跳到主要内容

How to Use the Model

Step 1: Assess Current State​

For each dimension, compare against the fluency level descriptions to identify where your team currently stands. Different dimensions may be at different levels, which is normal.

Tips:

  • Be honest about your current state
  • Involve multiple team members in the assessment
  • Document specific examples that support your assessment

Step 2: Set Goals​

Based on your team's business needs and technical capabilities, set reasonable target levels for each dimension. You don't need to aim for the highest level across all dimensions simultaneously.

Tips:

  • Consider your organization's strategic priorities
  • Start with dimensions that have the greatest business impact
  • Set realistic timelines for progression

Step 3: Plan the Path​

Identify the gaps between current and target levels, and develop specific improvement plans. Prioritize investments in dimensions that have the greatest impact on business value.

Tips:

  • Break down large improvements into smaller milestones
  • Identify dependencies between dimensions
  • Allocate resources based on priority

Step 4: Continuous Evolution​

Regularly reassess and track progress. Fluency improvement is gradual, and each level transition requires time and accumulated practice.

Tips:

  • Schedule regular reassessments (quarterly recommended)
  • Celebrate progress and learn from setbacks
  • Adjust goals based on changing circumstances

Common Patterns​

Starting from Awareness​

Most teams start at Level 1 (Awareness). The key investments at this stage are:

  • Building awareness and enthusiasm for AI tools
  • Establishing basic guidelines and best practices
  • Creating a safe environment for experimentation

Moving to Assisted Coding​

The transition from Awareness to Assisted Coding typically involves:

  • Standardizing on specific AI tools (e.g., GitHub Copilot)
  • Integrating AI into daily workflows
  • Establishing code review practices for AI-generated code

Advancing to Structured AI Coding​

This is often the most challenging transition, requiring:

  • Significant investment in codebase readiness
  • CI/CD integration with AI feedback loops
  • Spec-driven development practices

Reaching Agent-Centric and Beyond​

These advanced levels require:

  • Organizational commitment to AI-first development
  • Significant infrastructure investment
  • Cultural shift in how developers view their role

Anti-Patterns to Avoid​

  1. Skipping levels - Each level builds on the previous one
  2. Focusing on tools over process - Tools alone don't create fluency
  3. Ignoring quality - Speed without quality leads to technical debt
  4. Neglecting human factors - Team buy-in is essential for success