Meta moved its AI advertising tooling into a new gear this month. The Meta AI Business Assistant, until recently a closed beta for select advertisers, is now generally available to every advertiser worldwide across Ads Manager and Meta Business Suite. The assistant answers performance questions in natural language, surfaces optimization recommendations, runs benchmarking against industry peers, and troubleshoots account issues. Meta also shipped AI connectors that let third-party platforms drive Meta ads through their own AI workflows. For the millions of small businesses, agencies, and enterprise marketers running Meta campaigns, this is the most significant change to the ads operating model in years.
What’s actually new
The Meta AI Business Assistant is a conversational interface built directly into Ads Manager and Meta Business Suite. It does not sit in a separate tool; it surfaces as a sidebar alongside the ad accounts, campaigns, ad sets, and ads the advertiser already manages. The advertiser types or speaks a question in natural language, the assistant reads the relevant account context, and it responds with an answer plus, where appropriate, actions the advertiser can apply with one click.
The previously gated capabilities have all expanded. Performance Q&A is the headline workflow: “Why did my cost-per-action increase last week?” produces a structured answer with the contributing factors (audience saturation, creative fatigue, competitor pressure, attribution model shifts), each backed by the underlying data. Optimization suggestions sit alongside performance Q&A: the assistant proposes specific, applicable changes (budget reallocation across ad sets, audience adjustments, creative refresh recommendations) with predicted impact ranges. Benchmarking compares performance against industry peers anonymized at the cohort level. Troubleshooting answers the operational questions (“Why is this ad rejected?”, “How do I fix domain verification?”) that legacy advertiser support used to handle through long tickets.
The language coverage expanded with the global rollout. The assistant supports English, Spanish, French, German, Portuguese, Mandarin, Japanese, Korean, Italian, Dutch, and several other major business languages at production quality. The previous beta was English-mostly; the global GA is genuinely multilingual.
The new piece of the May launch is AI connectors. Meta now exposes a structured interface that lets external AI platforms (ChatGPT for Excel, Claude, Gemini-based agents, custom in-house orchestration layers) read campaign data and execute changes through Meta’s APIs while honoring Meta’s policies. The connector pattern is the bridge between the Meta-owned ads operating environment and the external AI tools advertisers increasingly use for cross-platform work.
The performance context matters. Meta disclosed in 2025 that end-to-end AI campaign tools (Advantage+, Performance Goals, and now the Business Assistant) reached a $60 billion annual run rate, contributing materially to Meta’s 23.7 percent year-over-year ad revenue growth. The Business Assistant is the conversational interface on top of the deeper Advantage+ ML stack, not a separate product. The launch announcement frames the assistant as the natural-language entry point to capabilities the platform has been building for years.
Why it matters
- Small business advertisers just got a senior analyst. A solo founder running Meta ads without an agency can now ask the same questions a media buyer charging $4,000 per month would answer, and get usable responses in seconds. The structural economics of small-business advertising shift.
- Agencies have to reposition. The work agencies bill for that the assistant now does — performance analysis, optimization recommendations, troubleshooting — needs to move toward strategy, creative, and the work AI does not handle well.
- The reporting workflow changes. Weekly performance reviews shift from “build the dashboard” to “ask the assistant the questions the dashboard should have answered.” Speed of insight rises; depth of insight depends on how well the advertiser knows what to ask.
- Multilingual reach expands the addressable market. Brazilian, Mexican, Indian, German, and Japanese small businesses now have native-language access to a tool that previously required English fluency. The competitive pressure on local-language ad agencies in those markets rises.
- AI connectors signal a multi-platform agent future. Advertisers increasingly want one AI agent that drives campaigns across Meta, Google, TikTok, LinkedIn, and Amazon. The connector pattern Meta shipped is the substrate that makes that real.
- Meta’s competitive position improves on the most-used surface. Ads Manager is the surface advertisers spend the most time in. A capable AI inside it raises switching costs and increases lock-in versus competitors.
How to use it today
Adoption takes minutes if you already have an Ads Manager account. The first useful workflow is a guided diagnostic of one of your campaigns; that produces immediate evidence about whether the assistant is worth integrating into your weekly routine. The path below is the one we recommend for a first session.
- Open Ads Manager or Meta Business Suite and look for the new “AI Assistant” entry in the right-side panel. If you do not see it yet, the rollout is still propagating; check again over the next 24 to 72 hours.
- Pick a live campaign with measurable spend. The assistant produces better answers about real campaigns than about test or paused campaigns.
- Ask the diagnostic question. “Why is my cost per result up 32% this week?” The assistant pulls relevant signals and produces a structured answer with the top contributing factors.
- Ask for optimizations. “What three specific changes would you recommend to improve performance on this campaign over the next two weeks?” The assistant responds with prioritized recommendations and predicted impact ranges.
- Apply one or two recommendations. The assistant offers one-click application for many changes. Apply the ones you agree with; ignore the ones you do not.
- Set a follow-up window. The assistant offers to check back in a few days with results. Take the offer.
- Test a creative question. “Which of my creative variants is performing best for high-intent audiences?” The assistant interprets the question, breaks performance down by relevant segments, and produces a creative-level read.
For programmatic users who want to query Meta’s marketing API alongside the assistant’s behavior, the structured access pattern below works. The connector authentication uses the Meta Graph API with the new ai_connector_access scope.
import requests, os
META_TOKEN = os.environ["META_GRAPH_TOKEN"]
AD_ACCOUNT = os.environ["META_AD_ACCOUNT_ID"]
resp = requests.get(
f"https://graph.facebook.com/v22.0/act_{AD_ACCOUNT}/ai_assistant_query",
headers={"Authorization": f"Bearer {META_TOKEN}"},
params={
"question": "Which ad set produced the best ROAS in the last 14 days, and what is the next-best optimization?",
"context": "campaign_id=120211234567890",
"response_format": "structured",
},
timeout=30,
)
print(resp.json())
For agencies and enterprise advertisers who want their existing AI orchestration platforms (Claude, GPT-5.5, Gemini, or in-house agents) to drive Meta campaigns, the connector pattern looks like this. The advertiser authorizes the external platform through Meta Business Manager; the external platform then makes structured calls to the Meta connector endpoint.
{
"tool": "meta_ads_connector",
"operation": "create_campaign",
"params": {
"account_id": "act_123456789",
"objective": "OUTCOME_SALES",
"budget_amount": 50000,
"budget_currency": "USD",
"audience": {"saved_audience_id": "23861234567890"},
"creative_ids": ["6309876543210", "6309876543211"]
},
"approval_mode": "preview_only"
}
How it compares
Every major ad platform now ships some flavor of conversational AI inside the buyer-facing tool. The table below compares the leading offerings as of mid-May 2026.
| Platform | AI assistant name | Coverage | External connector | Strength |
|---|---|---|---|---|
| Meta Ads Manager / Business Suite | Meta AI Business Assistant | All advertisers globally as of May 2026 | Yes (new connector API) | Conversational performance Q&A + optimization |
| Google Ads | AI Max + Gemini-powered assistant | All advertisers | Limited; mostly Workspace integration | Search and Performance Max campaign workflows |
| TikTok Ads Manager | Smart Performance + AI Studio | All advertisers | Limited | Creative production at scale |
| LinkedIn Campaign Manager | AI Campaign Optimizer | Limited to certain markets | No | B2B audience targeting |
| Amazon Ads | Amazon Ads AI | Beta for select advertisers | Limited | Retail media and product targeting |
| Pinterest Ads | Pinterest Performance+ | All advertisers | No | Visual creative optimization |
The competitive read: Meta and Google now offer the most capable AI ad operations surfaces, with Meta marginally ahead on the conversational interface and Google ahead on the underlying ML stack for certain campaign types. The connector pattern is where Meta has opened a real lead; the others are not yet exposing the same external-AI-driven workflows. For cross-platform advertisers running custom agents, the practical answer in 2026 is Meta-first for the cross-platform agent stack, with Google integrations built next as Google opens its own connector surface.
What’s next
Three threads to watch over the next sixty days. First, Meta has signaled the next capability wave for the assistant will focus on campaign planning and creation, not just analysis and optimization. Expect the assistant to start drafting full campaign briefs (audience, budget, creative direction) within months, with the advertiser approving or editing rather than building from scratch. Second, expect Google to match the AI connector pattern, given the strategic importance of cross-platform agent compatibility. Third, expect early data on agency revenue impact; agencies that have not yet repositioned around strategy and creative are likely to feel the pressure first in the small-business and mid-market segments.
The longer arc is that the ad operations function is changing shape. The work that buyers and agencies bill for is shifting from media operations to creative strategy, brand-side measurement, and cross-channel orchestration. The advertisers who win are not the ones with the most expensive media-buying teams; they are the ones with the best creative engine, the cleanest measurement, and the most thoughtful brand strategy. Meta’s assistant accelerates the shift but does not cause it.
Frequently Asked Questions
Do I need a paid Meta Ads account to use the assistant?
Yes. The Meta AI Business Assistant lives inside Ads Manager and Meta Business Suite, both of which require an active ad account. The assistant itself is free; you pay for the ads you run, not for the AI that helps you run them.
Does the assistant make changes automatically, or does it ask first?
By default the assistant proposes changes and asks for one-click approval. You can configure autopilot mode for specific change types (budget reallocations within a single campaign, creative refreshes from approved libraries) if you want it to act without confirmation. Most advertisers should leave autopilot off until they have built trust with the assistant’s recommendations.
How does the assistant handle sensitive industry restrictions?
The assistant respects all the policy restrictions that already govern Meta ads. Health, finance, alcohol, gambling, political, and other regulated categories have additional constraints on what the assistant will recommend or apply. The assistant explicitly cites policy when it declines to act on a request.
Can I use the assistant in languages other than English?
Yes. The May 2026 global rollout added native support for Spanish, French, German, Portuguese, Mandarin, Japanese, Korean, Italian, Dutch, and several others. Quality is highest in English but materially production-ready in the other major business languages.
What data does Meta use to train the assistant?
Meta uses aggregated ad performance data across the platform to train the underlying models. Advertiser-specific data is used within the advertiser’s own session and account context, not for training the public model. Enterprise advertisers can request additional data-handling terms through their Meta account team.
Will this replace my media buyer or my agency?
Not entirely, but it does compress the work that media operations roles handle. The roles that survive and grow are the ones focused on creative strategy, brand-side measurement, cross-channel planning, and the judgment calls AI does not handle. Media buyers who define their role as “running the platform” face the most pressure; media buyers who define their role as “running the strategy” gain leverage.