ProsjektKlar
From Istoft
See the plot in 3D before you build.
Plot, house and changes in 2D and 3D, with zoning and flood and noise zones.
Try the sun and shade demoAdvisory · AI and data · product development
Tell us about a workflow that takes too much time, or an idea you want to test. We help you decide where to start and build the solution.
A night aerial of a road through a Norwegian landscape. Car lights follow the roads, sensors send data to one shared core, and a calm result panel without numbers appears at the end.
From Istoft
See the plot in 3D before you build.
Plot, house and changes in 2D and 3D, with zoning and flood and noise zones.
Try the sun and shade demoFrom Eilavo
Data and models for systems and AI agents.
One entry point through REST APIs and MCP, with documented sources and access.
GET /api/v1/addresses/search?q=Bergen
Search returns addresses with IDs for further lookups. The API requires a key; the documentation explains how to get access.
Clarify your needs, compare options and decide what to do next. You can engage us for advice alone.
Explore our advisory workNeed to automate a workflow or build a new product? We help with design, development and integrations. You own the code.
See our servicesIstoft develops ProsjektKlar for property and building, and Eilavo develops Datakilde for data and AI agents. Approach AI AS operates them and is the legal contracting entity.
See the productsYou get A concrete recommendation, priorities and a proposed next step.
We review your goals, workflows and data with your team. We compare options and recommend what to try first.
You get Workflows in operation, documented, with an alert when something fails.
We connect the systems you already have so repetitive tasks run on their own. Simple rules where they are enough, AI where it is needed.
You get Assistants your team uses day to day, and an overview of what they can access.
We set up ChatGPT or Claude with your own skills and MCP connections to the data and tools you use, with clear access rights.
You get A prototype tested with users, and a proposal for what to build first.
We make an idea concrete: insight into what users need, a prototype, and an understanding of what AI adds.
You get Tested, documented software, and code you own.
We build software, integrations, AI features and agents.
Let an assistant retrieve a support case or look up figures in your systems. We set up MCP and API connections, limit access and test that it retrieves the right information.
Ask about an internal procedure and find the document behind the answer. We connect document search to a language model using RAG, retain document permissions and test retrieval on questions from your team.
Why do two datasets disagree? We use statistics to find missing values, unusual records and possible links. An agent can help investigate, with calculations your team can check and run again.
Start with one concrete task. We agree what to connect, what you will get, and how to measure whether it works.
Discuss an LLM integrationProperty, buildings and planning information share one workspace. ProsjektKlar connects data sources and tools that would otherwise require several systems.
Try the sun studyA model and a floor plan make the proposal easier to explore. The product combines maps, drawing tools and data in one workflow.
Try the sun studyThe sun study shows how an extension affects shadows in a simplified example. Try different times and seasons in the public demo.
Try the sun studyUnderstand the problem, the users and the intended outcome.
Try assumptions and prototypes before building further.
Build in focused increments with visible results.
Use what we learned to set the next priority.
Much can be solved with simple rules between systems. We use AI where it adds something rules cannot, and say so when a process does not need AI.
Agree sources, access and which information may be sent to external services. Use anonymised examples in the initial discussion.
Agree representative test cases, acceptance criteria and what happens when an answer is wrong or uncertain.
Establish who approves results, how errors are followed up and who owns operation and further development.
founder, accountable for delivery
Leads direction and delivery on every engagement. Anyone else who contributes, and their responsibilities, are agreed before work starts.
Tell us briefly what you want to achieve. A few lines is enough.
Or write directly to post@approach.no