Thoughtful AI Decisions
Collective decision making through structured AI deliberation — multiple models reason through a question across rounds, critique each other’s answers, and a Chairman weighs the arguments into a final verdict. RAG via Qdrant or PostgreSQL/pgvector, MCP tools, live SSE streaming.
Why Delibera?
Six reasons teams choose structured deliberation over single-model answers.
Multi-Model Councils
Orchestrate any number of LLM participants across structured debate rounds, with critique and refinement.
Chairman Synthesis
A dedicated moderator opens, regulates, and synthesises the final balanced verdict from all arguments.
Knowledge Keeper
Per-round semantic retrieval with RAG — Qdrant or PostgreSQL/pgvector, with structured cited responses.
MCP Tools
An Operator micro-agent that delegates tasks to MCP servers on demand during the debate.
Context Compression
4 strategies applied to every round — measured at 11.7–12.1% prompt-token savings.
Production Ready
.NET 10, ASP.NET Core server with REST + SSE streaming, gRPC transport, Redis distribution.
Install
Three NuGet packages and two Docker images, published on every release.
NuGet
dotnet add package Delibera.Core
dotnet add package Delibera.Server
dotnet add package Delibera.RedisDocker Hub — no build required
curl -O https://raw.githubusercontent.com/techbuzzz/Delibera/main/deploy/docker-compose.hub.yml
export DELIBERA_VERSION=10.5.3 # omit to track: latest
docker compose -f docker-compose.hub.yml up -dThe API is unauthenticated and unthrottled, and a debate spends real LLM credits. Every published port binds to 127.0.0.1. Put an authenticating reverse proxy in front before exposing this stack to a network you do not control.
What’s new
Latest release: v10.5.3. Every number below was measured, not estimated — see measured behaviour.
Documentation
Start here, then go deeper.