Why your free seat is actually free forever

Free tiers usually have a catch. Quorum Lite's catch is that we pay the GPU bill. Here's the mechanism that makes that sustainable.

The policy

Quorum's cloud-cost actions — Ask-Claude, Re-analyze, the 10-seat Pro council, the Max Opus sweep — are owner-only escalations. Free users see a clean 403 with a redirect to the local council instead. The button isn't there to taunt you; it's gated server-side as defense in depth.

What free users get instead is the 10-seat local council running on our private GPU — a single physical machine in our box, one user at a time, the same canonical synthesizer and Executive Briefing every paid tier writes.

The numbers

• 20 runs per user per day, resets at 00:00 UTC.
• 3 watchlist tickers — focused, not infinite.
• ~3-5 minutes per ticker end-to-end. Slow because the model reads the filing instead of guessing.
• Zero cloud LLM cost for us — your free run never touches Anthropic, OpenAI, NIM or Gemini.

Why we keep it free

Two reasons, both pragmatic.

• The prompts sharpen faster on real portfolios. Five real users running 20 verdicts each against their actual holdings surfaces more sycophancy, more hallucination, more confidence drift than any synthetic eval. The free seat is how we get there.
• It's not a loss-leader for an enterprise upsell. There's no enterprise tier behind a sales call. Lite stays free because the local GPU and the prompts are the product — Pro and Max only add depth and breadth.

What we'd ask of you

If it gets a call right, tell us which dissent moved you. If it gets one wrong, tell us which falsifier we should have named. That feedback is the entire compensation we ask for the free seat. Sign in / join the waitlist →