ZalenoAn AI executive assistant that reasons across your whole dev stack.
01
Where the idea came from
Running my own projects, I kept feeling the same friction: the context I needed to make a decision was scattered across GitHub, a calendar, Notion, and deployment dashboards, and stitching it together was a tax I paid every single morning. I did not want another dashboard to check; I wanted something I could just ask. Zaleno is that idea, an assistant that treats all of those tools as one surface it can reason over, so the answer comes to me instead of me hunting for it.
02
How it works
You sign in with GitHub, connect whatever else you use, and ask something like what should I focus on today. A reasoning loop gives the model a live set of tools drawn from your connections, it decides which ones to call and calls them in parallel, the results stream back into its context, and it repeats until it can answer, then streams a tight markdown briefing that cites the actual repo, PR number, event time, or deploy state. Every claim is grounded in a real tool result, never a guess, and it is read-only, so it briefs you but never changes your work.
03
Private by construction
Zaleno is multi-tenant with no database. GitHub and Notion connect through one-click OAuth so there are no keys to find or paste, and the tokens are sealed in an AES-256-GCM encrypted, httpOnly cookie that only the user's own browser can read. Anything you do paste, like a calendar URL or a Vercel token, lives in your browser's local storage and rides along only with your own requests. Credentials are used in memory to answer a request and are never logged or stored on the server, so there is nothing on the server for anyone to see, and no user can ever reach another user's projects.
04
The free-forever architecture
Every layer has a zero-dollar path. Reasoning runs on the Gemini free tier as primary with Groq running Llama 3.3 70B as an automatic fallback, both reached through one OpenAI-compatible interface so a rate limit on one silently rolls to the other. The shared free quota is rate-limited per visitor, and anyone can drop in their own free key for unlimited use on their own quota. GitHub, Notion, Vercel, and Hacker News supply data through their free APIs, and Vercel's hobby tier hosts the whole thing. It never bills anyone, including me.