Development
Architecture and development of AI solutions, agents and assistants connected to core systems and integrations. From prototype to production — secure, observable and maintainable.
Hi, I'm
Kristoffer Risa.
An architect at REMA 1000, where since January I have been making artificial intelligence something people actually use — and the business can measure. Before that: fifteen years of integrations, retail systems and a stint at sea. In the evenings I build small tools, mostly to understand how things work.
Right now: AI adoption at REMA 1000 It is in Norway
01Now
AI at REMA 1000, since January 2026
My main job this year has been building REMA 1000's AI capability: from the first solutions in production to broad adoption, responsible governance and benefits we can measure.
Adoption
Making AI an everyday tool. Training, champions and use cases that meet real needs across the business — not just in IT.
Governance
A framework for responsible use: policies, risk assessment, data access and compliance with regulation such as the EU AI Act and GDPR — without slowing innovation down.
Benefits realisation
From potential to impact. Benefit maps, measurement and follow-up, so every AI investment can be explained in time saved, quality and business value.
The hard part of AI is rarely the model. It is getting it adopted, governed and measured.
02Person
The person behind the title
I like things that connect. It started with communications on a frigate and printers at Ricoh, became retail systems for a thousand stores, and is today language models talking to the core systems of one of Norway's largest grocery chains.
01 — At sea
It started on F310
In 2007 I served on the frigate HNoMS Fridtjof Nansen, in artillery and communications. There I learned what I still bring into every architecture: systems have to work when it matters.
59°54′ N · 10°44′ E
Photo: Wikimedia CommonsNeovim and dotfiles
I would rather never leave the keyboard. So this site has shortcuts —
Small tools, done properly
CLIs in Go for the weather, the power at home and Confluence — and a set I let AI agents work in, instead of MCP servers. All of it open on GitHub.
Problems are inevitable. And they are soluble — with better explanations.My reading of The Beginning of Infinity, David Deutsch
03Experience
Career
-
2024 — now
AI and Integration Architect
REMA 1000 NorwayJoined as integration architect, responsible for the integration landscape between stores, the chain and suppliers. Since January 2026 my primary focus has been AI: development, adoption, governance and benefits realisation across the business.
-
2020 — 2024
Senior Solutions Architect
Value Retail ConsultingAdvisor and architect for retail clients. End-to-end solutions for stores, point of sale and integrations, from needs and target architecture to delivery.
-
2019 — 2020
Solutions Architect
The Norwegian Directorate of eHealthArchitect on projects and programmes for national eHealth solutions, including health sector messaging over AMQP.
-
2009 — 2019
Consultant, Developer and Solution Architect
Visma RetailTen years from support to architecture. Store Solution Manager for a Norwegian chain of about 1,000 stores, then Solution Management and Professional Services, where I became a solution architect. Built everything from internal tools to integrations and larger end-to-end solutions.
-
2008 — 2009
Service Technician
Ricoh NorwayField service on multifunction and production printers, and training in Ricoh's solutions for printing, scanning and OCR.
-
2007
Artillery and Communications
Royal Norwegian Navy, HNoMS Fridtjof NansenServed aboard the frigate HNoMS Fridtjof Nansen.
04Writing
Notes from the workshop
On kristoffer.dev I write about AI agents, developer setups and what I learn along the way. Usually with code.
-
A Claude Code Status Line That Tells Me When I'll Run Out
The status line as a small dashboard: weather from my own CLI, rate limits that know my pace, and a bill that is right.
-
CLI-first AI Agents: Why I Build Small Tools Instead of MCP Servers
Why I give AI agents small, isolated Go tools to work in, rather than MCP servers.
-
Building a Hardened Linux Environment for Your AI Agent
An autonomous AI assistant on its own home lab server — with isolation, network control and custom tooling.
-
Spark from NDC Oslo: Events as Internal APIs
A spark from NDC Oslo every integration architect should keep: treat your events like internal APIs.
05Expertise
Where technology meets business
AI and data
- Generative AI, LLMs and agents
- RAG, tool use and MCP
- AI platform and operating model
- Evaluation and observability
Governance and value
- AI governance and risk
- EU AI Act and GDPR
- Benefit maps and measurement
- Change management and adoption
Architecture
- Integration and solution architecture
- APIs and event-driven systems
- Cloud: Azure and Cloudflare
- Retail: store, POS and chain
06Education
Education
Bachelor in IT, Administration and Management
University of South-Eastern Norway
College Engineer, with craft certificate
Horten Technical College
Contact
Let's talk.
About AI that has to run in production, integrations that have to fit together, or just a good tool. Drop me a line.
hei@kristoffer.dev