We find where AI actually pays inside your operation, and we build it.

Most companies can feel that time and money go somewhere they cannot see, but not where. We go through the operation step by step, find the places it happens, and build the systems that take the load off. Where the answer is not AI, we say so.

Request a pre-study
The shipyard crane and the old brick buildings at Hietalahti, Helsinki.

What we look for

The problem is rarely that people are doing the work badly.

It is almost always that information does not move, so the work gets done twice or gets done on a guess. Three things we look for first.

The same information entered more than once

A job goes into one system, a document into another, and a note on paper travels along with it. Systems that were integrated years ago still leave the manual step in place. It costs real hours every day and shows up in no report anywhere.

Decisions made before the information arrives

A delivery date is promised before anyone knows when the materials land. Work is scheduled before anyone knows what it will really take. The estimate is not wrong because people are careless. It is wrong because the information does not exist yet at the moment someone has to commit.

Margin nobody can see

Costs are known at the level of the company, but often not at the level of the job, the product, or the site. Teams end up being asked to improve a number they have no way to measure, which is also why nobody can tell whether last year's fix worked.

Work

Autoklinikka

One of Finland's largest vehicle damage repair chains. A national network of workshops and over a hundred thousand repairs a year. Pre-study, June 2026.

We spent a week inside the company looking at where AI and automation could be used.

We interviewed people across every role, from the parts team to regional managers, and worked from the company's own operational data. The report covered the whole repair process from the first customer contact through estimation, parts procurement, the shop floor and invoicing. It identified the bottlenecks, traced them back to their causes, set out what could be built for each one, and put the work in a staged order so that the measurement comes before the optimisation.

The figures in that report are Autoklinikka's own and stay confidential. We do not publish a client's numbers to sell the next engagement.

The signed reference, in the original Finnish. Click to read.

“They got to grips with our business quickly and produced a clear, well structured piece of work in a short time. They had a strong grasp of both the data and the operational side, and the collaboration was straightforward and professional.”
Antti Toikkanen Business Controller, Autoklinikka‑yhtiöt Oy. Signed reference, 30 June 2026. Translated from Finnish.

What we do

Two things. We look, then we build.

Every engagement starts with the pre-study. We will not quote a build for an operation we have not seen from the inside.

Pre-study

Two weeks. Fixed scope and price.

Two weeks inside the operation. We interview the people actually doing the work, not only management, and we work from your own data rather than from what the org chart says should be happening.

You get a written report: where the time and the money go, what is causing it, what can be fixed, what each fix is worth wherever that can honestly be calculated, and the order the fixes should be done in.

Build

Scoped from the report.

We build what the report justified, one piece at a time, on top of the systems you already run. Each piece works on its own and is useful on its own, so there is no single large project that everything else depends on.

Usually that is a monthly report somebody assembles by hand, two systems that do not talk, a decision being made without a number in front of it, or documents a person has to read before approving.

Deployed, documented, and handed over so your own people can run it without us.

The Helsinki waterfront seen from the sea.

How we work

Everything in the report has to survive you checking it.

Four rules follow from that, and they are the reason some numbers in our reports are missing.

01

Your data, or nothing

Every figure in the report comes out of your own systems, with the period it covers written next to it. We do not use industry averages, and we do not invent improvement percentages.

02

If we cannot measure it, we say so

Some benefits are real but not yet measurable. The report says so, and states what measurement would settle it.

03

AI where AI is the answer

Some of what we find is solved by a model. Some of it is solved by a script, an integration, or a change to how work is scheduled. We recommend whichever one works.

04

You know the running cost before you commit

Running these systems is usually a few thousand euros a year in model usage, not a licence in the hundreds of thousands. That number goes in the report, calculated on your volumes, before anyone signs anything.

Helsinki Central Station and the surrounding streets in snow at dusk.

The firm

Two people. The ones you meet are the ones who do the work.

Rossblad is Tobias Rossi and Samuel Liljeblad, based in Helsinki. There is no team behind us that the work gets handed down to, which is the point: at a larger firm you meet the partners and get the juniors.

More about us

A building corner and tram wire in Helsinki.

Starting

Every engagement starts with the pre-study.

Fixed on both sides before the first day, so the only thing left to find out is whether the work is worth doing.

Length
Two weeks. Scope and a fixed price agreed before the first day.
What you get
The written report and the analysis behind it, yours whether or not we build anything afterwards.
Obligation
None. If the report says the work is not worth doing, or is worth doing by someone else, that is what it will say.
Confidentiality
We sign your NDA before the first meeting. Access to systems and data is scoped and agreed in writing beforehand.

Tell us what you think is going wrong, and we will tell you whether we can help.

contact@rossblad.com

Engagements run in Finnish. Interviews, workshops and the report itself are all in Finnish unless you would rather they were not.

Reference, Autoklinikka‑yhtiöt Oy, 30.6.2026
Signed reference letter from Autoklinikka-yhtiot Oy confirming a one week AI pre-study carried out in June 2026.