AgentMark began as a startup helping small brands but evolved into a no-code AI platform that helps marketers and publishers automate operations and scale their strategies with their AI tool, AgentMark.
Wentao Xiao, co-founder and CEO of AgentMark formally known as Cactivate, entered the world of entrepreneurship to solve real problems for small businesses, in partnership with his two co-founders, Zhe Lu, his CTO, and Peng Wu, his Chief Scientist.
So he built Cactivate to help small brands get the most out of their customer acquisition campaigns. But the company has since evolved into an AI-powered campaign optimization solution for small businesses struggling to navigate digital marketing without incurring significant costs.
In 2021, well before ChatGPT became a household name, Xiao and the AgentMark team were already building natural language models to help automate the messy, expensive work of launching and optimizing ad campaigns.
The idea worked. AgentMark grew to serve 200 customers and reached half a million dollars in annual recurring revenue from 2021 to 2024. But the deeper Xiao and his team dug into the market, the clearer it became: Small businesses didn’t just need automation; they needed support.
Rather than becoming an agency itself, however, AgentMark made a sharp turn into software as a service. Now the tool helps brands and publishers scale their marketing operations through custom AI agents.
AgentMark is betting that the next phase of digital marketing won’t be a battle between humans and machines, but a collaboration between the two. “The future is going to be marketing strategists plus AI agents,” Xiao said.
AI agents are especially suited to automating the many small campaign adjustments made by ad ops teams on both the buy side and sell side, according to Xiao. “Button-pushing shouldn’t require hiring full-time people to do that,” he said.
Building the Backbone of AgentMark
Cactivate’s hands-on experience with AI infrastructure and the real-world needs of marketers and ad ops teams shaped AgentMark.
To build the solution, AgentMark trained natural language processing models, developed custom tools for image analysis and built systems to predict ad performance.
Rather than surfacing suggestions in response to user prompts like a simple AI chatbot, they designed AgentMark to execute real marketing operations—from quality assurance (QA) and pacing to creative swaps and reporting. It’s a no-code AI agent builder tailored specifically for paid advertising.
“We see ourselves as the plumbers,” Xiao said. “We’ve built the pipes between agencies and AI platforms so you can automate repetitive work and focus on strategy.”
To train AgentMark, Xiao and his team took a hands-on approach by using the platform to service clients. “We pretty much just dogfooded it,” he said, referring to the practice of a company using its product internally to test it in real-world situations.
In testing the solution, the team noticed that while most ad ops teams perform similar tasks—such as reporting, QA and optimizations—each team carries out these tasks with its own style and nuance. That insight shaped one of AgentMark’s core principles: customizability. Instead of a rigid, black-box system, the platform functions as a white-box tool that can be tailored to fit each agency’s specific workflows, Xiao said.
With that flexibility in mind, a three-person AgentMark team was able to support 200 clients and automate the bulk of their back-office work, Xiao said. But true automation required more than just task execution.
“We had to train the system to understand how marketers think,” Xiao explained. That meant teaching it how different platforms define key metrics, like recognizing that “conversion over cost” in Google Ads corresponds to “ROAS” in Meta or that calculating average CTR requires weighting by spend, not just averaging percentages.
That level of marketing-specific intelligence is what Xiao sees as AgentMark’s biggest differentiator: a platform trained to act like a sharp, reliable junior employee—minus the overhead.
The Push for Scalable Efficiency
Agentic AI tools “might feel like an unnecessary cost” for smaller brands or startups, Xiao said. “But once you’ve hit repeatable processes, that’s when automation becomes essential.”
He added, “We work best with agencies, publishers and large brands managing 25 or more accounts or highly complex campaigns.”
AgentMark’s main value proposition—allowing users to build AI agents without hiring developers—has resonated particularly with publishers, according to Xiao.
While Xiao didn’t have permission to name any of AgentMark’s current publisher clients, he described a typical onboarding: “Let’s say you’re a publisher managing dozens of campaigns and tracking a wide set of KPIs. We’ll conduct two short calls, build a custom budget pacing agent that pulls data from platforms like Google, Meta, TikTok and Bing, and map that against your source of truth, whether it’s a media plan, GA4 or another platform. The agent runs daily checks and flags issues automatically.”
What starts as a simple alert system quickly opens doors to creative rotation, smart budget reallocation, automated client reporting and even generating reports in a publisher’s preferred tone and format.
“Suddenly,” Xiao said, “you’re not spending your day checking dashboards. You’re spending it making decisions.”
Adaptation and Access
As AI becomes more integrated into marketing workflows, Xiao sees a shift already underway: Agencies, publishers and marketers are moving from hourly or margin-based models to project- and outcome-based pricing.
“That naturally pushes people to look for ways to improve efficiency and expand margins,” he said.
Xiao compared the trend to the rise of Squarespace—it didn’t replace web developers, but it enabled more people to participate in web development. In the same way, Xiao believes AI tools like AgentMark won’t replace marketers, but they will empower a wider range of players to enter the field, especially small businesses and lean teams.
Looking ahead, Xiao’s road map for AgentMark centers on two main areas: more integrations and smarter delivery. The platform already connects with Google, Meta, TikTok, Bing and Google Workspace, but plans are in motion to expand to platforms like The Trade Desk, LinkedIn, Klaviyo and Shopify. The goal is to enable marketers to automate end-to-end workflows, from media planning to campaign optimization and creative generation.
Making AgentMark accessible across a wider range of workflow touch points is also a major priority for the team. Xiao said he envisions a future where AgentMark operates more like a team member, one you can CC on an email, tag in Slack or call on-demand to generate insights, answer client questions or even draft responses in real time.
“That shift could cut response times from days to minutes,” he says, “freeing up marketers to scale their work and focus on more strategic, creative tasks.”