Fello is a desktop app that sits on top of the software you already run. It installs in minutes, answers the how-do-I tickets out loud, and shows where things are, so your people do the click and learn it.
A big slice of every new-hire cohort's tickets is the same handful of "how do I" and "where is" questions. Fello answers those on the spot, and because your new hire does every step, the same ticket doesn't come back next week.
Fello opens the right page and walks the setup step by step, pointing at each field and waiting for each click. Your new hire does it; you never see the ticket.
It hands over the right link and guides from there. Fello can only ever open links you saved, so it can't send anyone to a made-up URL.
Teach Fello your access-request flow once by doing it, and every new joiner gets the guided version: the form filled in the right order, first try.
The eternal one. Fello points at the actual System Settings panes on your new hire's own screen, native apps included, not just the browser.
Nothing gets embedded in your tools and there's no per-app configuration. Fello reads the screen your people already see, and that's the entire integration.
Browser tabs and native apps alike: the HRIS in Chrome, the VPN client, System Settings. One guide can cross all of them.
A signed, notarized desktop app you distribute like any other. Company knowledge is a plain file you can read, diff, and version.
Plain answers about how the product works today. Your reviewer is welcome to verify every one.
Company knowledge, saved workflows and personal profiles live in plain files on your own machines. There's no Fello account and no Fello backend holding a copy.
Model calls route through Fello's proxy. Fello's AI provider keys never ship in the app. The optional bring-your-own-key live-voice mode is exactly that: you supply your own key and it stays in the macOS Keychain. The mode ships off by default.
While someone teaches Fello a task, it detects typed passwords and discards them. They aren't stored, and a guide never reads them aloud.
While Fello guides, pointing is all it can do. Acting on your behalf is a separate capability that ships switched off, and a guiding session can't escalate into it. The code itself enforces the boundary.
Microphone for push-to-talk, screen to see what you see, accessibility to find the real button it points at. It only looks when you ask; between questions it sits idle.
Each question goes through Fello's proxy to the AI model with what it needs to answer: a screenshot of the screen, the recent conversation, the company profile, summaries of your documents, and the employee's own profile and memory. A document is sent in full once, when it's added, to be summarized. Voice goes to speech-to-text only while push-to-talk is held.
Thirty minutes with the founder. Bring the security questions, the reviewer, and even the printer ticket.