This telegram bot gives access to an ai agent that can search for information on the internet, use WolframAlpha, summarize youtube videos and search for pictures.
- aiogram - library for interacting with telegram
- Gemini - Google gives free access to the smartest models. Documentation
- Groq - Also totally free access to very fast llm (mostly llama). They also give you free access to whisper
- WolframAlpha - allows for complex calculations and much more
- DuckDuckGo - A fast and free alternative to google (search links, pictures). Official Site
- Tavily - A very cool API to connect LLM to the web. Allows you to give context and also parse text from pages.
- YouTube Transcript API - Allows you to summarize youtube videos
- deep-translator - for free access to google translator. Detect Language for detect language
- Whisper - Free speech to text model (can run locally)
- e2b - Python compilator
- latexonline.cc and Equation Render as LaTeX compilator
- pythonanywhere - hosting
- PyMovieDb - for free IMDB api
- Hugging Face - for free image generate with FLUX
Download the necessary libraries (requirements.txt)
Create this .env
file and fill api keys
BOT_TOKEN=
GROQ_API_KEY=
TAVILY_API_KEY=
WOLFRAM_SIMPLE_API_KEY=
WOLFRAM_SHOW_STEPS_RESULT=
DETECT_LANGUAGE_API=
GOOGLE_API_KEY=
E2B_API_KEY=
Then run bot.py
- Local llm wia hugging face and ollama
- Work with files (removebg)
- Integration with Todoist, Calendar, Gmail, Notion
- Memory as in ChatGPT
- Embedings and vector datebase
- Improve picture generate (setting, delete files, more api's)
- ElevenLabs
- Working with multiple files (merge pdf) and sending files
- Upgrate
google_full_search
andwolfram_full_answer
- See what can be taken from https://github.com/theurs/tb1
- TODO in files
- CLI interface
You can try it here if it is enabled