AI Agent Assistant

AI Agent workflow diagram showing Telegram trigger, transcription, OpenAI model, memory, Gmail, and Google Calendar tools

Personal AI Assistant (Telegram + OpenAI)

AI Agent Assistant is a beginner-level personal AI assistant workflow that I built to automate simple daily tasks through Telegram. The goal was to create a single assistant that can understand text and voice messages, respond intelligently, and perform basic productivity actions.

The workflow starts with a Telegram Trigger that listens for incoming messages. A Switch step routes requests based on message type. Text messages are processed directly, while voice messages are first downloaded and then converted to text using a transcription step.

After normalization in Edit Fields, the request is handled by an AI Agent connected to the OpenAI Chat Model (4o mini via API). The agent then returns an answer back to Telegram as a text response.

To make the assistant useful beyond chatting, I connected tool integrations: Gmail (read multiple messages and send emails) and Google Calendar (read events and create events). I also added Simple Memory to retain context, enabling more consistent and personalized interactions.

This project is a foundation for expanding into more advanced capabilities like structured command handling, safer permission scopes, richer memory, and additional integrations.