YC W25 | Government AI | San Francisco
City planning, finally automated.
Two Stanford engineers who looked at city governments spending $50 billion a year on planning studies that take up to four years to complete - and built the AI to do it in months at 30% of the cost.
"Spend less time on routine tasks and more time on the meaningful work only planners can do."
- Waypoint Transit
The Problem
American cities are drowning in paperwork. Every new bike lane, every safety audit, every bus route adjustment requires a study. Those studies get handed to consultants who spend months doing what amounts to data entry: pulling crash records, overlaying GIS maps, writing boilerplate recommendations that look identical from city to city.
A typical transportation planning study takes up to four years. By the time it's done, the world has moved on. The road it was studying has been repaved twice. The consultant has billed accordingly.
The core issue isn't expertise - it's repetition. Skilled urban planners are burning hours on tasks that should take seconds. Literature reviews. Data formatting. Report templates. The work that requires a human brain - judgment, creativity, community listening - gets squeezed into whatever time's left.
Waypoint Transit looked at that dynamic and decided the repetitive part should be automated, so the human part can actually matter.
The Founders
Varun Tandon and Ryan Johnston met at Stanford. One studied computer science and went deep on AI at Microsoft. The other studied electrical engineering and went to Apple - then came back home to northern Minnesota and started building transit technology for his own city. That's how Waypoint Transit was born.
Varun holds a B.S. and M.S. in Computer Science from Stanford, where his focus was AI. Before Waypoint, he led applied machine learning work at Microsoft - developing diffusion models and LLMs for Copilot, Designer, and PowerPoint.
Ryan holds a B.S. and M.S. in Electrical Engineering from Stanford. After graduating, he worked on chip design synthesis CAD at Apple. But the pull of home - and a problem he could see clearly - brought him back to build real-time transit signage for Duluth, MN. That project became Waypoint.
The Product
Waypoint's core product is the Waypoint Assistant - a unified AI platform that plugs into a city's existing data sources and generates complete infrastructure planning studies through plain language commands. Planners describe what they need. Waypoint does the analysis, builds the maps, writes the report.
The system ingests crash reports, GIS layers, satellite imagery, transit feed data, and municipal code - then produces formatted memos, safety recommendations, compliance analyses, and interactive maps. Work that previously required months of consultant time now takes days or hours.
Data-driven analysis for transportation corridors. Pulls crash records, traffic patterns, and multimodal usage to produce comprehensive planning reports.
AI-driven risk identification with evidence-based recommendations. Covers roads, bike lanes, pedestrian crossings, and school routes.
Automated analysis of new development proposals against municipal codes. Generates formatted compliance memos for permitting decisions.
Complete streets analysis supporting Vision Zero goals and ADA compliance reviews. Built around transportation equity principles.
Impact analysis for transportation investments. Connects infrastructure decisions to broader economic outcomes for stakeholder reporting.
Connects to existing city data: crash reports, GIS layers, transit feeds, satellite imagery. One platform to unify fragmented municipal data sources.
Customers
Waypoint isn't a prototype. It's in production. The company works with municipalities, transit agencies, regional planning bodies, and departments of transportation across the United States. From small mountain counties to Marin County's commuter corridors, the platform handles real planning decisions for real cities.
The Opportunity
Every American city has a planning department. Every planning department outsources most of its analytical work to consultants. Those consultants are skilled, expensive, and doing a lot of repetitive data work that software can handle.
The $50B annual US municipal planning market moves slowly partly because there's no dominant software platform. Cities mostly still run on PDFs, spreadsheets, and phone calls. Waypoint is the first AI-native platform built specifically for this workflow.
The company targets municipal governments, transit agencies, and regional planning bodies - the buyers who control infrastructure budgets and approve projects. Getting into city hall is hard. Waypoint is already there.
Under the Hood
Waypoint's platform does something genuinely technical: it uses computer vision to analyze satellite imagery of urban infrastructure at scale. The team trained models to identify curbs, crosswalks, bike lanes, and pedestrian infrastructure from aerial images - with high accuracy from datasets of as few as 400 training images.
This means a city planner can upload a satellite image and ask "where do we need ADA ramps?" and get an answer based on actual visual analysis of the streets - not self-reported data or outdated surveys. That's a capability that didn't exist in this form for municipal planners before.
The system integrates with Roboflow for cloud-based computer vision processing, handling thousands of satellite images without requiring cities to maintain their own machine learning infrastructure.
"Roboflow is incredibly useful for rapid development of vision models."
- Ryan Johnston, Co-founder & CTO, Waypoint TransitThe broader platform connects to GIS layers, crash report databases, transit feeds, and municipal code libraries. Planners interact with it using plain English - describing what analysis they need. The system does the retrieval, synthesis, and report generation. The planner reviews, edits, and publishes.
It's not replacing the planner. It's replacing the 80% of their day that shouldn't require a planning degree.
Track Record
Funding
A small round with serious names. Y Combinator doesn't invest in ideas - they invest in founders and early traction.
For the record