EnrichNode
For EnrichNode AB we built EnrichNode, a service that lets sales and marketing teams search among Swedish companies and public tenders to reach the right decision-makers. All data comes from free government sources. We delivered company search by sector, industry code and location across the 1.8 million organisations in the Swedish business register, plus a summary of each company’s website. The service also includes tender monitoring, where new Swedish notices are imported every hour from TED, the EU’s tender database, and matched to companies by registration number. Salespeople see who is buying, who won and whom to contact. Every data point shows its source, when it was retrieved and how certain it is. A contact channel only counts as confirmed when the registration number or the exact domain appears on the company’s own website. That rule is in the code, and no AI can get around it. The service is built for GDPR from the ground up, with compliance as a step in the data flow itself. Only government data is handled, with exclusion of anyone who has registered an advertising block (reklamspärr), a source trail on every data point and a time limit on personal data. The live product covers company search, website summaries and tender monitoring. Confirmed contact channels, the advertising block and the retention period are built and will go live in the fourth quarter.
Tiny House Factory Sverige
For Tiny House Factory Sverige, which sells organic Attefall houses, we are building the entire digital presence, which includes the website, image production, customer flow, visibility and back-end systems. We rebuilt the website from the ground up in Swedish, with dedicated pages for five house models. We also set up AI-driven image production that shows each house in different colours, facades and seasons, from the ground, from the air and on video. Enquiries arrive through a protected contact form and a meeting booking tool. The website is built to be found both on Google and in AI assistants such as ChatGPT, and an admin panel raises an alert if search engines and AI services stop visiting it. Development runs on reviewed and tested code, and every claim about the houses is checked against the Swedish Marketing Practices Act. Anything that cannot be substantiated is not published. We moved the website to a new domain and host it in our own server environment. We are currently building a Finnish version of the website, the interiors for the fifth house model and an in-house customer database on a server within the EU. The result is a live website at tinyhousefactory.se that receives enquiries and only claims what the company can back up. We have also built an in-house CRM tool and a goal-management platform with a range of dashboards.
Articulate IT
For Articulate IT AB, we are building the technical foundation and the application layer of a product for sales teams. We built the first version of the platform and continue to develop it. One technical challenge was that one of the product’s data sources lacks information users need. We solved this with an architecture that automatically fills the gaps using free public registers. Development runs on reviewed code that is merged continuously into the production codebase, and the product has been designed from the ground up with compliance and data security built in, including documented architecture decisions and strict separation of each customer’s data. The product is live, and development is ongoing.
The product uses AI, and that component is subject to the same compliance and data security requirements as the rest of the product.
Gap analysis of self-inspections against Swedish building rules
A building contractor’s self-inspections had been rejected by the client. The documentation existed, but there was no way to show that the inspections met what the regulations actually require. We built a system that reads the construction contract and the inspection documents using OCR. The system tests each individual inspection against the following three regulated requirements: which document it was performed against, which method was used, and whether the work was approved. The system also flags the items in the inspection plan where evidence is missing altogether, the most common reason a self-inspection is rejected after the fact. The output is a finished contract text in Swedish that the client can pass on without rewriting.
The system uses AI to test each inspection against the requirements.
Technical due diligence
For an investor client, we reviewed the technology and code of an investment target (a communications startup).
The work answered three questions. First, whether the product lives up to its claims, which was determined through code review. Second, whether the company actually owns what is being acquired, which we tested by reviewing the chain of title for its intellectual property, including open-source and licence terms. Third, whether the deal covers its own costs, for which we reviewed the unit economics behind the figures in the investment material.
Our deliverable was a written report to support the investment decision.
