What Klaviyo Composer and K:AI Actually Do (vs. Marketing Claims)

Composer is Klaviyo's flagship AI marketing agent: describe a goal in plain language and it analyses your account, surfaces revenue opportunities, and builds full multi-channel campaigns, all grounded in your customer data. Introduced in March 2026 and moved into public beta on 30 June 2026, Composer sits inside Klaviyo AI (K:AI), the umbrella name for every AI capability built into the Klaviyo B2C CRM, alongside Customer Agent, which resolves customer service queries without human involvement. The platform uses OpenAI and AWS Bedrock for content-creation tools, and proprietary in-house models for its predictive features.

Most AI marketing announcements deserve a fair amount of scepticism. Big claims, glossy demos, then the real product arrives and half the features are in beta or locked behind enterprise tiers. We spend a lot of time inside Klaviyo as a Klaviyo Master Platinum Partner, so we wanted to write an honest account of what Composer and the wider K:AI suite actually do, feature by feature, and where the marketing language gets ahead of the reality. No hype. Just what's there.

Composer, K:AI, Marketing Agent: Sorting Out the Names

The naming has moved around, so it's worth pinning down before anything else. Klaviyo AI (K:AI) is the collective name for every AI-powered capability inside the Klaviyo platform, spanning three distinct modes: assistive AI (tools that help you do tasks faster), predictive AI (models that forecast customer behaviour), and autonomous AI (agents that act on your behalf).

Composer is the flagship agent within that suite. If you used Klaviyo in 2025, you'll remember K:AI Marketing Agent, the autonomous agent that built campaigns from a website URL. Composer builds on that work as a new conversational, agentic experience, launched in March 2026 alongside more than 75 other platform features, and it entered public beta on 30 June 2026, so it's now sitting in every Klaviyo account with nothing to install or activate. Marketing Agent still exists as its own product during the beta period, but Composer is clearly where Klaviyo's positioning, and its search advertising, is pointed. If you Google "K:AI" today you'll likely land on the Composer page, which is Klaviyo directing attention rather than a rename.

The three-mode distinction is worth holding onto. Assistive AI gives you a better first draft. Predictive AI tells you which customers are about to churn. Autonomous AI actually goes and does something without waiting for you to click a button. Most platforms are still mostly assistive. Klaviyo is pushing into autonomous territory with Composer and Customer Agent, which is where the more significant claims live.

Klaviyo was named a Leader in the IDC MarketScape: AI-Enabled Marketing Platforms for Small Businesses 2025 Vendor Assessment, which gives the platform a credible third-party anchor for its positioning. That said, "Leader" in an IDC assessment covers a wide range of maturity levels. It tells you Klaviyo is a serious contender. It doesn't tell you every feature works perfectly out of the box.

The platform sits in a B2C CRM category rather than a pure ESP category. That framing matters for how Composer is built. The AI tools pull from unified customer profiles containing purchase history, browsing behaviour, email engagement, and transactional data. Without that data depth, most of the predictive and autonomous features wouldn't function. It's the data foundation that makes the AI credible, not the AI layer itself.

Klaviyo's customer base grew to over 193,000 paying customers by the end of fiscal year 2025, up from 167,000 the prior year. That scale feeds the AI models constantly with real behavioural signal across retail, D2C, and ecommerce verticals, and Klaviyo says Composer's outputs are grounded in over 14 years of marketing performance data across that base.

Composer: Full Campaigns from a Single Prompt (Now in Public Beta)

Composer is Klaviyo's autonomous AI marketing agent: a marketer describes a goal in plain language and Composer builds the audience, copy, channel selection, and send timing using your customer data and brand voice, turning hours of campaign setup into minutes.

That's the claim. And it's a meaningful one if it holds up in practice. The prompt-to-campaign model is the right direction for a B2C CRM tool serving brands that run dozens of sends a month. The bottleneck for most email marketing teams isn't ideas, it's execution speed.

What Composer Actually Generates

Describe a goal ("re-engage lapsed buyers who haven't purchased in 90 days") and Composer maps out the campaign structure: which segments to target, what copy approach to use, which channels to activate, and when to send. It's not just a subject line generator. The output is a launch-ready draft with executable components, handed back to you in Klaviyo's standard editor, fully editable.

Composer also works in the other direction, and this is arguably the more interesting half of the product. Rather than waiting for a prompt, it continuously analyses your campaigns, flows, segments, and forms against historical performance, surfaces the opportunities with the greatest revenue potential, and recommends what to do next. It can audit an existing flow you're looking at and suggest fixes. That shifts it from a content generator into something closer to an always-on strategist reviewing your account.

Two practical details the launch coverage tends to gloss over. First, channel coverage: Composer currently works across email, SMS, and MMS, with mobile push and WhatsApp still on the roadmap. If your programme is push-heavy, the "full omnichannel campaign from one prompt" promise isn't quite there yet. Second, Klaviyo's own documentation notes that content generation is weaker on image-heavy or fully image-based templates, so brands running visually dense creative should review those drafts particularly closely.

Nothing Goes Live Without You

One thing the marketing pages actually undersell: Composer does not send or publish anything automatically. Everything it creates is staged for your review inside Klaviyo. You approve it, tweak it, or throw it away. Brand voice, tone, and compliance requirements are set once and applied to every draft it produces. For anyone nervous about handing campaign execution to an agent, that human-in-the-loop design is the reassurance, and it's genuinely how the product works rather than a marketing caveat.

Worth knowing on cost: Composer runs on a credit system. Klaviyo customers get free credits to start (new customers receive 10,000 complimentary credits, available for up to 90 days), with different actions consuming different amounts. Support and general questions are free; generation and analysis tasks draw down credits. Once the beta terms settle, credits will be a real line item for high-volume users, so factor it into any tooling comparison.

The quality of what Composer produces depends heavily on the quality of your customer data inside Klaviyo. Thin profiles produce generic outputs. Rich profiles with product purchase history, channel affinity data, and browsing behaviour produce something much more specific. This isn't a criticism of Composer specifically. It's true of every AI tool that draws on CRM data.

Email AI, Segments AI, and Forms AI

Composer sits at the top of K:AI's assistive and autonomous content stack. Beneath it are three supporting tools worth knowing about separately.

• Email AI generates subject lines, preview text, and email body copy based on your brand voice and campaign objective. It handles smart translations for multilingual sends and includes an Image Remix feature for adapting visual assets across variants.
• Segments AI lets you describe an audience in natural language ("customers who bought twice in the last six months but haven't opened an email in 30 days") and generates the segment logic automatically. For teams not fluent in Klaviyo's segment builder, this is a genuine time-saver.
• Forms AI generates opt-in form copy and layout suggestions aimed at conversion optimisation. It's the lightest of the three tools in terms of depth, but useful for teams spinning up list growth campaigns quickly.

For a broader view of how these tools fit into an email marketing workflow, our post on Klaviyo AI tools for email marketing covers the practical setup in more detail.

Customer Agent: AI-Powered Customer Service That Operates 24/7

Klaviyo's Customer Agent is an autonomous AI service agent built to handle inbound customer queries, including order tracking, returns questions, and product enquiries, across email, WhatsApp, and other connected channels, with 65% of questions resolved without a human agent involved.

Sixty-five per cent resolution without human involvement is a meaningful number. For a brand receiving a few hundred support queries a day, that's the difference between needing a full support team and needing a smaller one.

What Customer Agent Can and Can't Do

Customer Agent draws on what Klaviyo calls "retail skills": pre-built logic for common ecommerce queries. Order status. Estimated delivery. Return eligibility. Product availability. These are the questions that make up the bulk of post-purchase support volume for most D2C brands, so the coverage is well-targeted. Since the June 2026 release, Customer Agent and Composer also share intelligence: preferences and intent signals captured in service conversations feed the customer profile, which Composer then uses when building future campaigns.

What it can't do, at least not yet, is handle genuinely complex or emotionally charged service situations. A customer who received a damaged product and is frustrated needs a human. Customer Agent's strength is volume and speed on routine queries, not empathy and discretion on edge cases. That's fine. It's honest. The mistake would be deploying it as if it could handle everything.

The 24/7 availability element is the real operational win. Most ecommerce brands have a gap between late evening orders and the next morning's support shift. Customer Agent closes that gap without staffing cost.

AI Personalisation: Reaching the Right Customer at the Right Time

Klaviyo AI personalisation works by using real-time customer data, including channel affinity, purchase history, and predicted next order date, to tailor message content, product recommendations, and send timing at an individual level across every channel the platform supports.

Personalisation in most CRM conversations means first-name merge tags. That's the floor, not the ceiling. K:AI's personalisation operates at a different level because it draws on the B2C CRM data layer underneath. The gap between "Dear [First Name]" and "this product, to this person, on this channel, at this time" is substantial. K:AI is genuinely attempting the latter.

Personalised Send Time

Personalised Send Time predicts the optimal delivery window for each individual contact based on their historical engagement patterns. Klaviyo claims top-performing campaigns saw a 35% lift in click rate using this feature. Take "top-performing" with appropriate context. Average lifts will be lower. But even a mid-range improvement in click rate from timing alone is worth the feature being switched on.

Product Recommendations

K:AI's product recommendation engine uses purchase and browsing behaviour to surface next best product suggestions inside emails and flows. This is where the B2C CRM positioning pays off. The model isn't working from a generic interest category. It's working from actual purchase sequences across Klaviyo's customer base.

Klaviyo's Consumer AI Persona Research, which surveyed 7,998 consumers across 8 countries, found that 41% of global consumers had purchased a product an AI system recommended within the past six months. That's a significant share of buyers who are already comfortable acting on AI-generated suggestions. Product recommendations inside post-purchase or browse-abandonment flows are well-placed to capture that behaviour.

Channel Affinity

Channel affinity modelling identifies whether a given customer is more likely to engage via email, SMS, WhatsApp, or mobile push. Rather than sending every message on every channel, K:AI can route contacts toward the channel where they historically respond. This matters most for omnichannel flows where duplicating messages across channels adds cost without adding engagement.

Predictive Analytics Inside Klaviyo AI

Klaviyo AI's predictive analytics features generate individual-level forecasts for customer lifetime value (CLV), churn risk, next order date, and spending potential, all derived from in-house models trained on transaction and engagement data across the Klaviyo customer base.

Predictive AI is where K:AI is most mature. These aren't new features bolted onto an existing ESP. They're models that Klaviyo has been developing since before the AI marketing wave, which gives them more training data and more refinement than most of the newer entrants in the space. It's also the data layer Composer draws on when it recommends which audience to target and when to send.

Customer Lifetime Value Prediction

CLV prediction estimates the expected future revenue from each customer over a defined window. Inside Klaviyo, this feeds directly into segment logic. You can build an audience of "predicted high-CLV customers who haven't purchased in 60 days" and trigger a retention flow at exactly that junction. That's where predictive analytics and omnichannel automation connect properly.

Churn Risk and Next Order Date

Churn risk scoring identifies customers whose engagement and purchase cadence suggest they're moving toward lapsing. Next order date prediction estimates when each customer is likely to buy again based on historical purchase intervals. Both outputs feed into flow triggers. A customer flagged as high churn risk can enter a re-engagement flow automatically. A customer approaching their predicted next order date can receive a timely prompt.

This is the kind of predictive functionality that used to require a separate data science tool or a significant custom build. Having it native inside the B2C CRM layer, tied directly to segmentation and flows, removes a significant integration burden for mid-market ecommerce brands.

For context on how to put these capabilities to work practically, our post covering 50 ways enterprise D2C brands can use Klaviyo includes specific flow and segment setups that pair well with K:AI's predictive outputs.

Klaviyo AI Across Email, SMS, WhatsApp, and More

Klaviyo AI operates across email marketing, SMS marketing, WhatsApp, mobile push, and RCS from a single platform, applying personalisation, predictive routing, and autonomous content generation consistently across every channel rather than treating them as separate tools.

The omnichannel angle isn't just a features list. It's the thing that makes autonomous AI viable in a marketing context. Composer can only generate a coordinated campaign if it has something to coordinate. A platform that spans email, SMS, WhatsApp, and push gives it the channels to work with. A platform that only does email produces an email. The caveat from earlier applies here too: Composer itself currently drafts for email and SMS, with push and WhatsApp support still rolling out, even though the wider K:AI personalisation and predictive layer already spans all of them.

SMS and WhatsApp

SMS marketing through K:AI carries the same personalisation and predictive timing logic as email. Channel affinity modelling means SMS isn't just a fallback for non-email openers. It's a primary channel for customers who genuinely prefer text. WhatsApp support extends Customer Agent's service resolution capability into a channel that increasingly handles post-purchase queries in markets outside North America.

Deliverability and Flow Monitoring

K:AI includes flow anomaly detection and guided deliverability warming tools. Anomaly detection flags unusual drop-offs in open or click rates within active flows. Volume warnings alert senders approaching thresholds that could affect inbox placement. These are assistive AI features, not autonomous ones, but they're practically important for any brand running high-volume email marketing campaigns.

Deliverability is one of those areas where most people assume the platform handles it. It doesn't, not entirely. Our email marketing deliverability service exists precisely because the tools assist but the strategy still requires human judgement.

How Klaviyo AI Integrates with ChatGPT, Claude, and MCP

Klaviyo AI connects to external AI tools including ChatGPT and Claude through an MCP (Model Context Protocol) server, allowing marketers to query customer data, generate campaign briefs, and build segment logic from outside the Klaviyo interface, including directly inside their preferred AI assistant.

This is worth flagging because it changes how the tool fits into a broader marketing workflow. Previously, using Klaviyo meant working inside Klaviyo. The MCP integration means a marketer can ask Claude to draft a campaign brief using live Klaviyo customer data, then push the output back into the platform. For teams that already run AI-assisted workflows, this removes a significant context-switching cost.

Klaviyo's financial trajectory reflects how seriously the market is taking this positioning. Full-year 2025 revenue reached $1.2 billion, representing 32% year-over-year growth. Platforms growing at that rate are reinvesting heavily in product, and Composer is clearly where the development resource is going.

If you want a practical framework for prompting AI tools like Claude and ChatGPT in your email marketing work alongside Klaviyo, our guide to 10 AI prompts for improving email marketing performance covers the RISEN framework in detail. It applies directly to how you'd brief Composer or instruct Claude via an MCP connection.

Frequently Asked Questions About Klaviyo Composer and K:AI

Is Composer the same thing as Klaviyo AI (K:AI)?

No, though the names get muddled, not least because Klaviyo currently routes a lot of "K:AI" search traffic straight to the Composer page. K:AI is Klaviyo's umbrella brand for its full suite of AI capabilities: assistive tools, predictive models, and autonomous agents. Composer, launched in March 2026, is the flagship marketing agent within that suite, sitting alongside the earlier K:AI Marketing Agent and K:AI Customer Agent. If someone says "the Klaviyo AI thing", they usually mean Composer; if they mean the predictive analytics or Customer Agent, those are separate parts of the same K:AI layer.

Is Composer available on all Klaviyo plans?

Composer is in public beta and already sits inside every Klaviyo account, with nothing to install or activate. It runs on a credit system: Klaviyo customers get free credits to start (new customers receive 10,000 complimentary credits for up to 90 days), and different actions consume different amounts, with support questions always free. Beta terms will change, so check Klaviyo's current Composer documentation for credit costs and availability.

Does Composer send campaigns automatically?

No. Composer recommends, drafts, and QAs, but nothing goes live without your approval. Every campaign it generates is staged for review inside Klaviyo's editor, where you can edit the copy, swap images, and adjust settings before anything reaches a customer. You remain responsible for the final content, and depending on jurisdiction you may need to disclose that content was AI-generated. Klaviyo provides ready-made disclosure language for this.

How does Klaviyo AI compare to other marketing automation platforms?

The main differentiator is the B2C CRM data layer. Klaviyo's AI tools draw from purchase history, browsing data, transactional events, and cross-channel engagement in a unified profile. Platforms that bolt AI onto a pure email tool are working with a much shallower data set. The predictions are only as good as what feeds them.

Can Customer Agent replace a human support team?

No, and Klaviyo doesn't claim it can. Customer Agent handles high-volume routine queries, which is where the 65% resolution rate applies. Complex, sensitive, or unusual queries need a human. The right framing is that Customer Agent handles the queue that used to eat your support team's time, not the cases that required judgement.

What is predictive analytics in Klaviyo?

Predictive analytics in K:AI generates individual-level forecasts including customer lifetime value, churn risk score, predicted next order date, and spending potential. These outputs feed directly into Klaviyo's segmentation and flow triggers, so the predictions connect to action without requiring an additional tool or manual data export.

Does Klaviyo AI work for B2B brands?

Klaviyo AI is built for B2C and ecommerce use cases. Its predictive models are trained on purchase and transactional data. For B2B lifecycle marketing, HubSpot or Salesforce Marketing Cloud are better-suited platforms. Klaviyo's positioning as a B2C CRM is genuine, not just marketing language.

Making Composer and Klaviyo AI Work for Your Programme

Composer is the most credible attempt yet at an autonomous marketing agent inside a B2C CRM, and with the public beta live since June 2026, it's no longer a demo-stage promise. Customer Agent is the service-side counterpart, and the two now share customer intelligence. Predictive analytics for CLV, churn risk, and next order date are mature and genuinely useful. The omnichannel reach across email, SMS, WhatsApp, mobile push, and RCS gives the AI layer meaningful channels to work across, even if Composer's own drafting currently covers email and SMS.

The honest caveat: the quality of every output scales with the quality of your underlying data. Sparse profiles produce generic results. Rich, well-structured customer data produces specific, actionable outputs. Getting the data layer right is the prerequisite.

AI is a force multiplier for marketers, not a replacement. Composer doesn't replace a strategist. Customer Agent doesn't replace a support team. But both meaningfully change the output-to-effort ratio for teams that know how to use them properly.

If you want to get more from Composer and Klaviyo's wider AI features, our AI for email marketing service covers implementation, data structure, and flow strategy for brands looking to run K:AI properly rather than just turning it on. Or take a look at our broader guide on AI email marketing trends for marketers for the wider context. If you'd like to talk through where your programme stands, get in touch with the team.

‍