At 19, Zain Javaid was not planning to build an AI hardware company. For much of his life, he had wanted to become a quant, using mathematics and data to make investment decisions. He eventually became one, but the work did not hold his attention for long.Then ChatGPT arrived in 2022. Javaid told TechCrunch that he felt “fundamentally excited” in a way he never had before. He became interested in personal AI, particularly the idea of having an AI system that could understand a person’s digital life and actually do things on their behalf.That interest has since taken him into the world of startups. Javaid’s company Ghost has emerged from stealth with $11 million in seed funding led by Andreessen Horowitz. Its first product, Core, is a $3,499 computer built specifically to run personal AI agents.
What does a $3,499 AI computer actually do?
Core is not designed like a conventional desktop computer. There is no screen. Instead, users interact with it through a phone or an app, with voice mode also available through the Core app.Javaid describes it as a “brain in a box”.The idea is to have one device that can bring together information from a person’s computer, apps, wearables and smart home devices. AI agents running on Core can then use that information to perform tasks, rather than waiting for the user to give them instructions every time.The machine includes an Nvidia RTX Pro 4000 SFF Blackwell GPU, a chip designed for running AI models. Three models come pre-installed, while users can also install other models from Hugging Face or load models they have trained themselves.The first batch is available for order at $3,499, with shipping expected in the last week of October.
Why a 19-year-old thinks personal AI needs its own computer
Javaid’s argument is that existing computers were not really designed for AI agents.A Mac Mini, for example, can be configured to run AI tools, but it remains a general-purpose computer. Core, he says, has been designed from the beginning around AI, with the models, software and hardware working together.The company also wants the device to keep improving as AI models change. Javaid said Ghost can send updates to Core over the air, allowing users to move to newer models. They can choose to make those changes themselves or switch on automatic updates.But the more interesting part of his pitch is what happens when the AI starts connecting different pieces of information.Javaid gave an example of Core noticing that a user’s resting heart rate has increased, that they were awake late working on a project and that flu season had arrived in San Francisco. Instead of waiting for a question, the system could alert the person that they might be getting sick.
Your data stays on the device, Ghost says
Because Core is intended to know a great deal about its owner, privacy is a central part of the company’s pitch.Javaid said personal information, memory and AI processing can remain on the device. He also said the source code, logic and model weights live on Core rather than being dependent on Ghost’s services.The company is developing a firewall to monitor what an AI agent does with credentials such as logins and passwords. According to Javaid, the system can block a network request if it detects an action that should not be allowed and flag it for the user.Ghost’s founding team includes Nicholas Chua, 19, Yifei Chen, 24, and Gautam Sharda, 23. Together, they are betting that the next personal computer may not look much like the computers people use today.For Javaid, the bigger question is what happens when AI stops being something people open and start using it as something that is simply there, remembering, reasoning and occasionally acting on their behalf.At 19, he has already convinced investors to put $11 million behind that idea.