An LLM-powered Telegram chatbot that leverages personal context to provide tailored conversational experiences.
This bot interacts with users by processing messages through Large Language Models, utilizing either local inference or cloud-based providers. It is designed to maintain a specific persona or context provided during setup.
- Context-Aware Chat: Uses provided background data to inform LLM responses.
- Flexible Backend: Supports multiple LLM execution paths:
- Cloud: Provider-based inference via AWS Bedrock.
- Local: Integration with Ollama (Legacy).
- Reliable Hosting: Configured for seamless deployment on Railway.
This project uses Continuous Deployment via Railway.
- Every push to the
mainbranch triggers a new build. - To skip deployment: Include
[skip cd]in your commit message if you are pushing changes that do not require a live update (e.g., documentation or README edits).
Note: More advanced versions of this bot, including agentic workflows and specialized tool-calling, are hosted in private repositories.
- Infrastructure: Railway
- Interface: Telegram Bot API
- AI Orchestration: AWS Bedrock / Ollama
Ulises Rey