Klaviyo’s AI Marketing Composer Agent is a feature that generates complete email and SMS marketing campaigns from plain language descriptions, automatically determining audience segments, messaging for each channel, and send timing. Launched into public beta on June 30, 2026, this agent moves beyond simple template suggestions to create end-to-end campaign structures that reflect marketing best practices learned from 14+ years of data across Klaviyo’s customer base. A marketing manager at an e-commerce brand, for example, could describe a campaign goal in conversational language—”I want to win back customers who haven’t purchased in three months with a discount offer”—and the agent would generate the audience segment definition, craft channel-specific messaging, and propose an optimal send schedule, all within minutes.
What distinguishes this tool from existing marketing automation features is that it combines customer data, behavioral patterns, and campaign performance history into recommendations grounded in billions of real consumer interactions. The agent doesn’t work in isolation; it’s part of a broader ecosystem where a companion Customer Service Agent handles support inquiries, and both agents share real-time customer profile data, allowing insights from one workflow to enhance the other. All generated campaigns require human review and approval before sending—there is no autonomous deployment—which preserves marketing control while reducing the manual work of campaign setup.
Table of Contents
- How Does the Composer Agent Generate Complete Campaigns from Plain Language?
- Understanding the Data Foundation Behind Composer Recommendations
- How the Composer Agent Integrates with Klaviyo’s Customer Service Agent Ecosystem
- Getting Started with Composer Credits and Pilot Testing
- Approval Workflows and the Role of Human Review in Campaign Launches
- Channel Support and the Future Roadmap
- Real-World Performance Data from Private Beta Partners
How Does the Composer Agent Generate Complete Campaigns from Plain Language?
The Composer Agent converts conversational marketing briefs into structured campaign workflows through a process grounded in 14+ years of aggregated customer data and billions of tracked consumer interactions across Klaviyo’s 193,000 to 196,000 paying customers. When you describe a campaign objective, the agent analyzes that request against historical patterns to recommend audience segments, message variations optimized for different customer behaviors, and channel sequencing. For a re-engagement campaign, this might mean identifying inactive customers based on purchase history, suggesting a tiered message approach where newer inactive customers receive different messaging than long-term inactive customers, and recommending staggered send times that maximize open rates for each segment. Currently, the agent supports email and SMS channels, with mobile push and WhatsApp integration roadmapped for future release. The platform doesn’t generate copy in a vacuum; it draws on performance data specific to e-commerce and subscription businesses to inform recommendations about tone, urgency, and offer structure.
A fashion brand running a seasonal clearance campaign would receive suggestions based on patterns from similar clearance cycles, not generic templates. New customers can test the agent’s capabilities with 10,000 complimentary Composer credits, available for up to 90 days, which provides meaningful room to experiment before committing to paid usage. One important limitation to recognize: the agent generates campaign structures and recommendations, not guaranteed revenue increases. A well-structured campaign by the agent still depends on accurate audience data, compelling product offerings, and appropriate offer timing for your specific business. Small businesses with incomplete customer data or atypical revenue models may see less tailored recommendations than brands with deep historical customer behavior records.
Understanding the Data Foundation Behind Composer Recommendations
klaviyo‘s Composer Agent operates on a foundation of 14+ years of marketing intelligence accumulated from consumer interactions across its customer base. This isn’t general-purpose AI trained on publicly available marketing content; it’s an agent specifically trained on patterns from successful and unsuccessful campaigns within the e-commerce, subscription, and direct-to-consumer space. The agent analyzes billions of consumer interactions—email opens, click-through behavior, purchase attribution, and post-purchase engagement—to identify which campaign structures, message timings, and segment compositions drive the strongest outcomes for different business scenarios. The depth of this data advantage means the agent can recognize subtle patterns that individual brands might miss. For instance, it might identify that for beauty brands specifically, weekend sending outperforms weekday sending for certain audience segments, or that post-purchase follow-up campaigns perform differently depending on whether the original purchase was a first-time or repeat transaction.
However, this data aggregation also means the agent’s recommendations are skewed toward businesses similar to Klaviyo’s existing customer base. If you operate a niche business model with unusual customer behavior patterns, the agent’s suggestions may be less precise because it lacks sufficient comparable reference data. The agent requires human judgment to validate that its recommendations align with your business context. A recommendation to target customers who haven’t purchased in 90 days might be appropriate for a fast-moving-goods retailer but inappropriate for a luxury furniture brand where purchase cycles naturally span 18+ months. Marketing teams need to maintain oversight over what the agent recommends, not treat its output as automatically correct simply because it’s data-backed.
How the Composer Agent Integrates with Klaviyo’s Customer Service Agent Ecosystem
Klaviyo launched its Marketing Composer Agent alongside a Customer Service Agent, and the two systems are designed to work in concert within your customer communication infrastructure. The Customer Agent handles inbound customer service inquiries and has achieved significant traction in private beta: early adopter Naked Wardrobe reached 86% autonomous resolution of customer questions within 90 days of implementation, while NANUK achieved a 100% increase in resolved product queries within the same timeframe. Across Klaviyo’s testing, the Customer Agent resolves approximately 65% of customer inquiries automatically, without human intervention. What makes this ecosystem approach valuable is that both agents operate from the same real-time customer profile database.
When the Customer Agent resolves a support inquiry—say, processing a return or answering a question about product availability—that interaction updates the customer’s profile with new behavioral and preference data. The Composer Agent can then access that updated profile when generating new marketing campaigns, ensuring that the marketing outreach reflects recent customer service interactions. Similarly, when Composer sends a campaign and customers engage with it, that engagement data flows back into the customer profile, informing how the Customer Agent prioritizes or personalizes future support interactions. This integration reduces siloing between marketing and customer service, but it also means teams need to establish clear policies about how these systems interact. If the Customer Agent handles a complaint and offers a discount, should Composer avoid sending a separate discount offer the same week? These operational questions fall to human teams to define, while the agents handle execution within those guardrails.
Getting Started with Composer Credits and Pilot Testing
New customers to Klaviyo can access the Composer Agent immediately with 10,000 complimentary credits available for up to 90 days from their account activation. This allowance is substantial enough to run multiple campaign experiments—testing different audience segments, comparing email-only versus email-plus-SMS approaches, and refining how you brief the agent to generate campaigns closer to your intended outcomes. For a mid-market brand, 10,000 credits might cover a full quarter of marketing campaign generation, providing genuine operational value during the evaluation period.
Understanding your credit consumption rate is important for budget planning beyond the trial period. The cost structure isn’t detailed in public materials, but typically credits are consumed based on campaign complexity, the volume of audience segments being generated, and the number of channels being composed for. A simple single-segment email campaign consumes fewer credits than a complex multi-segment email-plus-SMS sequence. Testing different briefing approaches with the agent during your pilot—comparing detailed descriptions to brief summaries, requesting specific segment definitions versus letting the agent suggest segments—helps you understand which usage patterns align with your workflow and budget tolerance.
Approval Workflows and the Role of Human Review in Campaign Launches
A core operational requirement of the Composer Agent is that all generated campaigns require human approval before launch; the system contains no autonomous sending capability. When the agent generates a campaign, it produces a draft that appears in your Klaviyo interface for review. Your team reviews the audience definition, message content, send timing, and any other campaign parameters before deciding to approve and schedule the campaign, modify it before approval, or reject it and brief the agent to try again. This approval requirement protects against several real risks: audience definitions that accidentally target the wrong customer segment, messaging that doesn’t align with brand voice or current business offers, send timing that conflicts with other scheduled communications, or campaigns that violate regulatory compliance for your industry.
A marketing manager reviewing a generated campaign might notice that the agent included a segment definition that would email customers who recently requested to reduce email frequency, which violates customer preference signals. The approval step catches this before it damages customer relationships. One downside of the approval-required model is that it reduces the time savings the agent provides if your review process is slow or inconsistent. If campaigns sit in draft status for days waiting for sign-off, or if reviewers frequently reject and re-prompt the agent, the operational efficiency gains diminish. The biggest value emerges when teams establish clear approval criteria and review cadences, treating the agent more as a collaborative tool requiring structured handoff rather than a fully autonomous system.
Channel Support and the Future Roadmap
Currently, Klaviyo’s Composer Agent generates campaigns for email and SMS channels. This combination covers the primary direct communication channels for most e-commerce and subscription businesses, allowing the agent to create cohesive multi-channel sequences where email might handle longer-form storytelling and SMS handles time-sensitive urgency or transactional notifications. However, this channel limitation means the agent cannot currently generate campaigns that incorporate mobile push notifications or WhatsApp, both of which are on the public roadmap for future release.
The planned expansion to mobile push and WhatsApp is significant because these channels enable different messaging cadences and contexts. Push notifications can reach customers with higher frequency and shorter messaging without creating email fatigue, while WhatsApp operates in a permission-based, conversational context. Brands that rely heavily on push or WhatsApp for customer communication will not be able to use Composer for those channels until the roadmap delivers them. This gap may influence early-adopter strategy; some brands might focus their initial Composer pilots on email and SMS workflows while waiting for the broader channel set to mature.
Real-World Performance Data from Private Beta Partners
Klaviyo’s private beta for the Composer Agent included established consumer brands such as AS Beauty, SPANX, and Dermalogica. These partners worked with Klaviyo’s team to refine the agent’s recommendations and validate that the generated campaigns performed as predicted across diverse product categories and customer bases. The presence of recognizable brands in the beta suggests the agent was stress-tested against complex, real-world scenarios rather than just simple template campaigns.
The performance results available from the companion Customer Service Agent provide context for what’s possible in the integrated agent ecosystem. Naked Wardrobe, an early Customer Agent adopter, achieved 86% autonomous resolution of customer inquiries within 90 days—meaning the agent handled the majority of incoming support volume without human escalation. NANUK’s experience showed a 100% increase in the volume of resolved product queries within the same 90-day window, indicating the agent not only handled inquiries more completely but also processed higher volume. While these metrics reflect customer service rather than marketing campaign performance directly, they demonstrate the capability and reliability of Klaviyo’s underlying agent technology, which underpins both systems.
