Investors

LCOA is a private AI assistant for Windows that runs on your own computer. It is a working beta from two co-founders, and it is early. This page is about the company: who we are, what exists, how it earns money, what is not proven yet and how to reach us. It is accurate as of November 6, 2026.

The company at a glance

• Name: LCOA (said “el-co-uh”). Website: lcoa.ai.
• What it does: a desktop AI assistant for Windows. Chat, voice, writing and code run with models on the user’s own PC, so conversations and files stay there.
• Stage: beta 2.5.4. The public launch has not happened yet. Downloads open at launch.
• Team: two co-founders, Ethan Ganz (CEO) and Arieh Kohn (CTO).
• Legal status: not yet a registered company.
• Slogan: Ideas are many, credits are none.

The problem

Most AI assistants send a person’s chats and files to someone else’s servers, and give little say over what they change. LCOA starts from the other end: the model runs on the user’s own computer, so there is no meter running on every idea.

The product

LCOA runs the model on the user’s own PC. The user decides how much it may touch, from watching only to running programs. It shows every change before making it, and most changes can be undone.

• Chat, writing and maths with local models, and web search when asked.
• Voice: speak and it speaks back, with listening and voices that run on the PC.
• Tidying files, a task list, reminders, and calendar and mail through Google if the user connects it. Mail is only ever saved as a draft.
• Memory the user can see, keep or ask it to forget.
• LCOA Code, a project agent that proposes each edit and waits for a yes.
• Connections to other programs through MCP, off until the user adds one.
• The Trixe family of local models. A starter model picked automatically for each PC is still being finished.

The roadmap lists what works today, what is being finished and what is only planned.

How it earns money

• Free to start. Plus at $10 a month or $99 a year, when checkout opens.
• Pro, Family and Developer plans are planned. They are not for sale, and nothing is promised. See pricing.
• There are no paying customers yet. LCOA has not charged anyone.

The team

• Ethan Ganz, co-founder and CEO. Builds the app and leads the product and the research.
• Arieh Kohn, co-founder and CTO. Wrote the original website copy and legal text.

More about us is on the About page.

The research track

Alongside the product, Ethan has been testing a way to make local models cheaper to run, called Dynamic Neural Assembly. It is research, not a shipped feature. On one laptop, the measured speed of local models matched a memory-bandwidth prediction to within 6 to 20 percent, and a mixture-of-experts model ran about 40 percent faster with half its experts active while getting 19 of 20 easy test questions right, against 20 of 20 with all of them. These are small tests on one machine. They are not a claim that LCOA is faster or better than other assistants, and it has not been compared with them. The notes and the experiment code are available on request.

What is not true yet

• LCOA is not yet a registered company. It is run by two co-founders: Ethan Ganz (CEO) and Arieh Kohn (CTO).
• The installer is not code-signed, so Windows shows a warning. Signing and a Microsoft Store listing are in progress.
• There is no usage, retention or revenue data, because the public launch has not happened.
• Mac support is a first release that has not been tried on many real Macs. Chromebook has a demo only, with no AI. Linux is planned.
• No claim is made of being faster or better than other assistants. The speed and quality of the shipped models have not been published yet.

Main risks

• Big companies could ship the same idea inside the operating system.
• Local models are less capable than the largest cloud models, and people may not accept that.
• Two people carry the product, the support and the research.
• Paid plans are untested. It is not known how many free users will pay.

What we are looking for

Conversations with angels and early-stage funds who back local AI, privacy-first software or developer tools. There is no round open and no amount set: the first step is forming a company, and that needs advice. Intros, advice and criticism are welcome too.

Documents

• Press kit: descriptions, facts, logo and screenshots.
• Roadmap, pricing and the Trixe models.
• Privacy policy and terms of service.
• Service status.

Contact

Ethan Ganz, ethan@lcoa.ai. Write and you will get a real reply.