I build AI systems with engineering teams.
Far Horizons is my consulting practice. Production LLM systems since 2022, for teams like Scout24 and AutoScout24. Before that, VR and AR at REA Group. Independent since 2019 — Luke Chadwick, Berlin.
FAR
speculative
FH//03
speculative
FH//03
MID
in development
FH//02
in development
FH//02
AI-native property portal Conversational search, agent-run busywork. Own money. Multi-agent dev tooling Several coding agents, one repo, an automated merge queue. Agentic Ops Self-healing CI/CD — the agent fixes the pipeline that broke it.
LLM_ADVENTURE — SCENE 1, GRADED LIVE ▶ PLAY ALL TEN SCENES
·Scene 1 — The Basics of Prompting.
·Archmage Elara: "Vague questions yield vague answers. Ask me about the weather — be precise about location and time."
â–¸
NEAR
in production
FH//01
in production
FH//01
Five shapes of work, scoped and priced before anything starts:
Architecture audit What your AI system will do under load, and what to fix first — in writing. 2 wks · fixedFractional CTO The founder-level calls, made by someone still writing code in your repo. 1–2 d/wkRescue The stalled project, the vendor that walked. Done it before, under the original deadline. starts < 1 wkTeam enablement A working session on your repo; engineers leave faster with coding agents. 2–5 daysAdvisory retainer A standing call, an open channel, a second opinion before you commit. monthly
RECENT WRITING
07·11·2025 First Impressions: An Evening with sintra.ai18·08·2024 Shipping LLM Pilots in Days, Not Quarters18·04·2024 Railscamp Tanks
162 posts, 2004–2026 →SHIPPED FOR
Scout24 · AutoScout24 · REA Group · LuxuryEscapes · Partmax · Zolidar
the full timeline →HERE
today
FH//00
today
FH//00
hello@farhorizons.io
Describe the system, the deadline, and what's in the way. Reply within a business day — if I'm not the right person, I'll say so.