At Bloomreach, we’ve long been pioneers in AI-driven marketing automation. With the latest advancements in artificial intelligence, we’re pushing the boundaries even further. We’ve already brought some of these innovations to life, but I’m excited to share what our team has been diligently working on: campaign agents, which are built using our powerful Loomi AI platform.
By leveraging advanced AI capabilities, we’ve transformed how Loomi AI can empower marketers to fully streamline and optimize their marketing automation processes — from ideation to launch. We can now harness the power of AI to create marketing automation that achieves better conversions and clickthrough rates that can be deployed in a matter of minutes.
Setting the Stage: An Overview of Marketing Automation
Before diving into our developments, it’s important to understand the two primary types of marketing automation:
Batch Campaigns
Also known as one-off campaigns, these are standard marketing blasts sent at a scheduled time to a broad audience, regardless of individual customer behavior. They’re typically used for wide-reaching announcements, such as the start of a seasonal sale or a promotional offer. These campaigns are straightforward to build, often resembling something like this:
Because these campaigns have fewer triggered conditions based on personalized experiences on-site, they often drive less engagement. At the same time, most marketers can’t do without them due to their wide reach, ability to share a timely message, and the significant revenue they can drive.
Event-Triggered Scenarios
Referred to as automated or triggered scenarios, these campaigns are initiated by specific customer actions or events (e.g., cart abandonment or a first purchase). And, while they often start as simple scenarios, experienced marketers know that to really maximize their value, it’s best to personalize them based on customer segments and layer on multiple communication touchpoints across available channels (e.g., email, SMS, and push notifications).
Highly personalized and data-driven, these advanced scenarios nurture customer relationships by responding to individual behaviors. However, this personalization introduces complexity, resulting in scenarios that can look like this:
These campaigns require intricate logic, conditional statements, and filters — essentially resembling code. Building event-triggered scenarios demands more time and energy from marketers. This is where our advancements with Loomi AI come into play.
How We’re Evolving Marketing Automation With Campaign Agents
Developers across the industry have harnessed large language models to generate code and dramatically improve productivity. Recently, when I needed to create a simple Chrome plugin, instead of revisiting JavaScript and learning about Chrome manifest files, I simply described the desired functionality to ChatGPT. Within seconds, it generated the code for a plugin — and it worked flawlessly.
Building marketing automation scenarios shares similarities with programming. Traditionally, developers wrote code to build these systems. Even a simple abandoned cart email functions like a program. With Loomi AI’s marketing automation capabilities, we’ve simplified this process with our low-code scenario builder featuring Jinja support. Yet, constructing complex scenarios still requires significant thought and testing.
With campaign agents, we’re now leveraging Loomi AI’s extensive campaign and scenario training data to enable AI to craft sophisticated marketing programs. Essentially, we’re implementing AI coding and software development agents within our platform to deliver optimal marketing campaigns.
Here’s how the multi-agent architecture works. Marketers interact directly with a campaign briefing agent through a conversational interface. This agent works with marketers to define the best path forward, suggest proven use cases, and then delegate work to a team of purpose-built subagents.
Each of these agents handles a specialized task, including audience segmentation, content generation, recommendation strategy, channel selection, timing optimization, and more. A central orchestrator coordinates all these subagents to align with the broader goal. The agents can also learn from brand templates, using components (e.g., headers, footers, hero blocks, recommendation blocks, etc.) to create new on-brand emails without the need for raw HTML generation.
Importantly, each of these subagents has access to the entire feature set of our marketing automation solution, and can use deterministic platform capabilities (segments, Jinja, wait conditions, product recommendations, etc.) as building blocks to assemble campaigns.
Additionally, self-optimizing agents use reinforcement learning to continually improve and optimize future campaigns.
How This Transforms the Marketing Campaign Process
What does this mean for marketers? In short, incredible efficiency.
Previously, marketers faced a myriad of questions and decisions for each campaign: defining the audience, setting trigger conditions, creating content variants, and more — all before actually building a scenario. With campaign agents, campaigns that used to take hours (or even days) to launch can now go live in minutes.
Our customers are already seeing impressive results with campaign agents — Sideshow was able to go from idea to launch in 15 minutes and drove a 2x increase in value per email. Meanwhile, Revolution Beauty saw a 5x increase in revenue per email, with over 5% of total email revenue being driven by campaign agents.
Get Started With the Future of Marketing
With campaign agents, marketers now get access to autonomous marketing capabilities that greatly increase efficiency and effectiveness. And, this has been validated across a wide range of pilot customers. Learn more about campaign agents and start embracing the future of autonomous marketing.
