Most marketers use AI reactively: open a chat window, type a prompt, get an answer. Christopher Penn of Trust Insights argues that is the wrong model for ad account management. His latest video lays out how to build agents that watch your Google and Meta ad accounts continuously and surface waste without you asking.
What You Need Before You Build
Penn’s first requirement is documented standard operating procedures. Before an agent can act on your behalf, it needs a written definition of what counts as waste in your specific account. Vague intentions do not translate into machine-executable logic. If your team cannot agree on what a wasted dollar looks like, the agent cannot flag it reliably either.
The Trust Insights 5P Framework
Penn uses the Trust Insights 5P framework to turn business intent into user stories and decision logic a machine can execute. The framework bridges the gap between what a marketer wants and what an agent can actually evaluate against the Google Ads and Meta Ads APIs.
️ The Two-Part Architecture
The build has two components working together:
- Deterministic Python scripts that pull data from ad platform APIs on a schedule
- A scheduled agent that applies the decision logic and surfaces findings
Penn also adds MACD-style safeguards to the system. MACD (moving average convergence divergence) is a signal-smoothing technique borrowed from financial analysis. Applied here, it prevents the agent from flagging noise as waste and keeps it focused on statistically meaningful shifts in performance.
The Operator Angle
If you manage paid search or paid social at any scale, the pattern Penn describes is worth studying even if you do not follow his exact stack. The core insight is separating data collection (deterministic, scripted) from decision-making (agent-driven, policy-governed). That separation is what makes the system auditable and trustworthy enough to run without constant supervision.
The full walkthrough, including the call-to-action for connecting your own ad APIs, is available on the Trust Insights YouTube channel.
