Features How It Works Providers Get Started Pricing Download

Get Started with AI Council

Five steps from download to your first council verdict.

1

Welcome to AI Council

AI Council is a native app for macOS, iPhone, and iPad that consults multiple AI systems simultaneously. Instead of trusting a single model, it runs a three-stage deliberation: models answer independently, review each other's responses anonymously, and a chairman synthesises their best thinking into one reliable answer.

This is different from a standard chat interface. The council metaphor is intentional - you are convening a panel of experts, not asking a single advisor.

AI Council welcome screen showing the council metaphor

2

Pick your council

Open Settings → Council and tap Add Model to open the model picker. Three models is a good starting point - enough for meaningful peer review without excessive cost or wait time.

  • BYOK builds: models are grouped by provider. Tap any model to add it. You will need an API key for each provider you want to use.
  • Setapp builds: the picker shows two sections per provider - Setapp: <Provider> (consuming Setapp credits) and the plain provider name (using your own key). Pick the section that matches how you want to pay. See Setapp vs BYOK Routing for details.

Once you have added models, choose which one acts as Chairman for final synthesis. The chairman reads all responses and peer reviews and produces the concluding answer.

Council settings showing the Add Model picker with provider sections

3

Ask your first question

Press Command+N (macOS) or the compose button (iOS) to start a new conversation. Type your question in the input area and press Return or the send button.

  • Shift+Return inserts a line break without sending.
  • A pre-send cost capsule appears above the input showing an estimated token cost or Setapp credit count for the query. Tap or click it to see a per-model breakdown.
  • You can attach images (for vision-capable models) and documents using the paperclip button.
  • Toggle Web Search to augment the query with live results from a configured search provider.
Estimates only. Token counts are predicted using OpenAI's cl100k tokeniser (typically within 5% for OpenAI / Anthropic models, within 15% for other providers). Prices come from models.dev (BYOK) and Setapp's published credit ratios (Setapp). Setapp balance is derived from your tier minus this month's usage because Setapp has no public balance API. None of these numbers represent your actual bill - always confirm against your provider dashboards before relying on them.
Composer showing a typed question with the pre-send cost capsule above it

4

Read the verdict

While the council runs, a floating status bar shows how many models have completed. The response view has three stages:

  • Stage 1 - Individual Responses: Each model's answer appears in its own tab. Tap a tab to read the raw response. Expand Raw to see the unformatted text.
  • Stage 2 - Peer Review: Each model evaluates the other responses using anonymous labels (Response A, Response B, and so on). The rankings tab shows the aggregate order by average rank position. This is where you see whether the models agree or diverge.
  • Stage 3 - Chairman's Synthesis: The chairman reads all responses and peer reviews and produces a single synthesised answer. This is the verdict. A confidence signal reflects how much the models agreed.

Below the chairman's synthesis, the app may suggest follow-up questions based on the content of the deliberation. You can tap any suggestion to send it immediately.

Response view showing Stage 1 tabs, Stage 2 peer review rankings, and Stage 3 chairman synthesis

5

Track your usage

Open Settings → Usage to see the Usage Dashboard. It shows every token the app has sent or received, with estimated costs or Setapp credit consumption grouped by provider and day.

  • The four header cards show Today, This Week, This Month, and This Year totals.
  • The stacked bar chart groups usage by day with one colour per provider.
  • Use Export CSV to download a full record for your own accounting.
  • Enable Sync usage via iCloud to combine records across your Mac and iOS devices.
  • On macOS, enable the menu bar widget in Settings → Appearance to see a live spend summary without opening the app.
Estimates only. Token counts are predicted using OpenAI's cl100k tokeniser (typically within 5% for OpenAI / Anthropic models, within 15% for other providers). Prices come from models.dev (BYOK) and Setapp's published credit ratios (Setapp). Setapp balance is derived from your tier minus this month's usage because Setapp has no public balance API. None of these numbers represent your actual bill - always confirm against your provider dashboards before relying on them.
Usage Dashboard showing header cards, stacked bar chart by day, and provider breakdown

Ready to start?

Download AI Council for macOS or iOS and convene your first council in minutes.

Download AI Council

Need more detail? Visit the Help Centre for full documentation on every feature.