Helsinki · teaching · AI · automation
I turn AI into
real tools.
I'm a special education teacher who codes. I use AI every day for course production, assessment design, Wilma and itslearning routines, and personal agents that do actual background work.
Hermes agent · in production
Morning briefs, bookings and other routines ✓
15+ yrs
of vocational teaching
27
open AI lessons published
19,000+
teacher community admin
EDUFI
national AI working group member
Profile
A special ed teacher who got tired of doing the same thing by hand.

I teach ICT at Helsinki Business College: programming, cybersecurity, IT support and a lot of practical student work. I have an MSc in Computer Science and over 15 years in vocational teaching.
AI is not a keynote topic for me. It's a work tool. I've built my own Hermes agent, automated Wilma and itslearning routines, published an open AI course, and used agent pipelines to produce course materials, assignments, rubrics and teacher guides. I wrote a blog series about building my own agent.
Qualifications: vocational teacher (Haaga-Helia, 2013) · vocational special education teacher (Haaga-Helia, 2018).
Also: OAJ council deputy member · shop steward · admin of a 19,000+ teacher community · member of the Finnish National Agency for Education AI working group.
AI & agents
Not demos. Working tools.
I build agents that retrieve information, use tools and leave an audit trail. If it only works on stage, it is not done.
Automation
Less pointless clicking
Wilma, itslearning, booking systems and other login-protected routines. The human decides; the script handles the mechanical work.
Pedagogy
The course still has to teach
Bloom's taxonomy, constructive alignment, accessibility and differentiation — built into assignments, assessment and student instructions.
Training & pricing
AI training that leaves you with tools — not a slide deck.
Content is tailored to your audience: concrete tools, real examples, and skills that stay in use after the training. Start with a light introduction or book a full agent sprint.
Want to try it yourself first? AI Fundamentals is my open, free 27-lesson course (in Finnish). In trainings, the same content is tailored to your work.
What does a talk look like? How I use AI at work and at home is a browsable presentation (in Finnish) of my own tools, learning materials and automations.
Starter
2-hour introduction
A compact workshop: what AI can do, how to benefit in your own role, and what the risks are. For teams at the start of their AI journey.
- Remote or on-site in Helsinki
- Hands-on exercises with your own tools
- Content tailored to your needs
Most popular
PopularHalf-day workshop
A deeper dive: prompting, tool comparison, agent structure and a concrete plan for what your team can adopt the very next day.
- Industry-tailored example tour
- Materials and exercises to keep
- Recording by agreement
Delivery
1–2 day agent sprint
We design and build your first AI agent together: current-state mapping, tool selection, data sources and a working pilot.
- Needs assessment & architecture
- Working pilot in your environment
- Documentation & next steps
General terms
- Minimum billing 2 hours
- Includes content tailoring
- Remote or on-site by agreement
- Travel outside Helsinki billed separately
Custom program
Multi-day training program
For example an AI academy for leadership or an expert team. Designed together around your needs, schedule and budget.
Privacy: all work is privacy-first — minimal data, in your environment, and without unnecessary transfer of student data to external services.
Projects
These are not concept slides.
I built these for my own work or real use. That means the boring parts matter too: logs, errors, maintenance and limits.
Personal AI agent
2025–Hermes agent
My everyday agent: morning briefs, monitoring, bookings and markdown memory. No mysticism — just many small tools behind one interface.
Read the seriesTraining · 27 lessons
2025–AI Fundamentals course
An open course where AI is learned by doing. 27 lessons, student and teacher materials, and a site generator. The material was made with AI, and I don't hide that.
Read casePedagogical analysis
2025Moodle course audit
A tool that reads a Moodle backup and shows where the course leaks: structure, Bloom levels, curriculum coverage, duplicates and fix effort. A report, not a prose exercise.
Read caseSelenium · teaching
2024–Wilma & itslearning automations
Scripts for the moments when the system makes you click the same thing for 30 students: study plans, completion lists, absences and submissions.
Read caseVideos
Watch how the tools work.
Short animated videos with voiced characters, in Finnish. Each chapter can be watched on its own, and the videos work without sound thanks to subtitles.
Video · in Finnish · 8 min13 chapters
Pedagoginen agenttiputki
The pedagogical agent pipeline
How learning material moves through an AI agent pipeline: one agent plans, one builds, three reviewers test and two gates decide. The teacher approves at the end.
Watch the video
Video · in Finnish · 6 min7 chapters
Opettajan apurit
Small helpers for Wilma and itslearning
Small tools built with AI that make a teacher's work in Wilma and itslearning easier, and how you can build one yourself.
Watch the video
Video · in Finnish · 7 min6 chapters
Esteiden raivaajat
Clearing obstacles in special needs support
A team of AI helpers that differentiates assignments both ways, lets advanced students move faster, builds a student's own weekly schedule and an individual skills demonstration project, without lowering the goals.
Watch the videoBlog
Honest production experience — without the hype.
Series and standalone posts on agents, automation and pedagogy. Latest series: Building My Own AI Agent, 6 parts.
Standalone · PedagogyHow I Audited My Own Moodle Course with One CommandJuly 10, 2026 · 3 minFAQ
Answers before you ask.
The most common questions about training, agents and how collaboration works. If yours is missing, get in touch — I'm happy to answer.
Contact
An AI project that needs to work in production?
I take on training, consulting and delivery requests — especially when the solution needs to fit real work and stay in daily use, not just be a demo.
