PLATFORM // HOW IT FITS TOGETHER
A small OS for your AI life.
Leorna sits under the AI tools you already use and gives them what they're missing: a shared memory, helping hands, and a safe place to work.
THE PLATFORM // ASSEMBLING
Four layers, one machine. Keep scrolling — it builds as you go.
01 / MEMORY
One shared memory
Everything you tell one AI tool is there for the next one. Projects, preferences, decisions — said once, known everywhere.
02 / AGENTS
Niko & the Sprites
Niko is the one you talk to. The Sprites are the small agents Niko sends to do the work — each one focused on a single task, each one sandboxed.
03 / FLEET
Every machine you own
Desktop, browser, terminal, phone — the same memory and the same Sprites, wherever you are. Start at your desk, finish from the couch.
04 / SAFETY
Sandboxed by default
Nothing runs without your permission. Every action is logged, so you can see exactly what happened — and undo it.
ONE SYSTEM
Ask once. The whole platform answers.
Request in, work out, receipt back, memory updated — one loop, on machines you own. Click the diorama to run it again.
THE STACK
One request, acted out: a tool window drops it to Niko, who recalls context from Cortex — the fleet's one memory, built on MemoryOS — and splits the job between the Sprites. They hop down to your machines and work, their lessons fly back into Cortex, and an amber receipt returns to the tool that asked.
LEORNA the OS in the middle — under your AI tools, above your machines; one memory and one fleet shared by every tool
NIKO the orchestrator — plan → dispatch → handoff
SPRITES sandboxed workers — scoped grants, full logs
CORTEX the fleet's one memory — MemoryOS substrate: durable, with provenance + scope
FAQ // STRAIGHT ANSWERS
▸ What does “a fleet of agents” actually mean?
Instead of one AI in one window, you get many small agents running at the same time across your machines. One handles a build, another writes notes, another scans your backlog — all at once, all sharing what they learn.
▸ How do the agents share memory?
Through the Hive — one shared memory every agent writes to and reads from. Fix or learn something once, and the whole fleet has it. You decide what stays private and what goes in the Hive.
▸ Do I have to set it up separately for each AI tool?
No. You connect Leorna once and it shows up in all of them — Claude, Cursor, Gemini and more. Same memory, same fleet, no per-tool wiring.
▸ Can I see what the agents are doing and where my data is?
Yes — that's the whole point. You watch every agent live, see what each one knows, and your data stays on your machines. Every action is logged, so you can check it and undo it.
▸ Do I need to be technical?
No. You chat with Niko in plain words and the fleet handles the rest. When you want to go deeper, every piece has a programmable surface — see the Developers page.
▸ What does it cost?
Early access is free, and early users get the best deal we'll ever offer. Pricing comes later, shaped with the people using it. Running a fleet on your own machines means you're not paying a markup on someone else's servers — keeping it efficient is part of the design.