Klaviyo: Why trust still wins in the age of autonomous CRM

Customer engagement is going through two shifts at once. Brands are trying to untangle years of accumulated point solutions, each with its own data, channel, and workflow. At the same time, AI-driven tools are arriving faster than most teams can evaluate them, promising automation that is easy to demo and much harder to make work.

Greg Yu sits where those two shifts meet. As VP of Engineering at Klaviyo, the autonomous B2C CRM platform that brings marketing, service, agents, and data together in one place, he oversees the systems that hundreds of thousands of brands rely on to reach their customers. His view is that the fundamentals of unified data, reliable delivery, and a steady focus on what actually works matter now more than ever.

The fragmentation problem holding brands back

Most brands don’t struggle for lack of tools. They struggle because they have too many, and few of them work together. A single customer lifecycle might span email, SMS, a loyalty program, a service desk, and a growing list of messaging apps, each managed in a separate system with its own slice of customer data.

Greg describes this fragmentation as one of the biggest challenges brands face today. Every new point solution adds another integration to maintain and another place where customer context gets lost. Reaching customers consistently across the lifecycle becomes an exercise in stitching systems together rather than understanding the people on the other end.

Klaviyo brings channels together in one place while integrating a wide range of data sources, so brands can put their first-party data to work in every interaction. When data and channels live in the same system, orchestration stops being a coordination problem and becomes a strategic one. That matters more as the channels themselves become more capable.

Moving from outreach to outcomes on conversational channels

Messaging used to be largely one-way broadcasts. A promotion goes out, and success is measured in opens and clicks. The newer generation of channels makes that model look narrow.

He points out that these channels also open the door to genuinely conversational communication, where a customer can ask a question, browse options, or complete an action without leaving the thread. For brands, this changes what messaging is for. The goal moves beyond outreach toward driving outcomes like resolving a question or completing a purchase, which leads to a customer who feels understood.

That shift raises the stakes for everything underneath. A conversation only works if it reaches the customer, and if the customer trusts who it comes from.

Why trust and deliverability matter more than ever

Klaviyo was built from the ground up as a data platform, and Greg sees that foundation as central to what sets it apart. Brands bring together data from many sources and query it constantly to power segmentation and orchestration, which demands enterprise-grade scale and reliability.

Scale alone, though, isn’t the differentiator he emphasizes most. “With AI-driven solutions popping up all over the market, trust and reputation are more important than ever,” says Greg. In a crowded market, a platform that hundreds of thousands of brands have used to grow carries weight that a new entrant can’t easily replicate. Greg counts the investment in building that trusted platform as a core part of Klaviyo’s mission.

Trust also has a practical, technical side, and text messaging is where it becomes most visible. Deliverability depends on sender reputation, carrier relationships, and regulatory requirements that vary by country and keep changing.

Greg describes Infobip as a key partner in Klaviyo’s text messaging journey:

For Greg, these fundamentals are where platforms should stay most disciplined. Despite the pace of change across the industry, he argues this is not the moment to reinvent reputation and deliverability. They remain as important as they have always been.

The real hurdle with AI agents is adoption

Klaviyo recently launched a number of agentic products, including Customer Agent and Composer. Releasing agents, Greg has found, is the easier half of the problem.

Setting up an agent well takes real groundwork. Brands need to define custom skills, connect the agent to the right data sources, and make sure it respects their tone and guidelines in every interaction. According to Greg, Klaviyo is streamlining and automating much of that setup, so brands spend less effort configuring agents and more time seeing results from them.

It is a useful corrective to how AI is often discussed in customer engagement. What an agent can do is no more important than whether a brand can trust it to speak on its behalf.

Rebuilding engineering around an agent-first mindset

The same thinking shapes how Klaviyo builds software. Greg’s Research and Development organization defined clear levels of AI progression and set goals for engineers to build their AI fluency over time. The aim was not incremental tool adoption but a fundamental rethink of the software development lifecycle from an agent-first perspective, using agentic coding to accelerate work and increase impact across every team.

Alongside that cultural shift, Klaviyo has been investing in an internal agent platform. It provides shared primitives, such as memory, authentication, and streaming, that every team building agentic products can draw on. Greg explains that this lets Klaviyo build with shared efficiency, greater velocity, and consistency, rather than having each team solve the same foundational problems on its own.

For engineering leaders at other platforms, the lesson is that AI transformation is as much an organizational question as a technical one. Without shared foundations, teams end up recreating internally the same fragmentation their customers are trying to escape.

What comes next for customer engagement

Looking ahead, Greg sees two challenges defining the next phase. The first is making agents work in practice by helping brands move from experimentation to agents that are well configured, widely adopted, and trusted to represent the brand. The second is global reach. As Klaviyo continues to grow geographically, the question becomes how to keep consolidating the channels that matter most in each region, since the way customers prefer to communicate differs from one market to the next.

Greg sees collaboration between platforms playing a part in both, from combining the distinct data footprints each one holds to drawing on regional and channel experience when entering new markets and meeting new types of brands. Underneath it all is the same principle that runs through his approach to data, deliverability, and AI.