Cost reduction
Most companies have savings sitting in production, administration and logistics that nobody has gone looking for. We find the ones worth acting on, and we start with the cheap fixes.
Kurumi Innovations
We find the technology changes that pay for themselves: lower costs, better quality, higher throughput. Sometimes that means AI. Often it means something a lot simpler.
Get in touchWhat we do
Most companies have savings sitting in production, administration and logistics that nobody has gone looking for. We find the ones worth acting on, and we start with the cheap fixes.
Higher quality usually costs less than the defects it prevents. Often a single extra automated check is enough to pay for itself.
When you are running at full capacity, expanding is expensive. Removing the bottleneck usually is not.
Work
PostNL
Recognition software for parcels, similar to facial recognition but for shipments, used to follow individual parcels as they move through the sorting process.
015 Keramiek
A parametric design algorithm that reconciles contemporary design with additive manufacturing and artisanal ceramics. The collaboration is still running: it now includes a Python and Blender pipeline that takes a 3D design through to 3D-printed plaster moulds, cutting the route from concept to production.
Zero Foodwaste
Helped shape the AI roadmap for a food waste startup: which problems were worth solving with machine learning, in what order, and what each one would take.
Clare
The LLM pipeline behind an AI-driven communication coach for couples. Still in development, with launch planned for later this year.
We do not publish client numbers. Ask in a first conversation and we will walk you through what these projects delivered, and what we would do differently now.
Why Kurumi
Technology is what Kurumi is about. Even so, no team keeps up with everything new that gets built every day, and it does not need to.
We work out which technology matters for your business, and bring in a specialist from our network to carry it out.
That way you get something shaped around your situation, without being tied to one supplier with one answer.
There are ten ways to solve any problem. Usually only one or two pay for themselves within a reasonable time, and those are the ones we care about.
Kurumi starts from what you need, not from what we are good at selling.
If you want something that looks impressive, because it helps position your business for example, we will build it. But we would rather deliver the result by the shortest route.
We would rather work with a client for years than for weeks. That means you have to be able to rely on us, including when something goes wrong.
So we keep to what we agree, handle your data and systems carefully, and plan so the long term holds up as well as the short.
And it means we will talk you out of a project sooner than sell you one.
Approach
Traditional consulting works top-down: map the inefficiencies, optimise the processes, project the return. On paper it adds up. What it misses is the people who have to live with the result every day. We start at the other end, on the floor, with how the work actually gets done and where the time actually goes.
Free, around an hour. You tell us where it hurts. We tell you honestly whether we are the right people to help.
One to two weeks, mostly on site. We map the relevant processes and come back with a ranked list of opportunities. Each one gets an estimated investment, an estimated saving and a payback period. Fixed price, agreed before we start.
We take the strongest opportunity and prove it at small scale, before you commit to anything larger.
We implement together with our network of specialists, then hand over documentation and training so your own team owns the result.
You can stop after any step. The scan is useful on its own, and it is yours to keep whether or not you build anything with us.
Team
Lorenzo founded Kurumi after years of building machine learning that had to work in production rather than in a demo. He spent three years as an R&D computer vision engineer at Prime Vision in Delft, and later three years as its Product Manager R&D, leading a team of more than 25 computer vision scientists and engineers across the postal, logistics, retail and medical markets.
Before that he took a PhD in nanotechnology at the University of Cambridge, following a master's in condensed matter physics in Trieste. He has co-authored 23 peer-reviewed papers with more than 1000 citations, and reviews articles for Nature Communications, Applied Physics Letters and the Journal of Applied Physics.
The best part of the job is still the hour spent on the floor, watching how the work actually gets done. That is usually where the answer is.
Around this core, Kurumi works with a standing network of specialists in vision systems, data engineering, control systems and process automation. You get the specialist your problem actually needs, not the one who happened to be free.
Contact
We are glad to look with you at where your business could be stronger.
Ask your question through the form and we will come back to you quickly.
The first conversation is free, takes about an hour, and commits you to nothing.
Papenstraat 73, 2611 JB Delft
© OpenStreetMap