What it does today. What it's built to become.
Axom is already a research partner, a coding assistant, and a memory that's yours. It's also the start of something bigger: knowledge people own, and intelligence that goes where the cloud can't.
Working today. On your machine.
Everything in this section runs in Axom now.
A research partner
Ask about anything current. Axom searches, reads each source whole, and answers with who said what. Every page is dated, so old news is flagged as old.
A coding partner
Point it at a project and it reads, edits, and runs your tests, asking before it writes. Handed a failing test, it found and fixed the bug on its own in 92 seconds.
An assistant that knows you
Your projects, preferences, and people, remembered across every session. It all lives on your machine, where you review what it believes about you.
Work while you're away
Hand off a goal and Axom plans it, works it, and reports back. Ask it to keep watch ("tell me when this changes") and it decides how often to look.
One brain, every device
Run the model on your desktop and use Axom from a laptop that couldn't. It's sealed end to end, and your memory stays on the laptop.
A voice that fits
Terse or thorough, dry or warm, bullets or prose. Set it yourself, or let Axom learn how you write and suggest it.
The expert you need, grown from the work you do.
Axom starts as a generalist. Put it to work in your field and it becomes a specialist in it, learning from its own mistakes, on your machine.
Monday it guesses. Next month it knows.
Turn on a SysAdmin leaf and hand Axom your servers. At first it slips: a unit it didn't filter by, a log it misread, a restart it said it ran. Each slip is caught from the record, replayed at idle until the fix passes the same check, and kept. At 750 verified fixes the leaf is pressed, and Axom stops making those mistakes.
Accounting, security, your codebase, your lab's instruments: any work you do often enough becomes a leaf. How skill leaves grow
A mistake · a verified fix · a pile that becomes a leaf
Knowledge you own. Knowledge you can share.
Big AI trains on your data and keeps what it learns. Axom turns it around: a leaf you grew is a file you own, the skill without your data in it. Keep it private, give it to a friend or your team, sell it on the Leaf Store, or open-source it so everyone doing the same work starts where you finished.
The architecture was built for this. We're building it now.
Everything below is on the roadmap, not shipping yet. It's where chassis, leaves, memory, and the network lead.
Your lab. Your network.
Run your own relay and your own nodes on hardware you own. Every machine in the lab reaches the big models; nothing crosses your walls. The same sealed channel and node proofs as the Axom Network, with you holding every key.
Specialists that fit where the signal doesn't.
X1-Nano is a 2B model that runs in 5 GB. Snap on a specialty leaf for field medicine, equipment repair, or a local language, and carry an expert to a ship, a site, or a disaster zone, fully offline.
Remembered today. Simply known tomorrow.
The notes you accept about yourself graduate from memory into a personal leaf while Axom sleeps. What it once had to look up, it just knows, and you can see how much of you lives in the weights versus the memory.
Sealed compute
Hospitals, law firms, and labs running on Verified nodes: confidential hardware where not even the node's operator can read a prompt.
Know-how that stays
A team's hard-won judgment pressed into a shared leaf. A new hire starts with it on day one. The knowledge travels; the data doesn't.
Lend your machine
A spare GPU becomes part of the network in one command. Outbound only, its class proven by its own hardware.