Shown·Hermes Agent·Travel agencies
Travel agent that learns from trips
Hermes Agent plans trips using stored traveler preferences, researches flights and hotels by web search, and rewrites its own memory after a post-trip debrief.
Resultrc_CAGYW6CZ
After a debrief on a 6-hour layover, the agent added a 3-hour layover cap and applied it unprompted when re-planning the next trip.
ShownBlog · May 31 · Last checked
Overview
How it was set up
Hermes Agent with two custom skills (trip planning, post-trip debrief), a context file, MEMORY.md and SOUL.md. Runs on gemini-3-flash-preview with built-in web search and memory.
Worth knowing
Traveler profiles and prior-trip data were seeded before the demo, and trip dates are fixture data, not live calendar events.
A prompt to start from
Plan a trip for [traveler] to [destination] [timeframe]. Apply all stored preferences, research flights and hotels, check visa validity, and end with a pending-approval summary.
Written by Agentic Feats from the source. Not the author's exact prompt.
Sources 1
History
- Source published
- 2026-05-31
- Added to Agentic Feats
- 2026-10-13
- Pick
- No
- Markdown copy
- /c/travel-agent-that-learns-from-trips-hermes-agent.md