MoboAgent chats like an assistant and works like an operator: it runs commands, manages files, replies on Telegram and WhatsApp and runs jobs on a schedule — on a local model you control.
Most assistants can only talk. MoboAgent can act — locally, with your permission, on hardware you own.
The problem
Chat assistants answer questions, but the repetitive work — replying to customers, checking messages, moving files, running routine commands — still lands on a person. And sending all of it to a cloud model isn’t always an option.
The solution
MoboAgent runs a model through Ollama on your own machine and gives it a toolbox: shell, files, messaging, browser, screen, memory and schedules. Paths are scoped, and dangerous commands wait for your confirmation.
An AI operator that runs on your machine.
Features
Runs on a local model through Ollama — Llama, Mistral or any model you pull. Prompts stay on your machine.
Runs shell commands, reads and writes files, lists folders, reports system info and opens apps and URLs.
Dangerous shell commands need your confirmation, and file and shell access stay inside the paths you allow.
Reads new Telegram messages and replies, sends WhatsApp messages through Twilio and can answer inbound messages.
Run any prompt on an interval or a cron expression — like checking messages every fifteen minutes.
Remembers facts per user or chat — “our hours are 9–5” — and uses them in later answers.
Switch between modes like Support or Sales. Tools and memory stay the same; only the tone changes.
Answer from your FAQ or a folder of documents with built-in document search.
Capture the screen, read visible text with OCR and click at coordinates.
Talk to it: speech-to-text in, spoken replies out.
A macOS menu bar app for quick prompts without opening a browser.
How it works
Run the installer. It checks Python, installs dependencies and points you to Ollama.
Download a model with Ollama — Llama 3.2, Mistral or whichever you prefer.
Chat in the browser, from the menu bar, or by voice.
Turn prompts into schedules and let it handle Telegram and WhatsApp.
Technology
Agent
Backend
Integrations
Interface
Illustrations of the product’s interface, drawn from how it actually works.
Use cases
Answer routine Telegram and WhatsApp questions from your own FAQ.
Scheduled checks, reports and housekeeping commands.
Find, read, write and organise files inside allowed folders.
An assistant for sensitive work that doesn’t leave your machine.
FAQ
The model runs locally through Ollama, so prompts and answers stay on your machine. Integrations you turn on — Telegram, WhatsApp via Twilio, websites you ask it to open — talk to those services.
Only within the limits you set. Shell and file access are scoped to allowed paths, and dangerous commands require your confirmation. A full-access mode exists for machines you fully control — use it deliberately.
Any model you can run in Ollama, such as Llama 3.2 or Mistral.
Python 3.10+ and Ollama with a model pulled. The installer checks both and installs the rest.
MoboAgent is in preview. Tell us how you’d use it and we’ll be in touch.
MoboAgent is in preview. Tell us what you want it to handle.
or write to hello@mobodevelopers.com