How REQ Capital went from trying AI to building with it.
- Client
- REQ Capital
- Sector
- Investment management
- Where
- Oslo and Stockholm
- Format
- Ten weekly sessions, August to October 2026
- Status
- In delivery
The situation
REQ Capital is a partner-owned Nordic investment firm in listed equities and real estate, run by a small, senior team. Its edge is judgement built over decades, and its systems are the ones a firm of that size can carry: standard tools, spreadsheets, and a good deal of work done by hand.
The team had tried general AI tools and got what most firms get: a good demo, then nothing that stuck. The partners wanted something more ambitious than a course on prompting. They wanted the people who invest to be able to build for their own process, in their own language, and to know where the limits are.
What we did
We built a ten-week programme around how REQ actually works and delivered it inside the tool the team already uses. The course runs as a Claude plugin: the schedule, the pre-work and the recaps sit in the assistant, next to live connections to Norwegian public data, so Statistics Norway, the company register, the land registry and Norges Bank are one question away.
The first session was the only lecture. From then on the team built. Each person picked a task from their own week and turned it into a skill, a reusable instruction that Claude follows every time. They set up projects with persistent context for a company or a property, produced Word, PowerPoint and Excel documents straight from Claude, and by the fourth week were bringing their own initiatives to open work sessions and leaving with something that ran: portfolio trackers, internal websites, research tools built on their own material.
The later sessions go deeper. Writing in REQ's own template. Research with a verification checklist. Repeatable analysis in code. Building a small internal connector, and the rules for safe use of what the team has built. The course closes with the team presenting what they made and naming the champions who carry it on.
What changed
- From users to builders. Every participant ships working tools for their own workflow, reviewed by the people who use them.
- AI where the work happens. The course, the tools and the public data live inside the Claude the team opens each morning.
- The limits, found by building. The team now knows what it can build itself and which parts need engineers, because it has tried both.
- A practice that continues. Internal champions carry the capability forward when the programme ends.
What comes next
Building in the open showed the team where its own tools stop and engineering begins, and opened a conversation about how we keep working together once the programme ends.
Reporting is one example. The daily view of net asset value, holdings and investor flows is the kind of system a firm wants to own, and we are exploring with REQ what one built on its own data should look like.
Other projects
