Google And TikTok Just Put AI Agents Inside Their Ad Platforms. The Agency Middle Layer Is Being Automated Out

On 18 June 2026, Google launched "Ask Ad Manager" - a Gemini-powered agent built directly into Google Ad Manager. It troubleshoots underperforming campaigns in real time, generates custom performance reports from a single prompt, and configures the platform through plain conversation. The same month, TikTok shipped an agent layer that lets campaigns be created, managed, optimised and analysed without a human ever opening Ads Manager.

Both companies will frame these as productivity tools that help agencies do more. That framing is comfortable and incomplete. What these launches actually do is move the billable middle layer - the daily grind of ad operations - inside the platform itself, where the client can reach it directly.

The middle layer is where the money is

For a large share of marketing agencies, the monthly retainer is built on operational work: pulling reports, adjusting bids, flagging what is underperforming, setting up campaigns, optimising delivery. It is not glamorous, but it is the work that justifies the recurring fee. It is also exactly the work these agents now do on request.

When a client can ask a chatbot inside Ad Manager the same questions they used to ask their account manager - why is this campaign underperforming, build me this report, adjust this setup - the operational retainer starts to look like a charge for something the platform now includes. The middle layer is being automated out, and the middle layer is where most agencies make their margin.

This is a pattern, not a one-off

Google and TikTok shipping agents in the same month is not a coincidence. It is the direction of every major ad platform: build the AI assistant into the product, give the advertiser direct access, and remove the friction that an agency used to be paid to absorb. The connective tissue under all of it - the APIs and agent interfaces that let software run campaigns programmatically - is being laid down across the industry in 2026.

The strategic implication is blunt. Any agency value proposition that amounts to "we operate the platform for you" is on a clock, the same clock ticking under every account manager since Google quietly rewrote its Ads Terms to auto-authorise AI-generated campaign work. The platform now operates itself, on the client's command, for the price of the subscription they already pay.

What the agents cannot do

The work that survives is the work that sits above execution. An agent can pull a report. It cannot tell the client which number on that report actually matters for their business this quarter. An agent can adjust bids. It cannot decide whether the campaign should run at all, or what it should say, or how it fits the brand's bigger strategic bet.

• Strategy: what to do and why, before any campaign is built.
• Creative judgement: the idea and the message, which the platform optimises but does not originate.
• Interpretation: turning platform metrics into business meaning the client can act on.
• Accountability: owning the outcome, which a chatbot will never do.

Execution is being commoditised. Judgement and strategy are not. The agencies that survive will have moved their value - and their pricing - decisively from the first to the second.

What to do now

1. Audit your retainer for exposed work

Break your monthly retainer into its component activities and mark each one: can the platform's own agent now do this for the client directly? Be honest. The total of the "yes" column is the share of your fee that is now at risk.

2. Use the agents yourself

The agencies that win are not the ones that pretend the agents do not exist. They are the ones that use them to do the operational work in a fraction of the time, then reinvest those hours into strategy and judgement the client cannot get from the chatbot. Let the platform run the platform.

3. Reprice around the layer above execution

If your fee is justified by operations, rebuild it around the work the agent cannot do. Make strategy, creative, and interpretation the visible, paid core of the relationship, and treat execution as the cheap part it has become.

4. Get ahead of the client's question

Sooner or later a client will ask why they are paying you to do what the platform's AI now does for free, the same question every DSA agency is about to face when Google force-migrates the campaign type to AI Max by September. Have the answer ready before they ask it - and make sure the answer is a layer of value they genuinely cannot self-serve.

The bottom line

Google and TikTok did not hand agencies a better tool. They handed clients the middle layer. The operational work that filled the retainer is becoming a feature of the platform, available on demand. The agencies that treat this as a threat to defend against will lose. The ones that treat it as permission to move up the value chain - to sell the judgement and strategy the agents cannot touch - will come out of it with higher margins and stickier clients.