AI · Automation · Digitalisation
AI that answers routine questions, moves data between your systems and finds the one clause buried in a contract. It runs on your own servers or in your cloud, and you see a working version in days, not weeks.
01What AI costs
AI looks like something only corporations with their own IT and AI teams can afford. In practice, plenty of useful work costs so little that it never has to go to the board.
From a few dozen złoty a month
Sorting an inbox, summarising customer reviews or pulling figures off invoices can cost about as much as a mobile phone contract. The big platform project can wait until the numbers justify it.
Most business work runs on the computers you already own, or on capacity rented by the hour. We size the hardware to the job instead of buying cards that cost tens of thousands.
Suppliers charge very different amounts for exactly the same AI. We compare them and move your work to whichever is cheaper. Nothing else about the system changes.
02Example projects
Each of these is a separate project with a clearly defined scope. You start with one process and expand when you decide it pays off.
Emails, Google reviews, forms and survey answers in one place, marked positive or negative and grouped by the issue behind them.
You see what is going wrong before it becomes a trend
Competitor prices, offers, website changes and announcements tracked automatically, with a finished summary delivered to your inbox.
A weekly report replaces manual checks of competitor websites
It looks like an ordinary chat window. Behind it sits an agent that can open your systems, check the facts and finish the case, inside the limits you set.
Cases closed in the chat, with no callback
03What we do
We start with the process, then pick the cheapest tool that solves it. Sometimes that is software that acts inside your systems, sometimes a statistical method twenty years old, sometimes a properly built data import.
Agents that act in your systems, forecasting built on your own numbers, and any other method the problem calls for.
Simple, precise process automation. We automate the steps worth keeping and remove the ones that add nothing.
AI on your terms: in your cloud or on your own servers, built around the rules you have to follow.
04How we work
An MVP is the first working version, built on your own data. We start by learning how the process runs today, then hand that version over within days for you to judge.
No commitment
We sit down with the people who do the work and draw it out step by step. One session is usually enough.
Within days
We take the part where the effort pays off first and have it running on your data within days.
No lock-in
Extend it, roll it out to everyone, or stop there if it does not earn its keep. Whichever you choose, the code and everything we learned stay with you.
On your systems
It runs inside your own systems. We measure the quality of what it produces and keep an eye on the monthly cost long after launch.
05The first step
You do not need to know whether AI is the answer. A process that takes too much of your time is enough.
We talk to the people who do the work and draw the process out step by step.
You get one page: where the time goes, what the options are, a cost range, and what a first build would cover.
Order the build, do it yourself from the summary, or do nothing. The summary is yours either way.
06Blog
Short pieces on models, agents and cost, written while the work is under way.
TypeSafe AI released a model on 15 September that writes no text at all. It hands back a chosen option and a probability, costs $0.042 per million input tokens, and charges nothing for output. Here is what survives once the marketing is subtracted.
With five skills in the pool, 29.6% of the skills an agent actually uses are the right one; with a hundred, 3.3%. And in August a public skills registry served clones that stole SSH keys. Four rules for a team working with agents.
$12.93 billion for the platform most open AI models pass through. A year ago Hugging Face turned down an NVIDIA investment at a $7 billion valuation. The numbers, the promises, and four things worth doing now.
07Contact
You don’t need a finished brief or an idea for AI. A short description of a process that takes too long is enough, and we’ll work out the rest with you.
Email us directly
