Engineering for the Agentic Era: A CTO's Guide to Transforming Your Team into an AI-First Powerhouse

Introduction

Jon Hyman, co-founder and CTO of Braze, led his engineering organization through nearly 15 years of growth and then, in just a few months, overhauled it into an AI-first team. His journey offers a blueprint for CTOs and engineering leaders who want to prepare their teams for the agentic era—a time when autonomous AI agents become core to product development and operations. This guide distills his approach into actionable steps, helping you rethink your engineering culture, structures, and practices.

What You Need

• Leadership commitment – unwavering support from the C-suite
• Receptive engineering team – openness to change and learning
• AI tools access – LLMs, APIs, ML platforms
• Existing codebase & infrastructure – a foundation to build upon
• Experimentation culture – permission to fail and iterate

Step-by-Step Guide

1. Assess Your Current Engineering Culture

   Before any transformation, understand where you stand. Evaluate how your team currently approaches problem-solving, collaboration, and technology adoption. Hyman realized that Braze’s legacy of pragmatic innovation could be accelerated. Key action: Survey team members, review recent project outcomes, and identify resistance points. Document your culture’s strengths (e.g., agility, ownership) and weaknesses (e.g., silos, risk aversion).

2. Secure Executive and Team Buy-In

   Transformation requires top-down support and bottom-up enthusiasm. Hyman briefed Braze’s leadership on AI’s strategic importance and then presented a clear vision to engineers. Key action: Host a town hall explaining why the shift to AI-first is necessary for staying competitive. Share success stories from early pilots. Address fears about job displacement by emphasizing new roles and upskilling opportunities.

3. Define AI-First Principles

   Articulate guiding principles that will steer decision-making. For example: “AI should augment, not replace, human judgment” or “Every feature should ask: Can an agent handle this?” Hyman and his team established that AI would be embedded in their product from the ground up. Key action: Create a one-page document with 3-5 principles and socialize it across the org. Use them to evaluate new projects and tools.

4. Pilot AI Projects on Low-Risk Areas

   Start small to build confidence and gather data. Braze launched AI-powered experiments internal tools before customer-facing features. Choose an area where failure won’t be catastrophic but learning will be high. Key action: Identify three low-stakes problems (e.g., code review automation, test generation) and assign small teams to prototype AI solutions. Measure results against baseline metrics.

5. Rewire Team Structures for AI Integration

   Traditional engineering teams often separate data scientists, ML engineers, and software engineers. To operate in the agentic era, these roles must blend. Hyman restructured Braze’s teams around product verticals with embedded AI experts. Key action: Create cross-functional squads that include ML specialists, prompt engineers, and full-stack developers. Ensure each squad has autonomy to adopt AI tools and agents.

   

6. Invest in Continuous Learning and Upskilling

   Transformation happens when people grow. Braze launched internal workshops on prompt engineering, LLM fine-tuning, and agent architectures. Key action: Partner with training providers or build in-house curricula. Allocate dedicated time (e.g., 10% of work hours) for learning. Encourage engineers to get certified in AI/ML platforms.

7. Scale AI Across the Engineering Organization

   Once pilots succeed and skills are built, roll out AI-first practices broadly. Hyman made AI a standard part of the development lifecycle, from design to deployment. Key action: Integrate AI code assistants into your CI/CD pipeline, adopt agentic testing frameworks, and require AI feasibility assessments for new features. Track adoption metrics like percentage of code generated by AI agents or time saved in routine tasks.

8. Build Ethical Guardrails and Monitor Impact

   Agentic systems can behave unpredictably. Braze implemented strict governance on AI outputs, including human-in-the-loop checks for customer-facing actions. Key action: Establish a review board for AI deployments, define acceptable failure rates, and set up monitoring for bias, drift, and security vulnerabilities. Regularly audit agent performance and iterate on guardrails.

Tips for Success

• Start small, iterate fast. Don’t try to overhaul everything at once. Use agile sprints to test and refine your approach.
• Encourage cross-functional collaboration. Break down silos between engineering, product, and data science. The agentic era thrives on shared knowledge.
• Measure success with new KPIs. Move beyond lines of code or story points. Track agent efficiency, time-to-resolution, and customer satisfaction from AI features.
• Don’t forget human oversight. Even the most advanced agents need governance. Maintain a healthy skepticism and keep humans in the loop for critical decisions.
• Celebrate learning, not just outcomes. When experiments fail, treat them as valuable data. Share lessons learned openly to accelerate the whole team’s journey.

Transforming your engineering organization for the agentic era is a marathon, not a sprint. Jon Hyman’s experience at Braze shows that with strong principles, smart pilots, and a commitment to upskilling, any team can become AI-first in months—not years. Use this guide as your roadmap, and embrace the change.