19-Year-Old Founder Raises $11M for Ghost, Launching $3,499 Personal AI Computer
Zain Javaid, who abandoned a career in quantitative finance after ChatGPT's release, has secured seed funding for Ghost, a startup building dedicated hardware for running AI agents locally on a person's device.
Zain Javaid spent years pursuing quantitative finance, relying on mathematics and data to inform investment strategies. That path lost its appeal. When ChatGPT launched in 2022, his perspective shifted dramatically. "I felt fundamentally excited in a way that I never had been before," he told TechCrunch. The experience sparked an obsession with personal AI that ultimately led him to found Ghost.
Ghost emerged from stealth this week with an $11 million seed round led by Andreessen Horowitz. Abstract, Audacious Ventures, SV Angel, and Nova also participated in the funding. Javaid's co-founders are Nicholas Chua, 19; Yifei Chen, 24; and Gautam Sharda, 23.
The company's flagship offering is Core, a purpose-built personal computer engineered to run AI agents that operate on behalf of users. Javaid's vision consolidates data scattered across desktops, applications, and smart home systems into a single device capable of running continuous AI agents that serve as assistants, execute tasks, and develop understanding of a person's life. "You can run models on existing hardware, but it's a terrible experience on many levels," Javaid said, who serves as Core's CEO.
Hardware and pricing

Core houses an Nvidia RTX Pro 4000 SFF Blackwell GPU, the type of processor typically deployed for running AI models. The GPU comes integrated into the device's price, eliminating the need for separate purchase. The machine also features proprietary software, a custom browser, and a filing system designed to process information and execute tasks without requiring repeated user input. The device carries a $3,499 price tag. Preorders for the initial batch launched Monday, with deliveries expected in late October.
Design and functionality
Javaid describes Core as a "brain in a box" without a display. Users interact with it through smartphone applications or dedicated apps. Voice interaction is supported, though users must speak through the Core app rather than directly to the hardware. Three models arrive preinstalled: Qwen-3.8-Next, Qwen-3.8-27B, and Gemma-4-31B.
Users retain the ability to install additional models from Hugging Face, where the AI community shares model repositories, or deploy their own custom-trained models. When asked how Core distinguishes itself from Apple's Mac Mini—a compact desktop computer some users configure for AI workloads—Javaid emphasized that Core's architecture combines hardware and software optimized specifically for running AI agents, with models already available, whereas the Mac Mini functions as a general-purpose machine requiring AI configuration.
Ghost plans to distribute over-the-air updates as new models and improvements emerge, drawing comparison to Tesla's update model. When newer model generations become available, Ghost notifies users and asks whether they wish to upgrade. Users can enable automatic updates to maintain the latest model versions continuously.
Privacy and local processing
Ghost manufactures Core units at its San Francisco facility. Javaid emphasized the company's commitment to delivering a fully functional computer that demands minimal setup or configuration. All models and personal memory operate locally on the device. Personal information undergoes encryption, preventing cloud providers and third parties—including Ghost itself—from accessing it meaningfully. Users can calibrate the autonomy level granted to their Core agents.
The device is engineered to continue functioning even if Ghost ceases operations. "All of the source code, logic, and model weights live on the device, not the services," Javaid explained. "We build this device to not be dependent on having an outside entity maintaining the service."
Ghost is developing a firewall mechanism to constrain how agents utilize credentials such as usernames and passwords. The firewall "monitors every outgoing network request on the computer and if it sees the agent performing a sanctioned action or exfiltrating data it shouldn't, it rejects the request and flags it for user review," according to Javaid.
"The user owns all of it," Javaid stated regarding the device, memory, and data. "Once you buy the hardware, everything is stored, run, and processed on the device."
Competitive positioning
Javaid downplays competitive threats from established technology companies. "People sometimes give incumbents credit for having already built the product because they have the ingredients," he noted. While acknowledging Apple's strength in hardware and track record of disruption, he contended that Core's differentiation centers on software capabilities: "The model reasons over your life, how it remembers, how it takes action, how it knows when to reach out."
Javaid illustrated Core's intended operation with a scenario: "Core proactively sends you a notification that you might be getting sick because it sees your resting heart rate is elevated, sees from computer history that you were up late working on a project, and has seasonal context that it is flu season in San Francisco."
The company's primary objective remains building a product users genuinely value. "If we can't do that, Apple isn't the reason we failed," Javaid said.


