AI Overview
Generated by Klaviyo AI
Klaviyo announced four platform updates at K:BOS 2026: Klaviyo Headless, SQL support in Klaviyo Data Platform, a new Personalization layer that replaces Marketing Analytics, and expanded capabilities for its agents, Composer and Customer Agent. Together, they get brands closer to a truly autonomous B2C CRM — one you can run from any AI system, ask questions of in plain language, and let decide who, what, when, and where on every send.
- Klaviyo is now headless, so your team can run it without opening Klaviyo: More than 260 Klaviyo MCP tools and capabilities are callable from Claude, ChatGPT, or any AI system your team already uses, and an agent can read, write, and go live the same way you would.
- The new Personalization layer makes four decisions on every single send: It picks who to reach with Audience Optimization, what to show with 1:1 Content and Product Recommendations, when to send with Personalized Send Time, and where to reach them with Channel Affinity — then measures lift against a real holdout. In preview.
- Klaviyo's agents now bring finished work to you instead of waiting for a prompt: Composer adds scheduled audits, cited analysis powered by the new SQL engine, and custom skills you save once. Customer Agent goes global in 100+ languages — it already resolves 86% of customer queries and 94% of product recommendation requests for Naked Wardrobe.
The consumer economy has changed. Every major B2C industry is competing on data, attention, and customer relationships now. The customer experience matters more than it ever has.
AI should be changing how you connect with your customers. The real question is whether it actually is, yet.
What we hear from you, on calls and at our events, is that you want AI to make this easier, and you're ready to lean in. What's been in the way is practical, not philosophical.
You can't find the time to break your marketing into pieces you can tackle one by one. You don't have the tooling to automate, personalize, and optimize the way you need to. And every AI project seems to create a job you can't afford to hire for: AI project manager.
So this year, we built the answer to those problems. I can't wait for you to see what it does.
Klaviyo is now headless
You shouldn't have to work where we decide you work.
Some of you love the Klaviyo UI. Some of you have moved into an LLM interface and haven't looked back. Some of you want APIs and commands. So, we opened the whole thing up.
With Klaviyo Headless, more than 260+ Klaviyo MCP tools and capabilities are callable directly from Claude, ChatGPT, or any AI system your team already uses. An agent can read, write, and go live the same way you would inside Klaviyo, without anyone opening Klaviyo at all.
There are already a lot of ways brands are using this. Some are building custom reporting dashboards in Cursor or Lovable with live Klaviyo data. Others are using MCP to automate a weekly performance review and post the findings straight to Slack. Or asking ChatGPT which product is trending this week, then having it draft the campaign, ready for you to send.
That’s the Klaviyo we are building for you. One that works anywhere and in any way you need it to.
New SQL support in Klaviyo Data Platform
Klaviyo has always been a data platform. Today that means hundreds of APIs and integrations, processing more than 4 billion signals a day. As that foundation grew, brands started asking for more than access to their data. They wanted to actually understand it, and use it wherever their team and their agents already work.
The challenge is making sense of all that data has always required someone who can write a technical query. Today, Klaviyo closing that gap by introducing SQL inside the Klaviyo Data Platform (KDP). With SQL, a marketer can ask a question in plain language, through Composer, MCP, or another agent, and Klaviyo translates that prompt into a SQL query, runs it against the brand's own data, and returns the answer along with the SQL query that produced it.
Today, SQL can be accessed in Composer and MCP as a preview. It can query profiles, events, segments, lists, and catalogs, with forms, flows, and custom objects coming soon.
SQL in Klaviyo Data Platform is in Preview.
Marketing Analytics is now Personalization
We are updating and extending our Marketing Analytics product and making it a core layer of our platform under a new name: Personalization.
The Klaviyo Personalization layer will make 4 decisions on every send.
- Who to reach, using Audience Optimization against live recency, frequency, and monetary value (RFM) data.
- What to show them, using 1:1 Content, Product Recommendations, and discount decisioning.
- When to reach them, using Personalized Send Time.
- Where to reach them, using Channel Affinity.
Then it tests and optimizes across email, text, WhatsApp, mobile app, website, and social.
The Personalization layer runs on top of the Klaviyo data platform, which means it processes 4 billion signals a day across more than 9 billion profiles, and you and your team can see and configure it the whole way through. Built-in incrementality reporting measures lift against a real holdout, too, so you never scale on faith alone.
Personalization is in Preview.
Your agents just got a lot more capable
New Composer capabilities
Think of Composer, Klaviyo's AI marketing agent, as three roles you'd hire if you could. A data analyst who finds what matters. A head of CRM who decides what to do next. A marketer who gets it done.
Here is what is new in Composer:
- Analysis you can stand behind. Powered by the SQL engine, Composer goes deeper, and every answer comes with a citation you can check
- Scheduled tasks. Recurring audits replace one-time checks, so a monthly performance summary arrives on its own and an overlapping audience or a declining campaign gets caught before either costs you real revenue.
- Custom Composer Skills. Save your own proven approaches, like subject line standards, brand guidelines, or content hierarchy, as reusable context you don't have to repeat every time you prompt it.
- Coming soon: a new Klaviyo Mobile App, built around Composer, keeps you connected to a conversational analyst and campaign builder from anywhere.
Composer used to be something you prompted. Now it brings finished work to you.
New Customer Agent capabilities
Customer Agent has exchanged more than 1.3 million messages with customers since we first launched it, and brands like Naked Wardrobe use it to resolve 86% of their customers’ queries autonomously, and 94% of product recommendation queries.
Today, we’ve added three new capabilities to Customer Agent:
- One agent, every region. Customer Agent can now go global out of the box, supporting customers in more than 100 languages with automatic locale detection through Shopify Markets, and the right catalog, pricing, and currency for each market.
- Coming soon: Voice as a channel. It can take a phone call now, with the full context of a customer's history before it says hello. And, it handles the whole call from start to finish. No hold, no transfers, no repeating yourself.
- Smarter with every conversation. Each conversation becomes profile data available to Composer and the Personalization layer, and auto-training and simulations let you tell Customer Agent how to improve in plain language.
Additional K:BOS announcements
We also launched onsite personalization powered by Customer Hub, Clicks to Bricks capabilities, and a Compliance Hub to support all of these new features and ensure you can build a truly omnichannel customer experience you and your customers trust with Klaviyo.
No matter where you work, how you run your brand, or how you think about change management in the AI era, we are here to help. These new features and capabilities extend Klaviyo even further so it can work harder for you—literally for you, if you want it to.
That’s the power of a strong data foundation, a consistently optimized Personalization layer, and Agents that learn from every single data point, including from each other.
We’re here to help you power the next era of customer experience. And we’re just getting started.
Related content
