Launched this week
Council.
Open-source, serverless, multi-LLM group chat application
9 followers
Open-source, serverless, multi-LLM group chat application
9 followers
-Talk to multiple custom AI agents at the same time in one unified chat window. -Send a message once and watch the AI agents discuss, debate, and collaborate with each other automatically. Turn it on or off with dedicated controls. -Your API keys, chat history, and settings are saved locally in your browser or desktop sandbox. No servers, no tracking. -Instantly scan and import your local Ollama models. -Automatically minifies code blocks and older texts in API prompts to save your tokens.
"I've been running a similar setup with multiple AI personas debating each other before I make decisions, so this concept clicks immediately. The auto-debate where agents discuss and push back on each other is where the real value is. Everything stored locally is the right call too. Does it let you set different system prompts per agent so you can give them distinct perspectives?"
@a_kay2 yup, it does let you add an "AI System Instructions (Prompt)" in the "Add AI" or the Edit AI option.
Recommended Free & Low-Cost AI APIs
Since Council is completely client-side and serverless, you connect directly using your own API keys. You can add multiple agents from the same company/provider > Try mixing and matching different system prompts for the same model. For example, you can create a "Project Manager", a "Software Architect", and a "QA Engineer" all running on separate instances of Groq Llama 3, or multiple local models side-by-side.
Here are the best free-tier providers to get you started:
| Groq | https://console.groq.com/ (Best Overall)
| OpenRouter | https://openrouter.ai/
| Mistral AI | https://console.mistral.ai/
| Cohere | https://dashboard.cohere.com/
| Google Gemini | https://aistudio.google.com/
| Ollama / LM Studio |
100% Free & Local (runs entirely on your local GPU/CPU). | Total privacy, no internet connection required, and zero rate limits. Can integrate with Council.
| Ollama: https://ollama.com/
LM Studio: https://lmstudio.ai/
Council started with a simple problem I kept running into: every time I wanted to compare answers from different AI models, I had to jump between tabs and copy messages back and forth. It felt slow and awkward. I wanted it to feel like a real group chat where multiple AI models could all see the same conversation, reply to each other, and let me compare different ideas in one place.
As I built it, I decided to make it privacy-first by using a bring-your-own-API-key approach, so conversations and API keys stay on your own device instead of passing through any servers. I also added support for multiple AI providers, local models, desktop and Android apps, workspace backups, and a credit saving mode.