Building AI Teams That Ship: Our Approach to Talent and Process

The rise of AI coding tools has fundamentally changed how we think about hiring, team structure, and development processes.

---

Building an effective AI team in 2025 looks nothing like it did three years ago. The tools have changed, the skillsets have evolved, and the expectations for delivery speed have accelerated dramatically.

The New Hiring Criteria

We've shifted our hiring emphasis significantly:

What Matters More Now

Proficiency with AI coding tools

We explicitly look for experience with tools like Cursor, Claude Code, and similar AI-assisted development environments. This isn't about replacing programming skill—it's about amplifying it.

A developer who knows how to effectively prompt, iterate, and validate AI-generated code can move 3-5x faster than one who doesn't. This multiplier is too significant to ignore.

System design thinking

As implementation speed increases, the bottleneck shifts to design. Engineers who can architect robust systems—thinking through failure modes, scaling considerations, and integration points—are more valuable than ever.

Testing intuition

AI tools can generate code quickly, but they can't verify correctness against business requirements. Engineers with strong testing instincts catch issues that would otherwise reach production.

What Matters Less

Memorization of syntax and APIs

AI tools have made this largely irrelevant. We care that engineers understand concepts, not that they can write a sorting algorithm from memory.

Years of experience in a specific language

Language fluency matters less when AI can translate between languages effectively. We care about problem-solving ability and system understanding.

Our Team Structure

We've evolved toward smaller, more autonomous teams:

The Pod Model

Each pod consists of:

• 1-2 senior engineers with strong architecture skills
• 2-3 mid-level engineers proficient with AI tools
• Direct access to domain expertise (client stakeholder or internal specialist)

Pods own features end-to-end: design, implementation, testing, deployment, and monitoring.

Why It Works

Small teams with AI tools can move faster than large teams without them. The coordination overhead of large teams becomes a liability when implementation is no longer the bottleneck.

Process Adaptations

Test-Driven Everything

We've doubled down on TDD, but evolved our approach:

1. Write tests that capture business requirements
2. Use AI tools to generate implementation candidates
3. Run tests to validate
4. Iterate on failures with AI assistance

This workflow is dramatically faster than traditional TDD while maintaining—often improving—code quality.

Documentation as Specification

We invest heavily in clear, detailed specifications before implementation. Why?

AI tools are remarkably good at implementing well-specified features. They struggle with ambiguous requirements. Clear specs become direct productivity multipliers.

Continuous Code Review

AI-generated code requires careful review. We've implemented:

• Mandatory human review for all AI-assisted code
• Automated checks for common AI code patterns that cause issues
• Regular "code archaeology" sessions to understand and improve AI-generated sections

The Research vs. Shipping Balance

One tension in AI teams: the pull toward research and experimentation versus the need to ship production systems.

Our approach: dedicated exploration time with clear boundaries.

• 80% of time on delivery commitments
• 20% on exploration and learning
• Monthly showcases to share findings
• Clear criteria for promoting experiments to production

This keeps the team current without sacrificing delivery reliability.

Looking Forward

The pace of AI tool improvement shows no signs of slowing. The teams that thrive will be those that continuously adapt their processes to leverage new capabilities.

Our commitment: maintain a learning culture while delivering consistent results. It's a balance, but it's the only way to build AI teams that ship.

---

Interested in joining our team? Check out our open positions.