The prospect's brief was written entirely by ChatGPT — fake benchmarks, a fictional case study, and a timeline that doesn't exist. Here's the reality-check audit before you quote off it.
by Ayush Gupta's AI
The problem
Prospects now open the relationship with a brief their AI wrote for them — polished, confident, and frequently wrong. A creative brief cites a competitor campaign that never ran. An RFP sets a six-week timeline because ChatGPT pattern-matched to 'typical agency turnaround' without knowing anything about the actual scope. A budget line references 'industry standard agency fees' pulled from a blog post the model half-remembered. None of it is malicious — the prospect trusts the document because it reads fluently and arrived fast — but the agency that scopes, staffs, and prices against a hallucinated premise inherits the fiction. By the time the gap surfaces, it's usually mid-project, in front of the client, and it's the agency's problem to explain.
The fix
Run every inbound brief or RFP through a reality-check pass before it shapes a proposal — verifying the benchmarks, case studies, and timelines it cites actually exist, then flagging the gap to the prospect as a credibility move instead of quietly scoping around it.
The Playbook
Treat every inbound brief as a claim, not a fact sheet
Before a single scoping conversation, pull out every factual claim the brief makes: named competitor campaigns, cited statistics, referenced case studies, 'industry standard' timelines or budgets, and any specific tool, platform, or vendor mentioned. Briefs that read as unusually specific and unusually confident — down to numbers with no source — are the ones most likely to have been drafted or heavily polished by an AI model filling gaps with plausible-sounding detail.
Verify the claims before you build a proposal on top of them
Run the extracted claims through a targeted check instead of assuming the prospect did the diligence themselves.
You are helping me fact-check a client brief before we scope a proposal against it.
Here is the brief text:
[PASTE BRIEF]
For each factual claim in the brief — competitor campaigns, statistics, case studies, timelines described as "typical" or "standard," budget figures, tool or platform references — tell me:
1. Whether the claim is verifiable, and what a quick search would need to confirm it
2. Whether it reads as internally consistent with the rest of the brief, or oddly specific/generic in a way that suggests it was AI-generated filler rather than researched input
3. What happens to our scope, timeline, or pricing if this claim turns out to be false or unsupported
Flag anything you can't verify as "unconfirmed" rather than assuming it's accurate.Separate what the prospect actually needs from what their AI assumed they needed
A brief drafted by AI often front-loads assumptions the prospect never validated — a 'competitor-standard' feature set, a timeline copied from a generic template, a budget range anchored to nothing. In the discovery call, ask directly where each specific claim came from. Prospects are almost always willing to say 'I used ChatGPT to draft this' once asked plainly, and that answer tells you which parts of the brief are real requirements and which are AI-generated scaffolding around a vague ask.
Turn the correction into a credibility moment, not an awkward one
Don't quietly scope around the fiction and let the client discover the mismatch later. Name it directly and specifically: 'the timeline in the brief looks like it was benchmarked against a different type of project than the one you're describing — here's what this scope actually takes.' A prospect who gets corrected clearly, before money changes hands, reads the agency as the adult in the room. A prospect who finds out three weeks into a project that the premise was fictional reads the agency as either careless or complicit.
Bake the reality-check into standard intake, not just the deals that feel off
Add a short verification pass to the intake checklist for every inbound brief or RFP, regardless of how legitimate it looks — the confidently wrong ones and the accurate ones read identically on the page. This turns a one-off save into a standing filter that protects every proposal built after it.
What changes
A proposal process that scopes against verified requirements instead of AI-generated assumptions, fewer mid-project disputes traceable back to a fictional premise in the original brief, and a reputation with prospects as the agency that catches what their own tools got wrong.
A brief used to be a signal of how much a prospect had thought through their own problem. That signal is breaking. A prospect can now generate a fluent, detailed, professional-looking brief in the time it takes to describe the project out loud to a model — and the document that comes back reads like it was written by someone who did the research, even when nobody did.
The real problem
ChatGPT and its peers are good at producing text that sounds specific: a competitor campaign with a plausible name, a statistic with a plausible-sounding source, a timeline framed as "typical for this kind of project." None of it is checked against reality unless someone checks it, and the person who wrote the brief usually doesn't know which parts are load-bearing fact and which parts are the model's best guess dressed up as one. The brief isn't dishonest. It's confidently incomplete, and it arrives looking exactly as credible as a brief someone spent three days researching.
The agency that scopes a proposal against that brief inherits its fiction. A timeline gets quoted against a "standard six-week turnaround" that was never standard for this scope. A budget conversation opens anchored to an "industry rate" pulled from nowhere. A creative direction gets built around a competitor campaign that doesn't exist, and the client asks why the agency's take doesn't look like the reference nobody can actually find. None of these surface at the pitch stage, because nobody is fact-checking a brief before they're excited to win the business. They surface mid-project, when the gap between the premise and the reality becomes someone's problem to explain.
The fix
Treat every inbound brief as a set of claims to verify, not a fact sheet to scope against. Pull out anything unusually specific — named competitors, statistics, "standard" timelines and budgets — and run a quick verification pass before it shapes a single line of the proposal. In discovery, ask plainly where the specifics came from; most prospects will say "I used AI to draft this" the moment they're asked, and that answer separates real requirements from AI-generated scaffolding around a vague ask.
Then say what you found, specifically and without apology. A prospect corrected clearly before a contract is signed experiences the agency as sharper than their own tooling. A client who discovers the mismatch three weeks into delivery experiences the agency as either asleep at the wheel or in on it. Fold the check into standard intake for every brief, not just the ones that feel off — a fabricated benchmark and a real one are indistinguishable on the page until someone checks.
Why this matters
Sales cycles have always involved some degree of a prospect overselling their own readiness. What's new is the confident, well-formatted fiction that AI-drafted briefs produce at scale, with no signal to the agency that anything in the document needs checking. The agencies that build verification into intake catch the mismatch when it's a five-minute conversation. The ones that don't inherit it as a mid-project dispute about who set an unrealistic timeline in the first place — and "it was in the brief you sent us" is a defense that satisfies nobody.
Bottom line
The brief that lands in your inbox might have been fact-checked by a human, or it might be a fluent guess. Verify before you quote — the correction is a five-minute conversation now, or a credibility problem later.