Daniel Östling
Product leader who ships AI.
I turn fuzzy ideas into specs, and specs into shipped AI products. Comfortable in the code, but the job is product — strategy, build, and adoption.
- BACKGROUND
- 10+ years in B2B SaaS & e-commerce — IBM, Lime Technologies, CASAFARI, Moteefe
- SCALE
- Scaled Moteefe’s enterprise unit to €4M ARR as P&L owner; clients incl. H&M Group
- TEAM
- Led a 15-person data science team building ML on European real-estate data
- SHIPPED
- AI dev infrastructure at Direct Wines — 70% faster development, 3× fewer defects
- NOW
- FractalOcean (fractional AI product team), Karl, Monde — from Lisbon
FEATURED WORK
Karl
- PROBLEM
- Agents build the wrong thing when the input is fuzzy.
- OUTCOME
- <2 min per brief · 40+ hrs/month saved on PM documentation
- STATUS
- live
Monde
- PROBLEM
- TODO-CONTENT (proposed: brands have no visibility into how they appear in AI search answers)
- OUTCOME
- 100% visibility into AI answers
- STATUS
- live
Direct Wines — AI dev infrastructure
- PROBLEM
- TODO-CONTENT (confirm what is public — SPEC §7.2)
- OUTCOME
- 70% faster development · 3× fewer defects
- STATUS
- handed over
WHAT I HELP WITH
- AI product strategy
- Agentic workflow design
- RAG and internal knowledge systems
- Evals and quality loops
- Product discovery and spec writing
- Fractional product leadership
HOW I WORK
I work spec-first — write the spec, build in phases, review at every gate. It’s the same discipline whether the builder is a team or an AI agent, and it’s how this site was built.
TODO-CONTENT: paste the full quote verbatim from LinkedIn / the fractalocean.ai showcase (themes: zeroed in on highest-leverage opportunities; brought AI into the team in a way that stuck; insight to production in days).
Building Karl and Monde; fractional AI product work via FractalOcean. TODO-CONTENT: availability state.