The Brief is Dead. Long Live the Brief.

The traditional design brief has a structural problem that nobody talks about: it asks clients to articulate things they don't yet know. "What's your target audience?" is a reasonable question. But most small business owners haven't done a rigorous segmentation exercise. They have a feeling. They have a customer they like and one they don't. They have a competitor they admire and one they resent. None of that lives in a PDF form.

For years, the solution was the discovery call — an hour-long conversation designed to excavate the real brief beneath the stated one. A good strategist could walk out of that call with more clarity than the client had at the start. But the brief that followed was still a snapshot. A document. Static. Finished the moment it was saved.

AI is making both parts of this process better, and in doing so, forcing us to rethink what a brief is actually for.

Why Traditional Briefs Fail

The brief fails at the point of information asymmetry. The designer knows what questions are strategically important; the client knows the answers. Getting those two things aligned before work begins has always been the purpose of the discovery process — but the tools available have been limited. A form. A call. A document.

The result is predictable. The brief captures what the client can easily articulate, not what matters most. It documents stated objectives but misses unstated ones. It reflects the client's self-perception rather than how they're actually perceived by customers. And because the designer doesn't know what they don't know until the project is halfway done, the gaps only become visible when fixing them is expensive.

This is not a failure of either party. It's a structural failure of the format. A static document asked to carry dynamic, contextual, often contradictory information will always fall short.

The Information Asymmetry Problem

Here's the specific problem with most briefs: they're written from the client's perspective, filtered through what the client thinks a designer needs to know. This is understandable. Clients are not designers. They don't know which details are strategically load-bearing and which are irrelevant. So they default to describing their business, their history, their ambitions — and designers end up with a lot of background information and not much strategic signal.

The most important information in any brief is the thing the client doesn't know they're telling you. Everything else is preamble.

Good discovery conversations work because a skilled interviewer knows which threads to pull. They hear "we want to look more professional" and ask what specifically feels unprofessional right now. They hear "our competitors are doing well" and ask what the client thinks they're doing that you're not. This kind of inference is where the real brief lives — not in the answers, but in the space between them.

Using AI to Synthesise Research

The first place AI changes the brief process is in pre-discovery research. Before we speak to a client, we now run a structured research pass using Perplexity and Claude to synthesise their market, competitors, and public perception. This takes twenty minutes and produces something that used to take half a day.

What we're looking for isn't facts — the client knows their own business better than any AI. We're looking for hypotheses. Going into a discovery call with a set of informed guesses changes the dynamic entirely. Instead of asking open-ended questions and hoping something useful emerges, we're testing specific ideas: "We noticed your strongest competitor is positioning heavily on speed — do you see that as an area you can compete on, or are you differentiating elsewhere?"

That question only gets asked if we've done the research. And the answer to it is worth more than twenty minutes of open-ended exploration.

The New Discovery Call Format

We've rebuilt our discovery process around a simple principle: don't ask questions you can answer yourself. If something is findable through research, find it before the call. Reserve the conversation for the things only the client can tell you — their real anxieties, the decisions they regret, the customers they'd clone if they could.

Our current discovery call structure looks roughly like this:

1. Share back what we found in research — let the client correct our assumptions
2. Ask about the one thing they most want to change about their current site
3. Ask who their best customer is, specifically — not a demographic, a person
4. Ask what they would never say in their marketing, and why
5. Ask what they think is going to be different about their business in two years

Five questions. Ninety minutes. More useful than any form we've ever designed.

The Case for Living Briefs

The biggest shift AI enables is moving from a static brief to a living one. Instead of a document produced before work begins and filed away when it's done, a living brief is a continuously updated record of decisions, discoveries, and evolving understanding.

In practice, this means we maintain a shared document — a Notion page, usually — that gets updated throughout the project. When we make a significant decision, we record it and the reasoning behind it. When a client changes direction, we note what shifted and why. When we discover something during build that reframes an earlier assumption, we update the record.

The AI layer here is in synthesis. At the end of each phase, we run the notes through Claude and ask it to identify any contradictions with earlier decisions, any unstated assumptions we haven't examined, and any open questions that need answering before we proceed. This catches things that would otherwise only surface as problems in review.

How to Prompt a Client for Insight

One underused application of AI in the brief process is using it to generate better questions. Before a discovery call, we'll often give Claude a summary of what we know about the client and ask it to generate twenty questions we haven't thought to ask. Most of them are unusable. But three or four are genuinely better than what we would have come up with ourselves.

The questions AI tends to generate well are second-order questions — ones that probe the reasoning behind a stated position rather than the position itself. "Why do you want to look more professional?" instead of "What does professional mean to you?" The difference is subtle but significant.

• What decision would you make differently if you were starting the business today?
• What do your best customers say about you that you wish more people knew?
• What's the one objection you hear most often, and what do you say back?
• Who are you trying to stop working with, not just who you want to attract?
• If this website works exactly as you hope, what changes in your business six months from now?

Documenting Decisions in Real Time

The final piece of the modern brief is decision logging. Every significant design choice — a colour palette, a structural decision, a copywriting direction — gets logged with the rationale at the time it was made. This sounds bureaucratic. In practice it's transformative.

Six weeks into a project, when a client asks "why did we go with this approach?", having the answer on record changes the entire character of that conversation. You're not reconstructing reasoning from memory; you're retrieving a decision that was made thoughtfully and documented clearly. It builds trust. It also prevents scope creep — it's much harder to revisit a decision when it's recorded alongside the reasoning that made it the right call.

AI makes this easier because synthesis is cheap. A ten-minute Loom walkthrough of a design decision, transcribed and summarised by Claude, becomes a searchable record in two minutes. The overhead is low enough that it actually happens, rather than being a process that sounds good in theory and disappears under delivery pressure.

The brief isn't dead. The PDF form is. What replaces it is a dynamic, collaborative, AI-assisted process that starts before the first call and runs until the final handoff. It's more work upfront and considerably less work later. That trade-off is, finally, starting to feel obvious.