Quick answerWhen a customer pastes a competitor's AI agent response and asks for a direct, live comparison, the agent should not confirm, dispute, or evaluate the accuracy of the pasted content, since it has no way to verify it's genuine or representative, and should not disparage the competitor by name. The right response focuses only on what your own agent and product actually do for the customer's specific need, treating the comparison request as an opening to be genuinely useful about your own capabilities rather than a debate to win about someone else's.
This is not the same moment as recommending a competitor or a positioning framework
Proactively recommending a competitor when they're genuinely the better fit is the agent choosing, unprompted, to point a customer elsewhere. Competitive differentiation when every vendor has the same AI agent features is a marketing and product-positioning exercise, not something that happens live in a conversation. This is neither: the customer has initiated the comparison, mid-conversation, by pasting something a rival's agent said and asking yours to respond directly to it, in real time, with no time to consult marketing.
Why the agent should never verify or dispute the pasted content
A pasted screenshot or quoted response from a competitor's system cannot be verified by your agent; it could be accurate, outdated, edited, or entirely fabricated, and your agent has no reliable way to tell the difference in the moment. Confirming it ("yes, that's true, they do handle X better") risks endorsing something false or unrepresentative. Disputing it risks an inaccurate claim of your own if you're wrong about what the competitor actually does. Neither serves the customer or the company well, so the agent's job is to sidestep verifying the rival's content entirely.
What the agent should do instead: redirect to your own substance
Acknowledge what the customer is trying to figure out (which product fits their need better) without engaging point-by-point with the pasted content. Answer honestly and specifically about what your product does for the situation the customer actually described, including real limitations where relevant, rather than a generic sales pitch. This is a case very close in spirit to the customer who tells you they're a competitor or a journalist: the right response is calm, honest, and about your own product, not defensive or evasive.
From the team
We build production AI systems for startups.
LLM pipelines, RAG, and agent workflows that hold up under real traffic — not just in the demo.
Never let the agent trash-talk the competitor by name
Even if the pasted content makes the rival look bad on its face, the agent should not pile on with unverified negative claims about the competitor. This is both a brand-risk issue (an AI agent caught disparaging a named competitor reflects worse on your company than the original comparison ever would) and a factual-accuracy issue, since the agent doesn't actually know enough about the rival's real capabilities to make confident negative claims.
This is a different scenario from using a competitor's transcript as pricing leverage
A customer citing a competitor's stated terms specifically to negotiate a better price or contract concession is a commercial negotiation moment, handled by declining to match unverified terms and routing to a human with pricing authority. This is a product-comparison moment instead, about capability and fit, not price, and the right response is substantive engagement with your own product's strengths, not a refusal to discuss it. Recognizing which of the two is actually happening matters, since treating a genuine comparison question as a negotiation attempt (or vice versa) produces the wrong response either way.
Build this into training data and eval sets deliberately
Because this scenario is specific and easy to mishandle in the moment, include real or synthetic examples of pasted-competitor-response scenarios in your evaluation process, the same way benchmarking your AI agent's performance against competitors' agents requires deliberately constructed test scenarios rather than hoping representative cases show up naturally in production traffic.
FAQ
Should the agent ever acknowledge a competitor by name at all? Acknowledging that a competitor exists and that the customer is comparing options is fine and honest; the line is making unverified factual claims about what that competitor does or doesn't do well.
What if the pasted competitor response contains a factual claim about your own product that's wrong? Correct that specific claim about your own product plainly and factually, since you do have authority to speak accurately about your own capabilities, while still declining to verify or dispute claims about the competitor's product.
Does this protocol change if the customer is clearly trying to provoke a negative reaction about the competitor? No. Keep the response calm and product-focused regardless of the customer's tone; matching provocation with provocation is exactly the failure mode this protocol exists to prevent.