Frame a need, form hypotheses and validate them by prototype before building.
AI agents, multi-LLM workflows, automations, APIs: concept to MVP autonomously, removing feasibility risk early — and a clean handover to dev teams when it needs to scale.
From the data lifecycle to the metrics that drive trade-offs — aim for the outcome, not the volume.
I can train teams — or teach in schools — in AI and rapid automation, so they stay autonomous. On a full-stack engineering base.
AI SOCIAL GOD: weekly scan of web & socials, sector-trend detection, on-strategy angles and drafted content — the client just sorts cards for 20 min. Live with a first client; first fully autonomous weekly cycle passed.
3 Blasons project: from hardware purchase to POS software, plus a WordPress site, all synced in full-stack dev — then rebuilt "for the exercise" in no-code.
FR↔Hungarian voice translator, ultra simple: built in 1h on Make + Telegram, rebuilt in n8n in 20 min.