation subscription, β¬300 a month β we run and maintain it, and add one small automation a month. Cancel any month.
- A fixed-price build β when it needs to become a real product rather than a running script. Typically β¬12,000β25,000 depending on scope, quoted after scoping.
Data, NDAs and the awkward questions
What happens to the documents we give you?
We use them only to build and demonstrate the proof of concept, we keep them for the duration of the POC, and we delete them on request at any point. We do not use your material to train models.
Will our data be sent to an AI provider?
Usually yes β that is what makes these systems work β and we tell you exactly which provider and what leaves your premises before we start. If that is not acceptable, say so on the call: some processes can be built with models that run on your own hardware, and some cannot. We will tell you which yours is.
Will you sign an NDA?
Yes, and we would rather sign yours than argue about ours. Send it before the scoping call.
Who owns the code from the proof of concept?
It stays with MALI Labs until the first invoice is paid. When you pay the first invoice for the follow-on work, ownership of the code we wrote for you transfers to you. General-purpose tools and libraries we reuse across projects stay ours; you get a perpetual licence to use them as part of your system.
What if we stop after the POC?
Then we stop. You owe nothing, the findings from the demo are yours to use, the code stays with us, and we delete your data on request.
Why would you work a week for free?
Because describing what we do never convinces anyone, and showing it usually does. Roughly speaking, the ones that work turn into paid work often enough to cover the ones that do not.
Name one process
Twenty minutes is enough to tell whether it can be proved in a week. If it cannot, we will say what would be needed instead.