7 Ways Braze's CTO Is Reshaping Engineering for the Agentic Age

Jon Hyman, co-founder and CTO of Braze, has spent nearly 15 years steering the company's engineering organization through explosive growth and technological shifts. In a recent conversation, he revealed how Braze transformed from a traditional engineering team into an AI-first powerhouse in just a few months. This article breaks down the key strategies he used to prepare for the agentic era—a future driven by autonomous AI agents—while maintaining the startup culture that made the company successful.

1. Embrace the Agentic Era from Day One

Hyman emphasizes that engineering leaders must anticipate a world where AI agents are not just tools but autonomous collaborators. At Braze, this meant pivoting from a human-centric workflow to one where AI models handle decision-making and execution. The team redefined product requirements to include agentic capabilities, such as self-optimizing marketing campaigns and real-time customer interactions. This shift required a complete rethinking of system architecture, data pipelines, and even hiring priorities. Hyman notes that waiting to adopt an agentic mindset would leave companies scrambling to catch up. Instead, Braze embedded AI agency into its core engineering principles, ensuring every new feature could leverage autonomous agents. The result: a platform that not only responds to user commands but proactively improves itself—a key competitive advantage in the years ahead.

2. Restructure Teams Around AI Capabilities

Traditional engineering org charts—divided by frontend, backend, and infrastructure—don't work when AI becomes the central pillar. Hyman restructured Braze's teams into cross-functional squads focused on AI-driven outcomes: personalization, predictive analytics, and agent orchestration. Each squad includes data scientists, ML engineers, and traditional software engineers who collaborate on end-to-end agentic workflows. This blurred the line between model development and production engineering, speeding up deployment cycles. Hyman also introduced roles like "agent behavior designers" to ensure AI actions align with business goals. The new structure reduced silos and made it easier to iterate on agent capabilities. For example, a single squad now owns an agent from training to live monitoring—a stark contrast to the old handoff-heavy model.

3. Implement a Rapid AI-First Transformation Timeline

Braze didn't spend years planning its AI pivot. Hyman led the organization to an AI-first approach in under three months—a timeline he admits seemed impossible at the start. The key was a concentrated sprint focused on replacing legacy rule-based systems with ML-powered alternatives. Teams held daily stand-ups to review agent performance, and leadership removed all non-essential projects to free bandwidth. Hyman also used an internal hackathon to prototype agentic features, then fast-tracked the most promising ones into production. This compressed timeline forced rapid decision-making and prevented analysis paralysis. While risky, it allowed Braze to launch agentic capabilities ahead of competitors. Hyman advises other CTOs to set aggressive deadlines for AI transformation, as the market will not wait for perfect solutions.

4. Retain Co-Founder Culture During Exponential Growth

From a handful of engineers to hundreds, Braze's engineering culture could have diluted. Hyman actively preserved the co-founder spirit—where every team member feels ownership and is encouraged to question assumptions. He did this by maintaining open communication channels, including a monthly "Founder AMA" where engineers can pitch ideas directly to him. Even as the org scaled, Hyman kept decision-making decentralized, trusting squads to make AI-related calls without top-down approval. This autonomy accelerated the agentic transformation because engineers felt empowered to experiment. Hyman also hired for curiosity and adaptability over pure technical skill, ensuring new hires fit the fast-moving culture. The result: a cohesive team that operates like a startup within a larger enterprise, essential for navigating the uncertainties of the agentic era.

5. Invest in Agent Observability and Safety

With autonomous agents making decisions, Braze had to build robust observability to monitor their behavior. Hyman's team developed a suite of tools to track agent actions, flag anomalies, and enforce guardrails. Each agent generates a transparent audit trail that explains its reasoning—a necessity for both debugging and compliance with data privacy regulations. Braze also implemented a "human-in-the-loop" override system for high-stakes decisions, like campaign budget changes. Safety wasn't an afterthought; it was built into the agent design from the start. Hyman argues that agentic systems must be trustworthy, or customers will reject them. By investing in observability and safety, Braze provided confidence to its clients and reduced the risk of AI-related incidents. This proactive approach has become a selling point, as enterprises worry about the risks of black-box AI.

6. Rethink Data Architecture for Continuous Learning

Traditional batch processing couldn't support agents that learn in real time. Hyman shifted Braze's data infrastructure to support streaming data and online learning models. This meant migrating from nightly batch updates to event-driven architectures where agents receive and act on data within milliseconds. The engineering team also built feature stores and model registries to manage the lifecycle of hundreds of agents. Data quality became paramount: Hyman established automated data validation pipelines to ensure agents are trained on accurate, unbiased data. The new architecture reduced latency and allowed agents to adapt to user behavior instantly. For example, a marketing agent can now adjust a campaign mid-flight based on real-time engagement signals. This continuous learning loop is the engine behind Braze's agentic capabilities, making it possible for the platform to improve autonomously over time.

7. Lead with a Vision, Not Just Technology

Finally, Hyman stresses that engineering transformation isn't just about code—it's about storytelling. He spent hours communicating the agentic vision to the entire company, explaining why AI agents are essential and how they will change customer experiences. This clarity motivated teams to embrace the radical changes. Hyman also publicly shared lessons learned at industry events, positioning Braze as a thought leader in agentic engineering. He advises CTOs to articulate a compelling narrative that connects AI adoption to the company's mission. Without a strong vision, the technical overhaul can feel like a chaotic upheaval. By leading with purpose, Hyman ensured that engineers understood the "why" behind the shift, fostering enthusiasm rather than resistance. Today, that vision drives every decision, from hiring to product roadmaps.

Braze's journey from a traditional engineering team to an AI-first organization in just a few months offers powerful lessons for CTOs navigating the agentic era. By embracing autonomous agents, restructuring teams, moving fast, preserving culture, prioritizing safety, revamping data architecture, and leading with vision, Hyman has shown that transformation is possible—even under tight timelines. The agentic age is not a distant future; it's unfolding now. Engineering leaders who act decisively, as Hyman did, will be the ones to define what comes next.