24h Inside the $11M Startup Building a Personal AI Computer
Will Phillips · 22:05 · transcribed 1h ago
TL;DR Ghost is a startup building a personal AI computer that keeps all user data local for privacy. The company has raised $11 million and is preparing to manufacture thousands of units.
Ghost prioritizes local data processing to ensure user privacy
- The device processes information inside the home instead of sending it to cloud servers.
- It accesses apps, cameras, wearables, health data, and financial records locally.
- Founder Zein believes users must own their data and not share it with large tech companies.
- He argues that technology should serve people responsibly rather than just making money.
The hardware relies on expensive and scarce GPU chips
- The team holds $360,000 worth of GPUs in their office alone.
- They placed an additional order for $3.4 million worth of GPUs.
- Each chip costs $3,000 due to high global demand and limited supply.
- They use small System-on-Flex (SFF) chips to create a compact and efficient design.
Production involves manufacturing acrylic and stainless steel components
- The factory uses CNC routers to cut clear acrylic sheets for the casing.
- A laser cutter shapes stainless steel panels with high precision.
- The team assembles motherboards, RAM, SSDs, and GPUs into the final product.
- They are scaling up production to fulfill pre-orders within four weeks.
The AI provides real-time assistance through camera integration
- Cameras stream video 24/7 to help the AI understand the physical environment.
- The system can locate items like AirPods by analyzing live video feeds.
- It acts as a health coach by monitoring diet and sending notifications via Whoop data.
- Users can ask questions about their surroundings and receive immediate visual answers.
Next steps
- Launch the product and deliver thousands of units to customers.
- Host factory tours and showcases to build trust with early adopters.
- Execute a growth strategy using social media and PR channels.
- Maintain momentum after the launch with continuous content creation.