We want to deploy an AI-driven tool to draft our customer proposals, but we are terrified of the risk of hallucinated pricing or technical specs. How do we structure a Thinking Time session to map out the exact guardrails we need?

Question

Tyler Smith · Accepted Answer

Protecting your gross margins from AI errors requires disciplined planning, not gut reactions. To design the right guardrails, you should schedule a dedicated thirty minute Thinking Time session using the framework from Keith Cunningham's book, The Road Less Stupid.

Start by formulating a high-value question to focus your mind. For example, use the framing: How might we integrate AI into our proposal drafting process so that we can double our speed without ever sending an inaccurate price to a client?

During this session, sit with a pad of paper in a quiet room with zero distractions. Brainstorm every possible point of failure in your current estimation workflow.

Once you have identified the risks, design a simple, multi-layered quality control system. The first guardrail should be database segregation. Ensure the AI tool is only querying a locked, pre-approved pricing spreadsheet rather than generating numbers dynamically.

The second guardrail is human verification. No AI-generated proposal can ever be sent directly to a client. It must be reviewed by a team member whose seat on the Accountability Chart owns proposal accuracy. Write this verification step directly into your documented sales Core Process, and measure proposal error rates on your weekly Scorecard. This structured approach allows you to capture the speed of AI while eliminating financial risk.