Capstone: Trace a Reliable Answer

The raw agent in Chapter 1 read two fields and guessed. The capstone runs one leadership question through the system you built and exposes every decision between the question and the answer.

The question

Which high-value renewals need executive attention this quarter?

A defensible answer needs definitions for high value, renewal risk, executive attention, and the fiscal quarter. It needs authority when fields disagree, exclusions for test accounts, user-scoped permissions, a validated query, and a trace.

What the system should do

1. Load the stable organizational rules.
2. Select the field and workflow rules named by the question.
3. Resolve authority and freshness before planning.
4. Build an exact user-scoped query.
5. Validate the finished plan at the tool boundary.
6. Execute, answer, and log the evidence.

What a good answer says

It names the date range and permission scope. It explains why Acme qualifies, which conflicting field it rejected, and what evidence supports executive attention. It does not hide uncertainty behind a confident summary.

Finish the trace

Select every required layer in the GTM Lab. The result should be reproducible from the displayed rule version, query controls, CRM values, and trace.

Build it on your platform

The guide ends at the platform boundary. Continue with the implementation that matches your stack:

• Build it in Salesforce →
• Build it in HubSpot →
• Keep it portable outside the CRM →