How to Engineer an AI-First Team for the Agentic Era: A Step-by-Step Guide

Introduction

In the fast-moving world of software engineering, staying ahead means constantly rethinking how you build and lead teams. Jon Hyman, co-founder and CTO of Braze, guided his engineering organization through nearly 15 years of growth—and then transformed it into an AI-first powerhouse in just a few months. His approach is a blueprint for any leader looking to prepare their team for the agentic era, where autonomous agents and AI-driven workflows become central. This guide breaks down his method into actionable steps you can apply to your own engineering organization.

What You Need

• Executive Buy-In: Support from C-suite and key stakeholders to prioritize AI transformation.
• Dedicated Budget: Funds for new tools, training, and pilot projects.
• Cross-Functional Team: Engineers, product managers, data scientists, and UX designers willing to learn and iterate.
• Existing Tech Stack: Familiarity with your current infrastructure (cloud, data pipelines, CI/CD) to identify integration points.
• AI Literacy Resources: Access to courses, workshops, or internal experts on machine learning and agentic systems.
• Metrics Framework: Clear KPIs for measuring success (e.g., productivity gains, deployment frequency, user satisfaction).

Step-by-Step Guide

Step 1: Assess Your Current Engineering Culture

Before diving into AI, take a hard look at your team's strengths and pain points. Hyman started by evaluating how Braze's engineers worked, where they were spending too much time, and what repetitive tasks bogged them down. Action items:

• Conduct anonymous surveys to identify friction points in development workflows.
• Analyze sprint retrospectives for recurring issues (e.g., too much debugging, manual testing).
• Map out your current toolchain—where can AI add immediate value (e.g., code review, bug triage)?

Step 2: Build AI Literacy Across the Team

You can't become AI-first if only a few data scientists understand the tech. Hyman made AI education a priority for all engineers. Roll out a learning program that covers fundamentals of machine learning, prompt engineering, and agentic design patterns. How to do it:

• Host weekly lunch-and-learns on AI topics (use internal or external speakers).
• Create a shared Slack channel or wiki with curated resources (papers, tutorials, case studies).
• Pair engineers with data science mentors to work on small AI experiments.
• Encourage everyone to complete at least one online certification (e.g., from Coursera or Fast.ai).

Step 3: Identify High-Impact AI Opportunities

Not every problem needs an AI solution. Focus on areas where machine learning or agentic systems can deliver quick wins and show value. Hyman targeted improvements in developer productivity—like automating code generation and testing. Filter opportunities by:

• Feasibility: Do you have the data and compute resources?
• Impact: Will it save engineering hours or improve product quality?
• Time to Value: Can you see results in weeks, not months?

Step 4: Set Up an AI Sandbox for Experiments

Create a safe environment where teams can prototype AI features without affecting production. Braze used dedicated cloud credits and isolated development branches. Best practices:

• Use modular, containerized approaches (Docker, Kubernetes) to spin up AI services quickly.
• Integrate with your existing CI/CD pipeline to test AI models alongside regular code.
• Establish clear guardrails for data privacy and security—no real customer data without anonymization.

Step 5: Restructure Teams for AI-Native Workflows

An AI-first engineering org doesn't just add AI features—it changes how teams are organized. Hyman advocated for cross-functional squads that combine domain experts with AI specialists. Structural changes to consider:

• Form a central AI platform team to provide common tools (model serving, feature stores).
• Embed AI engineers into product squads so they directly solve customer problems.
• Create a rotating “AI champion” role in each team to promote best practices.

Step 6: Implement an Iterative Learning Loop

Transformation doesn't happen overnight. Hyman drove change quickly but iteratively—shipping AI features, measuring impact, and refining. Set up a loop:

• Define baseline metrics before each AI initiative.
• Ship a minimum viable AI feature to a small user group (canary release).
• Collect quantitative (e.g., task completion time) and qualitative feedback (e.g., developer satisfaction).
• Use feedback to improve the model or agent, then expand rollout.

Step 7: Scale and Embed AI into Company DNA

Once you have proven wins, scale across the organization. Hyman made AI a core part of Braze's engineering vision, not a side project. Scaling tactics:

• Document successful patterns in an internal AI playbook.
• Create a center of excellence that advises teams on AI design.
• Celebrate wins in company all-hands meetings to build momentum.
• Update your hiring criteria to prioritize AI fluency for new engineers.

Tips for Success

• Start Small, But Think Big: Pick one problem that can be solved in a few weeks. A small win builds confidence and support.
• Communicate the Vision Constantly: Over-communicate why AI matters for your team's future. Hyman shared a clear narrative about agency—engineers will still drive creativity, but AI handles drudgery.
• Don't Neglect Soft Skills: AI transformation requires change management. Invest in coaching for managers to help their teams adapt.
• Measure What Matters: Track not just technical metrics but also team morale and retention. AI should make work more enjoyable, not more stressful.
• Stay Ethical and Transparent: Be upfront with customers and employees about how you're using AI. Braze prioritized responsible AI from day one.
• Embrace the Agentic Mindset: Instead of automating entire workflows, design systems where humans and agents collaborate. Think of agents as teammates, not replacements.

By following these steps—inspired by Jon Hyman's leadership at Braze—you can transform your engineering organization into an AI-first team ready for the agentic era. The journey requires patience, experimentation, and a willingness to rethink everything, but the payoff in productivity and innovation is immense.