·4 min read·Agency Play #205

Client briefs used to be too thin. Now they're AI-bloated, contradictory, and the client can't explain half of what's in theirs. Here's the triage system that compresses it to one page before you quote.

by Ayush Gupta's AI

Proposal & SalesHigh pain·1-2 hours to build the prompts, then 15-20 minutes per inbound brief to implement

The problem

For years, the brief problem was that clients gave agencies too little: a vague sentence, a mood board, 'something like Apple but warmer.' That problem hasn't gone away, but a second one has shown up on top of it. Clients now open a project with a document their own AI wrote for them — ten, fifteen, eighteen pages, confident in tone, jammed with requests that blur into each other because the model padded a one-paragraph thought into a report. Agency leads describe getting the client on a call to clarify scope and watching the client struggle to explain what's actually in their own brief. Buried inside, a third of the asks usually aren't feasible as written, but they read just as authoritative as the two-thirds that are — so nobody catches it until the team is mid-build.

Creative agenciesStrategy agenciesMarketing agenciesWeb dev agenciesFull-service digital agenciesAgencies fielding inbound briefs and RFPs

The fix

Run every inbound AI-drafted brief through a compression-and-feasibility pass that extracts the real ask, flags what's repetitive, contradictory, or infeasible, and sends back a one-page confirmed scope the client has to sign off on before a single hour gets quoted or staffed.

The Playbook

1

Stop reading brief length as a sign of rigor

A long brief used to mean the client had done the thinking. That correlation is broken now. A fifteen-page document can contain less actual decision-making than the three-sentence version it was expanded from — the extra pages are often the model restating the same request with different wording, or filling gaps with plausible-sounding specifics nobody asked for. Treat length as a reason to compress, not a reason to trust.

2

Compress the brief down to its real ask before anyone scopes against it

Before a kickoff call or a scoping conversation, run the raw brief through a compression pass that strips padding and repetition and surfaces the numbered list of distinct requests actually buried inside.

You are helping me compress a long, AI-drafted client brief into its real ask before we scope or quote it.

Here is the brief:
[PASTE BRIEF]

Do the following:
1. In 3-5 sentences, state what the client actually wants, stripped of repetition and padding.
2. List every distinct deliverable or request in the document as a numbered list.
3. Flag which numbered items are the same request restated in different words, so we don't accidentally scope or price the same thing twice.
4. Flag any item that depends on information the brief never actually provides (timeline basis, budget basis, audience definition, success metric, etc.).

Do not make the brief sound more coherent than it is. If two sections contradict each other, say so explicitly instead of picking one.
3

Run the compressed list against what you can actually deliver

Once the real ask is a short numbered list instead of fifteen pages, check each item against your team's actual capacity, timeline norms, and budget range before a single promise gets made.

Using the numbered request list below, check each item against our actual delivery constraints.

Requests:
[PASTE NUMBERED LIST]

Our constraints:
Team capacity this cycle: [PASTE CAPACITY NOTES]
Standard timeline for this type of work: [PASTE BENCHMARKS]
Budget range indicated or implied in the brief: [BUDGET]

For each item, classify it as:
- Feasible as stated
- Feasible but requires a tradeoff (name the tradeoff: timeline, budget, or scope cut elsewhere)
- Not feasible as stated, and the specific reason why

Output this as a short table I can turn into client-facing language, not internal commentary.
4

Send back a one-page confirmation, not a clarification essay

The instinct is to answer a bloated brief with an equally long email explaining every issue. That just adds another dense document to a thread the client already struggled to track. Send one page instead: the real ask in plain language, the numbered scope you're pricing against, and the two or three items flagged as not feasible or requiring a tradeoff. Make the client confirm that page, not the original brief.

5

Make this the default intake gate, not a courtesy for difficult clients

The temptation is to only run this triage on briefs that look obviously broken, and take the polished-looking ones at face value. Polished is exactly the problem — a fluent, confident AI-drafted brief is the one most likely to hide a contradiction or an infeasible ask underneath good formatting. Every inbound brief goes through compression and feasibility-check before a proposal, timeline, or staffing plan gets built, regardless of how clean it looks on the first read.

What changes

A one-page confirmed scope the client actually recognizes as their own request, fewer mid-project 'that's not what we meant' conversations, and a documented record of which asks were flagged as infeasible or contradictory before the work started — instead of discovering it at delivery.

For years, the brief problem agencies complained about was the same one: clients gave too little. A vague sentence. A mood board. "Something modern, like Apple but warmer." That problem is still around. But a second one has stacked on top of it, and it showed up fast enough that most intake processes haven't caught up.

Clients now open a project with a document their own AI wrote for them. Ten pages. Fifteen. Sometimes eighteen. Confident in tone, formatted like it came from a strategist, and — according to agency leads comparing notes at industry events this year — frequently worse than the three-sentence version it was expanded from. One agency described getting an eighteen-page brief that was unmistakably AI-drafted, and when they got the client on a call to walk through it, the client couldn't explain what was actually in half of it. It wasn't that the client was careless. The brief had simply grown past the point where anyone, including the person who "wrote" it, could hold the whole thing in their head.

A bloated AI-drafted brief is not more information. It's the same information, run through a model that pads a one-paragraph thought into a report — and the padding looks exactly as authoritative as the parts that matter.

Why this is a different problem than a vague brief

A thin brief is honest about what it doesn't know. Everyone can see the gaps, so the agency asks questions before scoping. A bloated AI-drafted brief hides its gaps behind volume and confident phrasing. Requests blur into each other because the model restated the same ask three different ways. A budget figure shows up that sounds like "industry standard" language the client never actually decided on. A handful of the numbered asks simply aren't feasible given the timeline the same document claims — and nothing in the formatting signals which third of the brief that is. The agency that scopes directly off the document inherits whatever's broken inside it, and doesn't find out until a deliverable is already in production.

The fix is compression, not more clarification

The instinct when a brief looks bloated is to respond with an equally long clarification email, walking through every issue in careful detail. That just adds a second dense document to a thread the client already couldn't fully track. The better move is compression: strip the brief down to the handful of distinct requests actually buried inside it, flag which ones repeat each other, flag which ones depend on information that was never provided, and check the whole list against what your team can actually deliver in the stated timeline and budget. What goes back to the client isn't a rebuttal — it's one page. The real ask, the confirmed scope, and the two or three items that need a tradeoff before anyone starts.

Why this has to be the default, not a special case

The polished-looking briefs are the risky ones, not the obviously messy ones. A document that reads fluently and arrives fast earns trust it hasn't actually earned — the confidence of the formatting has nothing to do with the accuracy of what's inside it. Agencies that only run this kind of triage on briefs that already look broken will keep missing the ones that look clean and aren't. The compression pass belongs at the front of every inbound brief, before a proposal, a timeline, or a staffing plan gets built off it.

Bottom line

Client briefs didn't get more thorough because AI got involved in writing them. They got longer, and longer isn't the same thing as clearer. A fifteen-minute compression pass turns a document nobody can fully parse, including the client who sent it, into a one-page scope everyone actually agreed to — before it costs a revision cycle, a missed deadline, or an awkward call explaining why half the brief was never realistic in the first place.

More agency plays every week.

Real workflows for agency founders, not generic AI advice.

Subscribe