Email Is the Signal Agents Will Weigh with Ose Amiegheme of Intuit Mailchimp
Brent sat down with Ose Amiegheme, who leads email product at Intuit Mailchimp, and the conversation started with the oldest line in marketing, that email is dead. Ose has spent his career building products for creatives and marketers, first on Adobe Express, then at TikTok, and now at Mailchimp, and he came to the call with a clear case for why email matters more now than it did a few years ago.
Why Email Keeps Living
Brent called “email is dead” the old moniker and asked for Ose’s view of email today. Ose started with how old the technology is.
“Email is like a 40-year-old technology, and most of the technology, including like our inboxes, are like 20 years old. So everyone’s been trying to kill email for a while and it still lives on,” he says.
He pointed to two shifts in the market. Acquisition is getting more expensive, with the cost of paid social climbing, and search is changing as people turn to LLMs for answers.
“it’s dying with LLMs being like the way people find information now,” he says of SEO.
That leaves merchants leaning on channels they own, and Ose shared what Mailchimp sees in its numbers.
“people who do this actually get like 10 to 20% more ROI,” he says.
Nurture the List You Already Have
Brent asked about the nurture phase, since merchants tend to forget the backlog of customers they already have. Ose said every store sends a welcome offer, and then the list gets the same message again and again. The cost shows up in deliverability.
“people don’t click your emails, your deliverability score tanks. And people are like, yeah, email doesn’t work,” he says.
His fix starts with segmenting by who has bought, who has opened and who has gone quiet. It then moves to predictive modeling, which he notes is “not even Gen AI,” to learn which customers go for the more expensive items and which respond to a discount. Mailchimp builds some of those segments from a merchant’s own data.
Agents Will Weigh Engagement
Brent asked how agentic commerce interacts with email. Ose thinks agents buying on a person’s behalf are still a way off, because people do not yet trust AI with financial decisions. He sees recommendations arriving first, and he described how an agent that knows Brent could weigh what Brent has engaged with before against the reviews it finds on Reddit and Trustpilot.
He gave a hypothetical of how an agent might score a brand, with reviews at 40% of the decision, engagement with the brand at 30% and price at 30%.
“engagement now becomes a premium in what these agents would care about as they choose to make buying decisions in the future,” he says.
That is why he sees owned channels carrying more weight over the next 12 to 24 months. His advice for merchants who ask how to get an LLM to recommend their product is to treat it as a retention play as well as an acquisition play.
“email is gonna play like a really strong role in how recommendations come up in the future,” he says.
AI Raises the Floor
Brent raised the worry that people hand too much to an agent and lose track of what they want. Ose described an AI trust spectrum, where people trust AI to handle routine work and keep the biggest decisions for themselves. He then took on the argument that AI pulls everyone toward the average, using a bakery owner who started the business for the recipes and not for the marketing.
For that owner, moving toward the average is a gain, because it lifts the people who sit below it. For the people who are already strong, the work is guiding the model past its first answer.
“Human intelligence is what actually gets us to the next level,” he says.
The Analytics Agent and Insights to Action
Brent asked about the predictive work at Mailchimp, and Ose described an analytics agent that answers the question every marketer asks a data team, which is what a number means. The next phase he calls insights to action, where an insight such as a 31% open rate and a 5% click rate on a segment turns into a new segment and a campaign ready to send. He sees the same data-driven edge that large enterprises have had becoming available to small businesses.
Meeting People Where They Are
Brent asked about AI inside a tool compared with AI attached from outside. Ose called the difference building with AI versus building AI into the tool, and he described Mailchimp’s connector with ChatGPT and Anthropic, launched earlier this year. The reasoning is that most businesses already work in an LLM, and an LLM with a merchant’s own data returns advice that differs from what a competitor down the street receives.
“our philosophy now is meeting people where they are,” he says.
Where to Find Ose
Ose invited listeners to try Intuit Mailchimp, which he says has moved well beyond the email provider some people remember. For his own plug he added a request that listeners watch more soccer. He is most active on LinkedIn at linkedin.com/in/ose-amiegheme, and Intuit Mailchimp is at mailchimp.com.