Product · Salana Desktop
Talk to one Orb. Get the work of many agents.
The Mac app for your memories and your AI agents. Say what you need, the Orb sends agents to do it in parallel, and you read a short report instead of a long chat.
Fix the checkout bugRunning 42 tests
Add an FAQ pageReading the knowledge base
Last month's numbersRevenue up 9%. Ready to read.
Investor updateNeeds your yesA chat makes you the bottleneck. The Orb does not.
In a chat you ask, wait, read a long answer and ask again, one task at a time.
Tell the Orb everything at once. Each task gets its own agent, they work at the same time, and you read one line per task: done, here is the result, or this one needs you.
You wait, then read, then ask the next.
Fix the checkout bugDone
Add an FAQ pageDone
Last month's numbersDone
Investor updateNeeds youYou read four short reports.
Your AI agents, run from one conversation.
A request goes in, an agent takes it, and it comes back for your yes. You decide; you do not sit through the work.
A simulated demo of Salana. You ask for things, AI agents work on them, and each result waits for your approval. It approves itself after 5 seconds. Nothing is real: nothing is sent, and the phone number is fake.
Simulated, but realistic: a demo of the Orb as it works inside the app.
How the Orb works.
Claude, Codex, Gemini or Grok. Your own account.
Salana Desktop runs the AI accounts you already have, signed in by you. It does not include, resell or proxy any AI.
Switch model halfway through a chat and the next one carries on, because the context lives in your memory, not in the model.

From your notes in June: Saturday sends sell least, so the newsletter moved to Thursdays.
Inside Salana Desktop
The Orb is the front door. Behind it: chats and terminals in spaces, pages for your data, and skills for how you work.
Your memory, local or shared.
Set up a memory that is specifically yours.
Your notes and files, every model, yours forever.
