Backing applied AI, one hard domain at a time.
Talon Capital Partners backs founders taking AI deep into a particular domain — its data, its workflow, its edge cases — until it genuinely changes how the work gets done. That runs from robotics and autonomy, defense, and advanced manufacturing to the vertical, unglamorous problems most investors overlook.
The thesis
The next breakthrough isn't a bigger model. It's AI taken deep into one domain until it actually works.
For twenty years the frontier of AI lived on a screen — text, images, predictions. The models are finally good enough that the hard, valuable work is applying them: wiring a model into a real domain, with its messy data, edge cases, and stakes, until it changes how the job gets done.
We call this Applied AI, and it shows up wherever real work happens — clinics, supply lines, trading desks, the factory floor, the field. Some of it is physical: robots and autonomous systems, defense platforms, the plants remaking how things are built. Much of it isn't. What we look for is depth in a domain, not a demo.
Talon backs the founders early to this, technical to the core, and building the unglamorous infrastructure of an AI-native economy. We move with conviction, write real first checks, and stay close.
Where we invest
Domain-Specific AI
AI taken deep into a single domain — its data, workflow, and edge cases — until it changes how the work actually gets done.
Robotics & Autonomy
General-purpose robots and the manipulation, mobility, and control stacks that make them useful in the real world.
Defense & Autonomous Systems
Drones, uncrewed platforms, and the software giving democracies a durable technological edge.
Advanced Manufacturing
AI rethinking how physical things get made — materials and metals, process control, and the modern factory floor.
Embodied AI & Control
Foundation models for action — policies, world models, and sim-to-real pipelines that turn perception into competent behavior.
Robotics Training Data
Parsing, verifying, and selling the demonstration and sensor data that teaches robots to act — the fuel of embodied AI.
The partners
Shwetark Patel
Shwetark is an olympiad winner who studied at MIT before building a crypto trading firm from the ground up to many billions in trading volume.
He started Talon to back the next generation of founders bringing AI into the physical world. Having built a technical company from scratch himself, he knows what the early days ask of a team — so he invests early and works to help every company he backs.
If you're taking AI deep into a real domain, we want to see it.
Pre-seed and seed founders across applied AI — send us what you're building. We read everything, and we move quickly.
hello@taloncapitalpartners.com →