Five products, all live.
Five products shipped in 2026: a wellness app that drew investor interest, a transit platform a city forwarded to its operator, an enterprise email platform, a live shop taking payments, and the private assistant behind all of it.
I designed every flow and token, then built each one with a team of agents. This concept was built the same way.
My own live products, built outside this concept. Not work for Fin.
Native consumer app · 0→1 · 2026
ugoki
A wellness app pairing fasting with HIIT, coached by a governed AI. A cold outreach brought London investor interest.
The case study →
Transit operations · 0→1 · 2026
vtv
A transit platform for Riga’s buses, pitched cold to City Hall. Built on one rule: the agent advises, humans decide.
The case study →
Enterprise platform · 0→1 · 2026
email innovation hub
Nine AI agents build production email behind a ten-point quality gate. The hardest consistency problem I know.
The case study →
Commerce · 0→1 · 2026
gerboni
A live shop selling Latvian city heraldry, taking Stripe payments. The support agent acts on orders rather than only apologising.
The case study →
Personal ops · 2026
fredis
The assistant that runs my working day: 24 skills, persistent memory, and agents that draft everything and send nothing.
The case study →
d-050 · An eighth page shows the five products behind the method
- Decision
- An eighth page: five live products, thumbnail and link each.
- Because
- The concept argues a method; five shipped products show it working.
- Rejected
- A block on the proposal memo instead of a page
- Would measure
- Whether a reviewer opens a case study.