Colleagues

Colleagues is a concept for a panel of AI colleagues: several AI colleagues with a name, a role and company knowledge talk a decision through, instead of a single chatbot answering. The approach of several AI roles in conversation also feeds into LyvaMind.

A chatbot answers questions. A panel discusses them.

Most AI assistants used in companies are one-to-one conversations: one question, one answer, done. Real decisions work differently – sales contradicts finance, HR raises a point nobody had considered, and in the end there is a decision that has survived several perspectives. That is exactly what Colleagues models: AI colleagues with a name of their own, a fixed role (CFO, HR, sales, legal) and access to what is already known in the company.

The concept is meant for mid-sized companies in Germany, especially for managing directors who want to think a task through from several specialised points of view rather than with a single model – without buying in a real consulting team.

Technically, Colleagues is a FastAPI application with an HTMX/Alpine.js interface and a deliberately open AI connection: your choice of Claude API, local via the Claude Code CLI, or completely offline via Ollama.

Core idea

• Several AI colleagues with a name, a role and a personality discuss, instead of one assistant answering alone; the perspectives deliberately contradict each other
• The colleagues take company knowledge into account, so it does not have to be explained again with every request
• Several AI providers to choose from (cloud or local)

How to get started

The idea also feeds into LyvaMind. If you are planning a similar decision round with AI roles for your company, write to us via the contact page. We will discuss with you what makes sense for you.

Price

Colleagues is not offered as a product of its own, so there is no price list. For a similar project, we will give you a price after a conversation.