Anton Shvein
ཏོ་ཧ་T0ha aka Anton Shvein
Mindful Erlang and Elixir developer. Indiehacker. AI and LLM enthusiast.
Author and translator of articles on programming, startups and Buddhism.

Lama Bot (དླ་མ་བོཏ།) Project Reflection of 2025 and Plans

Lama Bot giving teachings

It’s January 16 today. The first business week of the year is mostly over, and I have a bit of time to write my thoughts about the past and future of the Lama Bot (དླ་མ་བོཏ།) project. I was reflecting on its path during my holidays.

Lama Bot 2.0 was first deployed a bit more than a year ago. This year, I was mostly focused on platform development and marketing. We launched on Product Hunt, posted to X, Mastodon, BlueSky, and sometimes to other social media. As a result, we currently have 70 users, and about 10% of them use the bot for a long time.

Right now, the platform code seems to be stable enough. Yes, there are a lot of things to do, but Lama’s users and friends are mostly unaware of them. The current platform has the following features:

  • Integration with Telegram for user interaction;

  • Integration with Google Gemini and Open Router for LLM;

  • An admin panel for web pages, LLM prompts, and periodic message adjustments;

  • An impersonalized dialog analysis panel;

  • Web pages: landing, blog articles, etc., editable from the admin panel;

  • Periodic follow-ups editable from the admin panel.

So, I decided to devote the next year more to UX, which is not just about design and appearance but more about LLM and communication tuning. Users should feel like they are communicating with a live, compassionate, and supportive person as much as possible, not an LLM. It should provide real value to users. On the other hand, it shouldn’t hurt or cause any harm.

My goals are:

  • 🎯 Choose the main and secondary LLM for the bot that feels most empathic and human-like;

  • 🎯 Optimize the main system prompt for better communication;

  • 🎯 Add periodic messages that provoke user dialogue with the bot, resulting in growth of inner calm;

  • 🎯 Analyze existing dialogs and analytics to improve the bot based on them.

Additionally, I’m going to write a couple of articles and translations for the blog to enhance SEO, fix bugs, and improve both the bot and the platform as much as possible.

PS. The project is completely non-profit and open-source. I pay $50 monthly for Digital Ocean right now, and this cost assumes that we won’t exceed the free tier LLM limits. If you have resources to support the project in any way, please contact me.