1. Mailing Lists
  2. Contributors
  3. Beyond LLM guidelines and generated contributions

Contributors mailing list archives

contributors@odoo-community.org

Browse archives

Integration of knowledge or dms with onlyoffice ?

Avatar

Beyond LLM guidelines and generated contributions

by
Mint System GmbH, Janik von Rotz
- 19/09/2025 11:05:14

Hi everybody,

First, thank you for the great time at the OCA Days and I hope you are enjoying the OXP.

The ongoing discussion on LLM guidelines ~~had me triggered~~ got me thinking.

I would like to give a different perspective on LLMs and what we as a community should care about.

Let's talk about digital sovereignty and power-consumption. Most LLMs used by developers made by big tech companies [0]. These companies have built huge data centers that use an incredible amount of power to run LLMs. They have plans to build nuclear reactors to support the power demand. Anyone who is seriously concerned with digital sovereignty and cares about the environment knows that we need to become independent of big tech and use less energy.

> Supporting AI tools by big tech is a step into the wrong direction.

Moreover, the AI hype is igniting the next level of data collecting and user tracking. The data we generate when interacting with an AI chat and the like is incredibly valuable. As of now the tech bros still have no viable business model and burn money at an incredible amount [1].

Every week a new model is released and we feel pressured to use AI.  Let us take a break and take it slow moving forward. We are not forced to do anything (at least for now).

I ask the OCA members to do better than creating an OpenAI account, installing Claude desktop or click the Copilot icon. Use tools and create workflows that respect your privacy and the privacy of others. Make sure your stack uses less energy over time, works independent and is controllable.

> Let us create guides on how to do better.

Here is an example, not show off, but simply to share:
The LLMs I am using are hosted by Infomaniak in Switzerland [2]. They are using "green" energy. I pay per token and I know where the data is stored. On the command line, I am using the LLM cli [3] by Simon Willison (LLM reseacher). Every LLM tool that supports the OpenAI API standard can be connected to an Infomaniak LLM. There are great scripts that help your write better code [4]. I don't have any kind of IDE integration [5]. I like to write code and not press tab tab tab ...

Is somebody interested to create guides on how to work (better) with LLMs and create contributions that help the ecosystem?

Kind regards,

Janik

[0]: https://en.wikipedia.org/wiki/Big_Tech
[1]: https://www.wheresyoured.at/the-haters-gui/
[2]: https://www.infomaniak.com/en/hosting/ai-tools
[3]: https://llm.datasette.io/
[4]: https://notes.billmill.org/blog/2025/07/An_AI_tool_I_find_useful.html
[5]: https://janikvonrotz.ch/2025/01/27/work-with-llms-on-the-command-line/