On April 30, Google announced AI Brief, a Gemini-powered control surface inside AI Max. The premise is simple: instead of bidding adjustments, audience signals, and asset groups, you describe — in sentences — how the campaign should behave. PPC Land's coverage walks through the launch details, but the punchline is that Google has finally given advertisers a place to write down what the model should do, rather than reverse-engineering it from settings.
The feature is shipping in stages. English plus Search at launch on April 30. Performance Max and Shopping later in 2026. And the headline timeline most SMBs should have on a sticky note: AI Max replaces Dynamic Search Ads in September 2026. If you've been running DSA, you have a runway, not an option. Getting fluent now is the conservative move.
The honest take: turn AI Brief on this week, write a real brief, and treat the guideline file as a living document. The teams that will win in the next two quarters are the ones whose guidelines are crisp. The teams that will quietly underperform are the ones that leave the box empty or fill it with marketing-deck adjectives.
What AI Brief Actually Is
Mechanically, AI Brief is a Gemini-backed editor inside AI Max that accepts three kinds of natural-language rules:
- Messaging Guidelines — what the ads should say, how they should say it, and what they should never say.
- Matching Guidelines — which queries the campaign should chase and which it should leave alone.
- Audience Guidelines — which user segments the campaign should treat as priority, and how the messaging should shift for them.
You write these in sentences. Gemini interprets them at generation time and applies them across headlines, descriptions, query selection, and audience prioritization. There is a preview step before anything ships, which is the second-most important feature in the release and the one most teams will underuse.
The mental model is closer to a creative brief than a settings panel. You're not configuring; you're directing. That sounds soft until you realize the alternative is a black-box Performance Max campaign with no place for institutional knowledge to live. AI Brief is the place. Use it.
It's also the second large platform to bet on plain-English steering this spring. Meta moved the same direction with its "Describe Your Audience" box in Advantage+ Targeting. The cross-platform pattern is real, and we wrote about it in the broader multi-platform AI ad orchestration piece. AI Brief is Google's version, and it goes deeper than the audience seed — it touches messaging and query selection too.
The Three Guideline Types, With Examples
Use a worked example throughout: a small ecommerce brand selling clean-ingredient pantry staples — cooking oils, baking flours, spice blends. About 90 SKUs. The brand has a strict no-discount policy and a strong "ingredient transparency" voice.
Messaging Guidelines
Messaging Guidelines tell Gemini what to write. They are the difference between generic Performance Max output and ads that sound like your brand.
For our pantry brand, the messaging block looks like this:
Never mention prices, percentages off, or sales — we don't discount.
Lead with ingredient transparency: name the source country or farm where space allows.
Use plain, food-forward language. Avoid "premium," "luxury," or "gourmet."
For oils, emphasize cold-press or unrefined origin. For flours, emphasize stone-milling.
Always include "non-GMO" when describing baking ingredients.
Never use words like "boost," "supercharge," or "transform."
Six lines. Each one is a constraint or a positive direction. None of them are aspirational ("be authentic and bold"). The model can act on the version above. It can't act on the aspirational version.
Notice what's missing: brand color, logo placement, hashtag rules. Those belong elsewhere — AI Brief is about words, not visual identity.
Matching Guidelines
Matching Guidelines tell Gemini which queries are in-bounds and which are not. AI Max already expands beyond your keyword list by default; Matching Guidelines are how you keep that expansion useful.
For our pantry brand:
Prioritize searches for healthy pantry staples, clean-label cooking ingredients, and organic baking supplies.
Pursue searches that include "non-GMO," "organic," "single-origin," "stone-milled," or "cold-pressed."
Do not pursue searches for restaurant supply, bulk wholesale, or food-service distribution.
Do not pursue searches for diet-specific meal plans (keto, paleo, carnivore) — we don't market that way.
Treat searches for specific recipes as in-bounds only when an ingredient we sell is mentioned by name.
Matching Guidelines are where you spend the budget that AI Max would otherwise spend for you. The instinct is to write tighter rules. The right move is usually broader rules with a strong negative list. Gemini is good at expanding intent; it is less good at guessing which adjacent verticals you don't serve.
Audience Guidelines
Audience Guidelines tell Gemini how to shift messaging for different user segments. This is the guideline type most teams skip, and it's the one with the highest leverage.
For our pantry brand:
For health-conscious users, highlight clean ingredients and sourcing transparency in headlines.
For home bakers, lead with stone-milling, freshness, and recipe applicability.
For repeat customers (existing CRM segments), highlight new arrivals and seasonal SKUs rather than category basics.
For first-time visitors, lead with the brand's no-discount, no-fillers position so the proposition is clear.
Do not target professional chefs or food-service buyers.
This is the surface where institutional knowledge — the stuff that used to live in your head and your media planner's spreadsheet — finally has a home. If you have ever said "we sell differently to repeat customers than to cold traffic," that sentence belongs in Audience Guidelines.
The Preview Step Is Your Creative QA Gate
Before anything publishes, AI Brief shows you the headlines, descriptions, and query interpretations Gemini intends to use. Most teams treat the preview as a formality. It isn't.
Use the preview as an editorial gate. Read every headline and ask three questions:
- Does it violate a Messaging Guideline? If "20% off" appears anywhere, the constraint isn't sticking. Tighten the wording. "Never mention prices, percentages, or discounts of any kind" beats "we don't discount."
- Does it sound like your brand? If it doesn't, the brief is too vague. Add a sentence about voice. "Plain, food-forward language" did real work in our example.
- Does it match an intent you actually fulfill? If you're a single-origin oil brand and the preview shows a query for "cheap cooking oil bulk," the Matching Guidelines need a negative.
The preview is also where you'll catch the most expensive mistake: ads that are technically on-brand but addressed to the wrong audience. A perfectly written headline pointed at a segment you don't sell to is worse than a generic headline pointed at the right segment. Audience Guidelines fix that, but the preview is where you discover you needed them.
One discipline that scales: keep a "what the preview showed and what we changed" log for the first month. Two paragraphs a week. By month two you'll have your own playbook for how Gemini interprets your brief, which is more valuable than any external guide.
The Trade-Off Most Teams Miss
Conventional wisdom on Performance Max has been: tighten the audience signals, narrow the asset groups, fence the campaign. AI Brief inverts that instinct in a specific way.
The trade-off most teams will miss: good guidelines outperform tighter targeting. Vague guidelines drift.
If your Messaging and Matching Guidelines are crisp, you can let AI Max expand more aggressively, because you've told it exactly what counts as a hit. The model has a clear fitness function. If your guidelines are vague — "be authentic, target our customers, drive conversions" — tighter targeting won't save you. It will just produce a more constrained version of the same drift.
This is the same lesson we keep relearning across platforms: the more capable the model, the more leverage clear instructions have. Vague instructions to a less capable model produced mediocre results. Vague instructions to a more capable model produce confidently mediocre results, which are harder to diagnose. The fix is upstream of the campaign settings — it's in the brief.
If you've been running both Google and Meta and want to keep the comparative frame in mind as both platforms move toward natural-language steering, our Facebook vs. Google Ads 2026 post tracks the diverging philosophies in detail.
The Six-Line Template Most SMBs Need
You don't need a thousand-word brief. You need six lines that actually constrain Gemini's behavior. Use this scaffold and fill in your specifics:
1. [Voice rule] — Use [tone descriptor] language. Avoid [words that are off-brand].
2. [Constraint rule] — Never mention [thing you don't do: prices, claims, certifications you don't hold].
3. [Positive lead rule] — Lead with [your strongest differentiator: ingredient sourcing, response time, certification].
4. [Matching positive] — Prioritize searches for [the 2–3 query themes that convert].
5. [Matching negative] — Do not pursue searches for [the 2–3 adjacent verticals you don't serve].
6. [Audience shift] — For [your most distinct segment], emphasize [the angle that resonates with them].
The template covers ecommerce, lead-gen, and local services equally well. A roofing company would write line 3 as "Lead with licensing, insurance, and same-week scheduling." A B2B SaaS would write line 5 as "Do not pursue searches for free tools, open-source alternatives, or downloadable templates." A local restaurant would write line 6 as "For dinner-time searches within 5 miles, emphasize availability tonight and the menu's seasonal items."
Six lines. Then iterate based on what the preview shows you.
Rollout Timeline and What to Do Now
Here is the calendar to plan against:
| Date | What Ships | What to Do |
|---|---|---|
| Live now (April 30, 2026) | AI Brief in AI Max for Search, English-language accounts | Write a six-line brief. Run preview. Log results weekly. |
| Mid-2026 | AI Brief expanding to Performance Max | Have your Messaging and Audience Guidelines polished by then. Performance Max will inherit them. |
| Mid-2026 | AI Brief expanding to Shopping | Plan SKU-level Matching Guidelines if you have a wide catalog. |
| September 2026 | AI Max replaces Dynamic Search Ads | If you still run DSA, your campaigns auto-migrate. Have a brief written before then or accept Gemini's defaults. |
The September deadline is the one that moves AI Brief from "interesting feature" to "operational necessity." Every account currently running DSA will be steered by AI Max with whatever brief is in the box on migration day — including, by default, an empty one.
What Stays Human
AI Brief is a directing tool, not a strategy generator. The decisions that still belong to a person:
- The differentiator you put in line 3. Gemini cannot guess what makes your business specifically worth choosing. If you can't write the differentiator in one sentence, the brief won't be the bottleneck — your positioning will be.
- The negatives in line 5. Adjacent verticals you don't serve are the highest-ROI rules in the brief, and they require knowing your business well enough to identify them. Pull a quarter of search-term reports and sort by spend with low conversions; that's where line 5 comes from.
- The audience segments you call out. Gemini will create reasonable segments on its own. The ones that compound for you — repeat customers, high-LTV cohorts, regional buyers — need to be named explicitly.
- The conversion definition. AI Brief steers the front of the funnel. The back of the funnel — what you count as a conversion, how you feed signal back to Google — is still a measurement decision. (Same lesson we made in the Meta one-click Conversions API piece: better steering can't fix bad measurement.)
- The decision to update. Briefs go stale. Pricing changes. New SKUs ship. A brief written in May should not be untouched in November.
Common Pitfalls
Writing the brief like a marketing deck
"Be bold, authentic, and customer-centric" is a slogan, not a guideline. Gemini cannot act on it. The model does not know what "authentic" means for your brand specifically. Replace every adjective with a constraint. "Avoid corporate language; use the voice of a small founder explaining the product" is something a model can use. "Authentic" is not.
Skipping Audience Guidelines
Most accounts will fill in Messaging and Matching and leave Audience Guidelines blank. That's the highest-leverage box, because it's where institutional knowledge that used to live in spreadsheets and meetings finally has a place to go. Skipping it means the messaging is generic across segments that should sound different.
Treating preview as a formality
Click-through-and-publish on the preview defeats the entire feature. The preview is the gate where you discover the brief was vaguer than you thought. Read every headline. If anything surprises you, the brief needs a sentence.
Over-constraining the matching rules
The instinct from years of keyword-based campaigns is to enumerate exactly what you want. AI Max performs better with broad positive matching plus a strong negative list. If you find yourself writing twenty positive rules, you've recreated keyword-level control inside a model that wants to expand. Three to five positives, three to five negatives, then let the preview correct you.
Setting it and forgetting it
The brief is a living document. Pricing shifts, new product lines launch, seasonal angles change. A quarterly brief review — same cadence as a creative refresh — is the minimum. Monthly is better while you're learning what Gemini does with your specific phrasing.
The Bottom Line
Turn AI Brief on this week. Write a six-line brief — voice, constraint, positive lead, matching positive, matching negative, audience shift. Run the preview, read the output critically, and tighten any guideline that didn't constrain the way you expected. Treat the brief as a living document and update it on the same cadence as your creative.
The accounts that will outperform over the next two quarters are not the ones with the cleverest bid strategies or the tightest audience signals. They are the ones whose guidelines are written like instructions rather than aspirations. AI Max replacing Dynamic Search Ads in September 2026 is not a feature deadline; it's a fluency deadline. The brief you have on file that day will steer everything that follows.
The honest framing for the next planning conversation: "AI Brief is where the strategy lives now. If we can't write our positioning in six lines, we don't have a positioning problem the campaign settings can fix — we have a positioning problem the brief is exposing."
Want AI Brief written and maintained for you?
Ads Agents drafts, previews, and iterates on AI Max guidelines as part of the campaign setup — including the Audience Guidelines most teams skip. See the integration or check pricing.
Get Started Free →