← Glossary

AI Fluency Pyramid

A tiered framework, originated at Brex, for assessing how deeply an organization has integrated AI into its workflows -- from basic individual tool use at the base to fully autonomous agent-driven operations at the top.

Context

The AI fluency pyramid was developed at Brex under CTO James Reggio as part of the company’s broader AI transformation strategy. Reggio detailed how the framework shaped Brex’s internal AI adoption in his Latent Space podcast appearance, which was discussed on Episode 11 of the ADI Pod. The pyramid provides a concrete way to evaluate where a team or organization sits on the spectrum of AI integration, moving the conversation beyond vague claims of “using AI” toward measurable levels of fluency.

Why It Matters

Most companies say they are adopting AI, but few have a shared vocabulary for what that means in practice. The AI fluency pyramid gives organizations a ladder to climb rather than a binary on/off switch. At the lower levels, individual contributors use AI tools like code completion or chat assistants for isolated tasks. Higher levels involve teams building workflows around AI — using it for code review, testing, and documentation as a standard part of the development process. At the top, AI agents operate with significant autonomy, orchestrating multi-step tasks with minimal human intervention.

The framework is useful because it exposes the gap between perception and reality. A team might believe it is highly AI-fluent because every developer has a Copilot license, but if no one has changed their workflows or processes to leverage AI systematically, the organization is still near the base of the pyramid. Real fluency requires cultural and process changes, not just tool access.

Practical Application

Engineering leaders can use the pyramid as an internal assessment tool: survey teams on how they use AI, map the results to pyramid levels, and identify what needs to change — tooling, process, or culture — to move up. The Brex case study demonstrates that moving up the pyramid is as much an organizational challenge as a technical one — Reggio describes it as a company-wide commitment, not a developer tooling checkbox.

Related Concepts

• Cognitive debt — a risk that increases when teams advance up the pyramid without building genuine understanding of AI-generated output
• Verification debt — the testing and review gaps that can accumulate as AI autonomy increases at higher pyramid levels