A board relies on your AI's analysis, makes a major decision, and it goes catastrophically wrong. Shareholders sue. Who bears that liability — because I promise you it won't be my directors' plan to say "the software told us to.?

The full executive answer

The liability architecture has three deliberate layers, and I'll give them to you in the order a court would. First, positioning — which is legal substance here, not marketing: Vouli IQ is decision support, and our terms of service say so in plain language: the platform provides decision support, not financial, legal, or fiduciary advice. That's not a disclaimer bolted on; it's how the product actually behaves. Nothing executes autonomously. Every analysis is an input to a human deliberation — the human-in-the-loop isn't a safety feature we added, it's the product category. Legally, that keeps your directors squarely inside the business judgment rule: informed reliance on expert inputs, with judgment exercised by the fiduciaries. Directors have relied on imperfect advisers — bankers, consultants, counsel — for a century; the case law protects informed reliance, and punishes only blind delegation.

Second, the product is engineered to make that reliance demonstrably informed, which is where we're unlike a chatbot: every output carries provenance — which model, which inputs, which sources; every number in generated prose must trace to a licensed source or the output fails validation; board-tier analyses are cross-examined by two independent AI providers before delivery; and the append-only, hash-chained audit trail preserves exactly what the board saw, when, with what caveats attached. In the shareholder suit you're imagining, that record is your directors' defence: it proves diligent process — they consulted a governed analysis, with documented sources and confidence bounds, and exercised judgment. A platform that made decisions FOR boards would be a liability machine; a platform that documents how boards decided is litigation armour.

Third, the contractual allocation, stated without flinching: our liability is capped at fees paid in the trailing twelve months — the standard SaaS position, and I won't pretend a seed-stage vendor's balance sheet backstops a nine-figure boardroom decision, because you'd never believe it and shouldn't. The honest allocation is: we warrant the platform performs as described (the controls, the provenance, the process integrity); the decision and its consequences belong to the humans with fiduciary duty and the D&O coverage built for exactly that. What I'd resist in negotiation is uncapped liability for decision outcomes — no decision-support vendor in history has accepted that, and one who claimed to would be writing a cheque their balance sheet can't cash. What's negotiable is everything else: service warranties, security breach carve-outs from the cap, indemnities for our own IP and our own negligence.