There and back again... but by I-5 or the ferry?
summary
Built an agentic AI that queries WSDOT ferry schedules, Google Maps drive times, and ElasticSearch-indexed calendar events to compare routes, costs, and arrival buffers; uses FastMCP servers with MCP resources, tools, and prompts to estimate travel time, generate expense estimate (mileage × $0.70 plus parking/ferry), save preferred routes, and push notifications. The system picks among ferries, I‑5 drives, and hybrids, records chosen routes, and reliably returns ranked options.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:14Commuting dilemma on Olympic Peninsula
- 3:41Google Maps ferry failures
- 4:26ElasticSearch and APIs solution
- 7:00Travel planning system design
- 9:17Technical stack
- 11:16MCP server core components
- 13:18LLMs with MCP tools
- 16:34MCP in agentic AI
- 17:21WSDOT server code walkthrough
- 20:28Elastic agent integration
- 21:31Safe ElasticSearch writes
- 23:28Live demo travel tool
- 26:01Calculating event travel options
- 29:52Debugging improvements
- 31:12Future use cases
related talks
41mThere and back again... by ferry or I-5?
58mFrom here to there and back again
22mMicrodosing the Economy
28mBuilding Agentic AI applications on AWS
01h 30mHow to build AI Agents: A step by step guide to creating AI agents
23mKEYNOTE: From Copilot to Flight Captain
23mShip It First, Fix It Later: How a Medical Crisis Led to a Global Tool
25mArtificial intelligence, meet location intelligence