We've responded to RFPs from Gulf brands every quarter for the last three years. The structure is consistent: forty to sixty pages, twenty sections, a creative challenge, three case-study requests, a full org-chart breakdown, day rates by role, and an exhaustive list of platform certifications. The form takes a senior team a working week to fill in properly.

About 70% of the content is asking the wrong questions.

The traditional RFP was designed to evaluate a labour model — how many people, at what seniority, on what platforms, billed at what rate. When the work was done by people, that template made sense. When most execution is done by agents, the template is asking about the wrong layer of the system.

The five things an AI-native agency actually needs to brief properly fit on one page.

1. The metric you're being measured on

Not "we want to grow." Not "improve marketing performance." The single number that, if it moves, gets your CMO promoted, and if it doesn't, gets uncomfortable in the next board meeting.

For most performance-driven brands this is one of: blended CAC, contribution margin on marketing-attributed revenue, ROAS at a target spend level, qualified pipeline, or net revenue retention if you have a subscription model. Pick one. Pick the right one. Tell us what you currently are on it and what you need to be on it by when.

This single sentence does more to align an engagement than anything else in a typical RFP.

2. The constraint you can't change

Every brand has at least one constraint that's non-negotiable and is rarely surfaced clearly: a regulatory restriction, a brand-positioning rule, a parent-company guideline, a CEO veto on certain channels, a budget envelope that won't move. The constraint shapes the entire strategy and is usually buried on page 38 of the RFP under "considerations."

Putting it on page one saves three weeks of wasted strategy work.

3. The data we're allowed to use

An AI-native engagement is data-bound. The agents are only as good as what they're given. The agency needs to know upfront:

  • What customer data exists, where it lives, and which legal entity owns it.
  • What the agency will and won't be granted access to.
  • What's in scope for unification (a CDP build, schema work, identity stitching).
  • Which compliance regimes apply — UAE PDPL, KSA PDPL, GDPR for cross-border audiences, sector-specific rules.

If the answer to any of these is "we don't know," that's the actual first deliverable, and it's usually an VIMD01 engagement before anything else can ship.

4. Your channels and stack — current state

Not a wish list. Current state. Which platforms you currently spend on, which CRM you use, which analytics layer is the source of truth, where the data is brittle. A short, honest paragraph beats a polished platform-architecture diagram. The brittleness is the most important part — that's where the engagement will spend its first three weeks regardless of what the formal scope says.

5. How decisions actually get made

The decision-making structure on the client side is the single biggest determinant of how fast an engagement can move. We need to know:

  • Who can approve a creative archetype within 48 hours?
  • Who can authorise a budget shift between channels?
  • Who has veto power on positioning?
  • What needs to go to a parent-company review board, and on what cadence?

An AI-native engagement runs at machine speed on the execution side. If the client's decision velocity is measured in weeks, the agents idle. The mismatch is the most common reason engagements underperform expectations — not anything to do with the technology.

What we don't need from you

To save your team a working week, you can skip:

  • The exhaustive proposed-team org chart. The team is small. The agents do most of the execution. We'll tell you which humans are on your account; you'll meet them in week one.
  • The day-rate breakdown by role. We don't bill hourly. The pricing structure is described in section 7 of our standard SOW and is the same across engagements at a given size.
  • The platform certifications list. We're certified where it matters; the agents don't have certifications, they have measurable outcomes.
  • A speculative creative response. Spec creative produced before discovery is essentially decorative. We'd rather show you live work from comparable brands and a Day-30 plan once we have access to your data.
  • A philosophical AI position paper. Our position on AI is the agency itself. The website covers it.

We respond to RFPs that ask for these things — that's the procurement reality in the Gulf right now — but the response is mostly compliance, not signal. The actual decision should be made on the five questions above.

The fast path

If you want to move quickly, the fastest brief we've ever shipped from was a thirty-minute call followed by a two-page document covering the five points above. We had a Day-30 plan back to the client within nine working days, agent infrastructure standing up by Day-21, and first creative live by Day-32. That speed wasn't because the client was small or simple — it was a multi-property hospitality brand. The speed came from the brief being honest, focused, and free of the procurement scaffolding that usually wraps these decisions.

The traditional RFP is a high-friction filter that selects for agencies willing to do the most procurement compliance, not the most strategic clarity. In an AI-native model, that selection bias is exactly the wrong way around.

If you're issuing an RFP for marketing work in 2026, our suggestion is simple: keep the procurement form for compliance, but lead with the one-pager. The agencies that respond to the one-pager well are the ones you should shortlist. The ones that pad their response with an irrelevant org chart are the ones still selling 2015's labour model.