- AngelsRound
- Posts
- Proception ✌🏿
Proception ✌🏿
Robotic hands that learn from human hands
Spotlight
What if the hardest problem in robotics isn't walking, seeing, or reasoning, but grabbing a doorknob?
Quick Pitch: Proception is building human like robotic hands and wearable data gloves to help robots learn the dexterity needed for real world tasks.
The Problem
Manipulation Remains Hard: Robots can walk, perceive, and reason, but still struggle with everyday tasks requiring human like dexterity.
Training Data Is Scarce: Physical AI needs data showing how humans touch, grip, and manipulate objects.
Teleoperation Doesn't Scale: Robot based data collection is expensive and slow, producing only 5 to 10 demonstrations per hour.
Why It Matters
Humanoid robotics is moving from walking and reasoning toward useful physical work.
That work overwhelmingly requires hands. 99.5% of civilian jobs require fine hand manipulation, from handling tools to operating equipment.
As humanoids move into real world environments, manipulation could become one of the most important layers of the robotics stack.
Snapshot
Industry: Robotics / Physical AI
Headquarters: Mountain View, CA
Founded: 2024 (YC W25)
Backed By: First Round Capital, Y Combinator, BoxGroup
Funding: $11M (Seed); Raising ~$5M (Seed+)
Traction: Early commercial shipments to paying robotics customers
Target Customers: Robotics companies and data providers
Founder’s Edge
Playing Field
Hand Manufacturers: Sharpa, WUJI, and others offer dexterous hands but rely on customers for data and learning.
Humanoid OEMs: Can build manipulation in house, but must develop the hardware, sensing, control, and training stack themselves.
Proception's Edge: An independent manipulation platform combining hardware and training data, rather than selling standalone hands or building a vertically integrated humanoid.
Analysis
Bulls Case 📈
Strong founder market fit with former Tesla Optimus engineers.
Early traction with hardware shipping to paying customers.
ProGlove could create a data flywheel as adoption grows.
IP extends across hardware, sensing, and data collection.
Bears Case 📉
Humanoid OEMs may consider manipulation too strategic to outsource.
Dexterity, strength, and durability remain difficult to balance.
Chinese manufacturers could pressure hardware pricing.
ProGlove's data advantage still needs to translate into better robotic performance at scale.
Verdict
The largest humanoid companies may never outsource their hands. With billions invested, manipulation could be too strategic not to own.
The bigger opportunity may be the broader robotics market, where building the hardware, sensing, data, and control stack from scratch is harder to justify. Proception's bet is that manipulation becomes a specialized layer most robotics companies buy rather than build.
The Startup Pulse
Another happening week in startup funding. Three signals from this week:
Making AI cheaper and easier to deploy is becoming as valuable as building better models.
Healthcare and enterprise software remain some of AI's biggest commercial opportunities.
Investors are increasingly funding the infrastructure around AI, not just AI itself.
Commonwealth Fusion Systems — Secured $1B to scale commercial fusion, addressing AI's growing energy needs.
Multiverse Computing — Closed a Series C of up to $570M at a $1.7B valuation to make AI models smaller, faster, and cheaper.
Gravie — Brought total funding to $463M with a new General Atlantic led round to modernize employer healthcare with AI.
Your team’s new support player
Stax Payments, a Hubspot customer, now books 80% more meetings with the exact same team they had before.
No new hires. No new territories. Just HubSpot's Prospecting Agent finding the right accounts, surfacing the right contacts, and drafting outreach that actually gets responses.
Your best rep is great because they know who to go after and when. Prospecting Agent gives every rep that same starting point.
Strategic breakdowns of how the biggest AI companies actually work — their business models, their moats, their vulnerabilities. This week: why no one can beat Nvidia. And it has nothing to do with the chips.
Written by Ashher
