Me in 10 seconds
- 2014-2017
- Sales. Cold and warm calling, online ad sales at Hitta.se, then sales manager with recruiting at a meeting-booking firm.
- 2016
- First company at 21: an e-commerce brand. Break-even in a year, closed to focus.
- 2017-2025
- Ran a quantitative promotions operation across the Swedish and Norwegian gambling markets: my own capital, a team of up to seven, standardized playbooks, monthly P&L.
- 2021-2023
- International business development at Clar, a digital loan broker: set up the Italian and Portuguese entities, launched clar.co, market-entry analyses for six countries, M&A research.
- 2025-
- Engineering Mathematics at KTH, and the systems below.
What I am building
oddslogik.se is a Swedish comparison publication for the licensed gambling market. I write and run it alone, with a set of AI agents doing the execution. Every agent reads the same doctrine file. Every material choice gets a numbered entry in a decision log, 365 so far. Work runs in up to four parallel sessions partitioned by file ownership, and nothing is promoted without verification. 84 pages shipped since June.
For a friend's services company I built a 4 230-row prospecting CRM with agents, then had a separate model instance audit it over four sessions. It found about forty defects. A dozen were errors in the builder's own rulings that had been accepted as correct, the class of error that is invisible in its own output.
TimeBox Stats is a Garmin watch app I wrote in Monkey C for focused-work sessions. It syncs through the phone to a Postgres database and a public dashboard, and it produces the record below. Code, architecture guide.
This site is run the same way: plain HTML, a doctrine file, and a public decision log that the agents and I both write to.
Focused hours, live
I log focused work from the watch. This is every day since 7 April 2026.
0, under 1, 1 to 3, 3 to 6, 6 or more hours
527 hours across 984 sessions on 121 active days; longest single session 8 h 05 min; this week 0.0 h so far. The September gap is real; logging restarted on 22 September.
Some things I believe
Read the rules more carefully than the people who wrote them.
Most systems pay the person who does.
A green build is not a deploy.
Verification is a separate step, done by someone who did not build.
Errors invisible in their own output need a second instance.
The builder cannot see them, and neither can I.
Intensity is not consistency.
A high minimum week beats a heroic day.
The record is the record.
An instrument that hides its gaps is decoration.
Governance is scar tissue.
Every rule I keep was paid for; keep the set small enough to obey.
Pre-commit the pivot.
The moment you need to change course is the moment you are least able to decide it.
Write it before you prompt it.
Otherwise you will not know which thoughts were yours.
How I work
- Decisions close with three lines: decision, rationale, next action.
- Every material choice gets a numbered log entry before the work starts.
- Plan first, halt, then write; layout changes are shown before they ship.
- Nothing is promoted without verification: a green build is not a deploy.
- Parallel work is partitioned by ownership so nothing collides.
- Redo it, do it right is a normal sentence, in both directions.
The offer
I want 20 focused hours a week inside a small AI-native team. The way I propose to start: one 8-12 week scoped build of an internal system you already know you need, at a day rate, with an option to continue at 50 percent if it works.