The challenge they faced:
"We need an AI plan, but I'm not sure where to start. People are already using these tools. We need to contain the risk, but we also need to enable the work." That's how the conversation started with a midsize professional services firm. Their team was experimenting, some using ChatGPT, others Claude, a few testing Copilot, with no consistency and growing anxiety about client data exposure. Leadership had basic questions: How do we make sure platforms aren't training on our data? What are our firm-wide guidelines? How do we align a team with wildly different AI knowledge?
Our approach:
We treated AI readiness as strategic work, not technical onboarding. We interviewed four leaders about their goals, current usage and risks, then surveyed the team. We started by tackling the fear head-on: we demystified AI terms, covered what's safe versus unsafe, and co-developed clear guidelines on what data can touch AI, how to verify platforms aren't training on client information, and what requires human review. Then we worked through real examples from their work (analyzing agreements, summarizing sensitive documents, interpreting financial data), compared platforms on their needs rather than vendor marketing, and built a phased roadmap together.
The business impact:
- Every employee moved to a shared Claude Team account and trained on what's safe to share and how to work together in it
- Documented AI guardrails and properly configured platforms
- A phased roadmap the firm is actively executing
- A first, low-risk application in use: AI drafts initial document summaries that analysts review and refine
