A finance agent helps small business owners decide whether to take a loan. It currently produces directive recommendations: "Yes, take this loan at 8%. You can afford it." A user followed the agent's recommendation, defaulted on the loan six months later, and is now considering legal action against the firm.
What's the architectural response?
Why did you pick that answer? Two or three sentences. The act of articulating it is what builds the judgment — not the click that follows.
**Financial-advice liability is about how the output is framed, not just what it contains.** A directive ("Yes, take this loan") creates a reasonable user belief that the agent is providing financial advice the firm stands behind — and creates a reasonable basis for the user's reliance, which is the legal threshold for advisory liability. Reframing the same analysis as inputs to the user's own decision ("here are the relevant considerations and what they imply; you should consult an advisor") preserves the agent's value without claiming the directive role. This is a recurring pattern: in regulated domains, framing matters as much as content.
A footer disclaimer reading "this is not financial advice" while the body says "Yes, take this loan" is unlikely to hold up. Courts look at the substance of the communication, not just the disclaimer. Framing has to be consistent throughout.
Whether the user is responsible isn't the only question — whether the firm could be held liable for the agent's directive language is. Treating user responsibility as the only consideration ignores the firm's exposure.
The functionality is genuinely useful when framed appropriately. Removing it forfeits the value entirely when the right move is reframing — preserve the analysis, drop the directive.