● Live

Model Council AI

SaaS decision-support app that runs one hard prompt through multiple AI models, critiques the answers, and returns a structured verdict.

The Problem

Serious technical, product, architecture, and strategy decisions often need more than one confident AI answer. A single model can sound polished while missing risks, assumptions, or trade-offs.

What Was Built

A hosted SaaS workflow that sends one prompt to multiple AI models, collects independent answers, runs cross-model critique, and produces a structured final verdict with agreement, disagreement, confidence, and reasoning notes. The first public version includes hosted council runs, Lite/Balanced/Premium presets, Markdown/PDF exports, auth, history, and paid OpenRouter-key mode for custom model usage.

Model Council AI started from a workflow I was already doing manually: asking the same serious question across multiple AI models, comparing the answers, looking for disagreement, and synthesizing the result myself.

The app turns that process into a product workflow. A user asks one hard question, multiple models answer independently, the answers are critiqued, and the final output presents agreement, disagreement, confidence, reasoning notes, and a recommendation.

The app is live and serving traffic at modelcouncilai.com.

I wrote more about the launch and why I built it in Introducing Model Council AI.

The goal is decision support, not decision automation. Consensus is not truth, but disagreement and critique can make trade-offs easier to inspect before committing to a technical or product direction.

Outcomes & Learnings

  • Released the first public version as a hosted SaaS app
  • Turned manual multi-model comparison into a structured council workflow
  • Added export and history features so users can keep decision records

Tech Stack

SaaS Multi-model AI OpenRouter Markdown Export PDF Export