Software that runs the new model.
Viher IT is a small software company in Oulu. We build the systems a business runs on — web platforms, product data models, and the devices in the field that feed them. The point is never the software. The point is the operating model it makes possible.
A new tool inside an old model changes nothing.
Most software projects hand back the same business, slightly faster. Who does what, which data moves between them, where the money enters and leaves — unchanged. That is a cost line, not a strategy.
The trap has a newer form. Generative AI made building so cheap that a founder can ship a working product in a week — and cheap building removed the one force that used to make anybody check whether the product was wanted. Faster tooling, same model, and now the mistake is affordable enough to repeat. That is the same failure as the enterprise rollout that changes nothing, arriving from the opposite direction.
The new model is usually sitting in data you already produce and have never modelled: origin, batch, delivery, usage. As free text it is overhead. Structured and provable, it sells — documentation that generates itself, a claim a buyer pays a premium for, a platform position instead of a supplier one.
So we write the software. A Rails or Next.js platform on PostgreSQL, the data model under it, the integrations to payments, invoicing, and logistics — and where the model reaches into the physical world, the devices too: 4G routers, edge agents that hold state through an outage, locks, sensors, all managed remotely. AI agents in the loop, one person’s name on the result. Read how we work or meet the founder.
Five fixed-scope engagements. Most of the work is the build. Three of the others exist because product data has to be right before software built on top of it means anything. The fifth exists because a product built in a week still has to be one somebody wants.
The system that runs it.
Web platforms, the data model underneath, the integrations either side, and the field devices when the model reaches into the physical world. Rails and Next.js on the server, edge agents on the hardware, built so the model on paper actually runs in production.
02 / Claims auditKnowing which claims survive.
An item-by-item audit of every environmental, origin, and sustainability claim on your webshop, product pages, packaging copy, and social channels — each one sorted into keep, keep with evidence attached, or remove before the new EU rules start being enforced. The register also shows which claims are worth building on.
03 / Product dataOrigin data that holds up.
Origin fields, batch identifiers, immutable records, and an audit trail — designed into your product data model, webshop, or PIM, so that a claim on the product page can be traced back to the batch it actually came from — and so the same data can carry a new business model, not just a compliance answer.
04 / DocumentationThe paperwork, produced automatically.
Omavalvonta and origin documentation for small producers and wholesalers — generated from records you already keep, instead of retyped by hand every time a professional-kitchen buyer asks for it. Paperwork stops being the reason wholesale accounts stay closed.
05 / ValidationProof before the build.
Two weeks that turn “we are building X” into a written list of what has to be true, the cheapest test for each of those things, and a decision rule agreed before the data arrives. Where the fastest test is a working artefact, we build the artefact — a landing page, a clickable prototype, a thin slice that runs.
What actually changed, and what only got faster.
Risto is running an interview study on how early-stage founders use generative AI across the whole innovation process — idea, concept, prototype, validation, business model — rather than in the build alone. Six to eight founders in the Oulu region, semi-structured interviews, qualitative content analysis, through Oulu University of Applied Sciences.
There are no findings yet. Fieldwork starts in November 2026 and the thesis is published openly in spring 2027. Until then the page describes a question and a method, and says so plainly — which is the same standard we hold the claims on the rest of this site to. Read what is being studied, or see the method applied to a product in the validation sprint.