Ads, AEO, CRM Agents, and Agent Infrastructure
OpenAI is expanding ChatGPT ads with beta self-serve buying, CPC bidding, and measurement while continuing to separate sponsored content from organic answers.
- Owner
- RevOps or GTM systems owner
- Workflow
- Buyer-question capture, AEO review, CRM approval, or agent-enabled GTM task
- Review
- Ads, page updates, CRM changes, external messages, and workflow outputs
- Metric
- Qualified intent, page improvement, field completeness, time saved, and accepted output rate
1. ChatGPT Ads Become A Buyer-Question Surface
Source: OpenAI, "New ways to buy ChatGPT ads" and "Testing ads in ChatGPT."
What changed:
OpenAI is expanding ChatGPT ads with beta self-serve buying, CPC bidding, and measurement while continuing to separate sponsored content from organic answers.
GTM use case:
Demand gen teams should treat conversational ads as buyer-question capture, not just another display placement. The useful unit is the decision conversation: "what should I buy?", "which tool fits this use case?", or "how do I solve this problem?"
Workflow:
- Pick one decision-stage buyer question.
- Write the answer a buyer would expect before seeing an ad.
- Draft ad copy that extends the conversation rather than interrupting it.
- Build a landing page section that continues the same question.
- Track click, signup, demo request, and downstream quality.
Tools needed:
- ChatGPT ads access or waitlist.
- Landing page builder.
- Conversion tracking.
Risk:
Early ad products change quickly. Do not move core search or paid social budget until one narrow test shows qualified intent.
Test this week:
Create a one-page campaign brief for a single decision question.
2. AEO Moves From SEO Theory To GTM Operating System
Source: HubSpot Spring 2026 Spotlight.
What changed:
HubSpot announced AEO tooling for tracking and improving how companies appear in answer engines like ChatGPT, Gemini, and Perplexity. The same update tied AI output quality to business context inside the CRM.
GTM use case:
Demand gen teams need an answer-engine workflow that connects buyer questions, competitive mentions, cited sources, and page updates.
Workflow:
- List five high-intent buyer questions.
- Capture the answer engine output for each.
- Record cited sources and competitor mentions.
- Compare the output to your current site copy.
- Add missing proof, use-case language, and limitations.
- Recheck monthly.
Tools needed:
- Answer engine checks.
- Spreadsheet.
- CMS.
Risk:
Teams can optimize for prompts that buyers do not actually use. Tie prompt selection to sales calls, chat logs, support tickets, and CRM notes.
Test this week:
Run the AEO check for three buying questions and update one page.
3. CRM AI Should Start With Rep Approval
Source: HubSpot Spring 2026 Spotlight.
What changed:
HubSpot announced Smart Deal Progression, which analyzes call transcripts with deal history to suggest CRM updates, follow-ups, and action items.
GTM use case:
RevOps should start with approved suggestions, not autonomous deal movement. The high-value workflow is reducing rep admin while keeping the rep accountable for final CRM state.
Workflow:
- Pick one stage of the pipeline.
- Collect transcript, previous notes, stage definitions, and required fields.
- Generate suggested CRM updates and follow-up email.
- Require rep approval.
- Track time saved and field completeness.
Tools needed:
- CRM.
- Call recording or transcript.
- Stage definitions.
Risk:
Bad history produces confident but wrong updates. Make uncertainty visible and keep approval in the workflow.
Test this week:
Run five calls through a transcript-to-CRM-update prompt and compare against rep-written updates.
4. Agent Infrastructure Matters More Than Model Demos
Source: Google I/O 2026 developer highlights and Anthropic's Stainless acquisition.
What changed:
Google announced Managed Agents with persistent isolated environments. Anthropic's Stainless acquisition reinforces a similar theme: agents become more useful when they can reliably connect to APIs, SDKs, CLIs, and MCP servers.
GTM use case:
GTM teams should prioritize workflows with clear tool access and state: research a target list, update a spreadsheet, generate a call plan, draft a follow-up, or build a campaign QA checklist.
Workflow:
- Pick one recurring GTM task.
- List every tool, file, and approval it touches.
- Identify the state the agent must preserve.
- Add a human checkpoint before external output.
- Run the workflow manually three times before automating.
Tools needed:
- Source system access.
- File or CRM state.
- Approval step.
Risk:
Agent workflows fail when tools are unavailable or permissions are vague. Build around narrow, auditable tasks first.
Test this week:
Document one repeatable workflow with inputs, tools, state, approval, and output.
Operator Checklist
| Field | Decision |
|---|---|
| Owner | RevOps or GTM systems owner |
| Workflow | Buyer-question capture, AEO review, CRM approval, or agent-enabled GTM task |
| Inputs | Buyer questions, cited sources, transcripts, CRM fields, tools, state, and approval rules |
| Human review | Ads, page updates, CRM changes, external messages, and workflow outputs |
| Metric | Qualified intent, page improvement, field completeness, time saved, and accepted output rate |
| First test | Run one narrow workflow for one week with inputs, tools, state, approval, and output, then ship, stop, or expand |