On 9 November 2026 Attentive, the SMS and email marketing platform used by retail and ecommerce brands, named co-founder Eric Miao as chief executive and made its Model Context Protocol (MCP) server generally available. Marketers can now connect ChatGPT or Claude to approved Attentive data and ask the assistant to pull reports, build segments and draft or edit campaigns.
- MCP is generally available from 9 November 2026
- Works with OpenAI's ChatGPT and Anthropic's Claude
- Approval controls, compliance and deliverability safeguards stay in place
Attentive announced the leadership change and the MCP release in a press release carried by Yahoo Finance, and CMO Tech covered it the next day. Miao has been with the company since the start, led its customer success team and has run AI product strategy for the past three years.
The CEO news matters less to most readers than the plumbing. MCP is an open protocol that lets an AI assistant call tools and read data inside another system. When a marketing platform ships an MCP server, the chat window starts to work as a second interface to your customer lists.
Not every team runs its customer data in Attentive, of course. If your contacts sit in a CRM, a database or a homegrown app, connecting ChatGPT or Claude to your own customer data is the same idea built on your systems, with the same need for permissions and audit logs.
What did Attentive announce?
Beyond the CEO change and the MCP release, Attentive used the 9 November announcement to pitch what it calls an agentic operating system. Under that model, marketers set goals, budgets, brand standards and approval rules, and the system makes and executes decisions inside those limits. Miao was careful to separate this from rebadged automation. "A lot of what's called agentic AI in marketing today is really just automation with a new name," he told CMO Tech. On the CRM side he argued that connecting customer intelligence "across the full customer lifecycle" is what would finally let CRM deliver one-to-one marketing.
CTO Andrew Sega put the technical bet in one line: "The more context AI has about a customer, the better decisions it can make."
What can the MCP server do?
With the server connected, a marketer can ask ChatGPT or Claude to look up customer intelligence, build an audience segment, or create and edit a campaign by prompt. Compliance and deliverability checks still apply, and changes go through the approval controls the brand has set.
Attentive named two early users:
- ReserveBar uses MCP to automate its weekly performance reporting.
- MintRX uses it to analyse historical campaign data, plan new campaigns and spot problems in existing journeys.
Both examples are reporting and analysis. That's the low-risk end, and probably where most teams should start too.
What results does Attentive claim?
Attentive quoted figures for its existing AI products. These are the vendor's own medians, not independent measurements:
- AI Grow: a median 25% increase in email and SMS sign-ups and 35% more welcome-series revenue.
- AI Pro and AI Journeys: a 20% lift in segmented campaign revenue and a 200% lift in journeys revenue.
Treat them as a reason to run your own test, not as a forecast.
What does it mean for small and mid-size teams?
- The chat window becomes a front end. After Zoho put AI agents into Zoho Campaigns, this is another sign that vendors expect marketers to work through an assistant rather than a dashboard.
- Segments get cheaper to build, and easier to get wrong. A prompt can produce a segment in seconds. If you can't explain who is in it, you can't trust the result. Our primer on segment marketing covers the rules worth writing down first.
- Permissions decide the risk. An assistant that can edit a live SMS campaign can also send the wrong one. Read-only access for reporting is a very different decision from write access.
- Your data has to be joined up. An assistant only knows what the connected system knows. If orders, support tickets and CRM records live apart, see our CRM integration guide before expecting good answers.
What should you do this week?
- Start read-only. If you use Attentive, connect the assistant for reporting first, the way ReserveBar does, and keep campaign edits behind approval.
- Write a prompt log. Save the prompts that produce segments you actually send to, so results can be repeated and checked.
- Compare one report. Ask the assistant for last month's campaign summary and check it against the platform's own numbers before you rely on it.
- Ask your other vendors. Check whether your CRM and email tools offer an MCP server or an API an assistant can use, and what permissions it exposes.
