Skip to content
ProjectsCase study

Helping colleagues work with AI

How I help colleagues in technical and non-technical roles start working with AI: by starting from their problem, and matching the help to what they already know.

Role
AI Enablement & Solutions Engineer
Period
2026 – present
Context
ACA Group, internal
Stack
  • Claude Code
  • OpenCode
  • n8n

Problem

AI can help almost every role, but not in the same way. A technical colleague needs a different start than someone in HR, sales or fleet, and a generic training gives them both the same answer. How I help depends on what the person already knows.

Approach

The problem first, then the tool

Whoever I help, I start by working out the problem they're facing, before anyone opens an AI tool.

Technical colleagues: coaching, so they carry on alone

I look for solutions with the tools they already use. One technical colleague had little to no experience with AI. After a day working side by side, he was using OpenCode, writing his own skills and automating repetitive tasks in his daily work.

Non-technical colleagues: building it together, step by step

For colleagues in HR, sales, fleet and other roles, I sit down with them until the problem is clear. Then I build automations with AI tools and n8n in small iterations, with short feedback loops, until the result is what they need.

Advice beyond the team

I follow developments in AI and automation and advise management, security and infrastructure on what they mean for the organisation: new tools, risks and the choices they bring.

Teaching outside the company

From December 2026, I'm one of the trainers in an AI course for business owners and managers, run by Voka and ACA Group. I guide the technical side: setting up each participant's AI workspace, the build plan for their own case, and a build day where they make a working prototype without programming.

Result

Colleagues who had hardly used AI now use it in their own work. The technical colleague above automates his own tasks and writes his own skills. For non-technical colleagues, repetitive work now runs as automations that we shaped together, one iteration at a time.

More case studies

Built something similar? I'd like to compare notes.

Get in touch