Conversations with context
Anthon is a chat application for sports mental coaching. It combines streaming responses with knowledge retrieval, persistent memory, and interactive routines. Conversations can take place on the web, Telegram, or WhatsApp, with a shared user identity.
Behind the conversation
Channel-specific history and rolling summaries keep the context bounded. Memory has explicit read, write, and deletion flows. Voice generation runs through a durable queue, while authentication and usage limits control access to the application.
The repository includes tests for the conversation infrastructure, database migrations, and scripts for integration and browser checks against temporary database branches.
From requirements to a first release
I have worked on Anthon since 2025, within Sincro Group. I gathered the client’s needs, defined requirements, designed the interfaces and application behavior, and chose the platform and tools. I also defined how the assistant would organize and use the coach’s content in its responses.
I developed and published the first version, integrating AI models, document retrieval through RAG, and access management.
My role
I am one of three co-founders and lead product and development. I now guide Anthon 2.0, set its priorities, and coordinate releases and beta testing. That includes collecting reports, verifying fixes, and discussing requirements and conversation privacy with the team.
Anthon is currently accessible through a private beta.