OnDeck AI
Analyze any footage, without training a model
Y CombinatorSummer 2025ActiveB2BAnalyticsMachine LearningSaaSComputer VisionB2B
What they do
Problem
Building and deploying custom computer vision models from video footage is slow, data-hungry, and fails to generalize across diverse camera setups and environments.
Solution
OnDeck provides an enterprise infrastructure layer using vision-language models to search and detect objects, behaviors, or events in any footage without training a model or collecting training data.
Who it's for
Enterprise organizations that need to analyze and search video footage across varied cameras and workflows.
Founders
Alexander Dungate
Founder
Solving vision @ OnDeck. BSc in CS + biology, National Geographic Explorer.
Sepand Dyanatkar
Founder
Solving vision. Masters @ Cambridge in multi-agent RL. Research in swarm robotics for space, and built software at Amazon and the European Space Agency.
Open roles 2
Company timeline
- Founded2025
- YC batchSummer 2025
- StatusActive
- Team size5 people
- LocationVancouver, BC, Canada
Similar startups
Flowstate
Powerful AI-driven visual understanding for enterprise-scale video ma…
Tenyks
The World's Most Versatile Video Answering Engine
Conntour
AI to monitor thousands of security cameras.
Standard Cyborg
Buid, test and deploy perception applications
Roboflow
🖼️ Give your software the sense of sight.
Dragoneye
Custom video detection models in less than 5 minutes