19-Year-Old Founder Raises $11M for Ghost’s $3,499 Personal AI Computer

Ghost, a startup founded by 19-year-old Zain Javaid, has emerged from stealth with $11 million in seed funding and a $3,499 computer designed specifically to run personal AI agents.

Andreessen Horowitz led the round, with participation from Abstract, Audacious Ventures, SV Angel and Nova. Javaid’s founding team includes Nicholas Chua, 19, Yifei Chen, 24, and Gautam Sharda, 23.

Core is a computer built specifically for personal AI

Ghost’s first product, called Core, is designed to centralise information from a user’s computer, apps and connected devices so AI agents can continuously process that context and perform tasks on the user’s behalf.

The screenless computer includes an Nvidia RTX Pro 4000 SFF Blackwell GPU and has its own software layer, browser and file system. Users interact with Core through a phone or app, while voice commands are handled through the Core app rather than directly through the hardware.

Core ships with several AI models preinstalled, and users can add other compatible models from repositories such as Hugging Face or load models they have trained themselves. Ghost also plans over-the-air model updates, with users able to decide whether to install newer generations automatically.

Preorders for the first production batch open Monday, with shipping scheduled for the final week of October. Ghost is currently assembling the systems at its San Francisco facility.

Ghost is betting on local AI and persistent memory

Javaid describes Core as a “brain in a box” built to run continuously rather than waiting for individual prompts. The broader idea is to give personal AI access to enough context to understand a user’s routines and proactively surface useful information or take approved actions.

For example, Javaid said Core could combine information from health data, recent computer activity and seasonal context to warn a user that they may be getting sick. The goal is for the system to reason across personal information rather than treating each interaction as an isolated request.

Ghost says its AI models and personal memory run locally and that stored information is encrypted. Javaid said the architecture is designed so neither Ghost nor cloud providers can meaningfully access a user’s personal data. However, connected services may still be required when agents interact with external applications.

Users can control how much autonomy agents receive

Core is designed to let owners determine how much authority its agents have to take actions. Ghost is also developing a firewall that monitors outgoing network requests and can block activity that appears outside approved permissions or attempts to send data somewhere it should not.

The company says the hardware is intended to remain functional even if Ghost itself shuts down because the core software, model weights and logic reside on the device rather than depending entirely on a remote service.

Ghost is entering an increasingly competitive personal AI market

Existing computers such as Apple’s Mac Mini can already be configured to run local AI models. Still, Javaid argues that Core differs because both its hardware and software were designed around persistent AI agents from the beginning rather than adapted from a general-purpose computer.

That distinction will be central to Ghost’s pitch at $3,499. The startup is betting that users willing to invest in powerful local AI will value a system that arrives with models, memory, security controls and agent software already integrated rather than assembling those pieces themselves.

Javaid said Ghost’s immediate priority is making Core useful enough that people want to keep using it. For the young company, the larger test is whether personal AI becomes compelling enough to justify dedicated hardware alongside the phones and computers users already own.