AI agent for executive directors
Strategy Decision Log Outcome Review Agent
Review every major decision against its expected outcome and capture the lessons
What it does
A company approves an acquisition, a pricing move or a market entry, and two years later nobody can say whether it worked or what the assumptions were. This agent closes that loop. It reads the decision log and extracts what was expected for each decision, such as target revenue, a date and the main assumptions. When a review date arrives, it pulls the actual data, compares it with the expectation and classes the decision as on track, off track or unclear. For off-track decisions it asks the owner for causes and checks whether an assumption failed. It then collects lessons into a list. The leader approves the lessons list. Edge case: an outcome cannot be measured. The agent marks it as unclear and asks which measure to use.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Review date arrives for a decision
- Read the decision log entry and extract expected outcomes and assumptions
- Is the outcome measurable with available data?If not: Ask the owner for a measure and mark the decision unclear until set. Back to step 2.
- Pull actual data for each measure
- Compare actual to expected and class the decision
- For off-track decisions test which assumption failed
- Ask the owner to confirm causes
- Did the owner confirm or correct the cause?If not: Remind once and note the cause as unconfirmed. Back to step 7.
- Strategy leader approves the lessons listThe agent waits here for your OK.
- Decision review report and lessons list
How it decides
A decision is on track when actual results are within 10 percent of the expectation, off track when worse by more, and unclear when no measure exists.
- On track means within 10 percent of the expected result
- An assumption is failed when its measured value differs by more than 20 percent
- Decisions without a measure are unclear, not on track
- Lessons must name the assumption, not blame a person
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- On-track tolerance (default 10 percent)
- Review frequency
- Decision types to include
- Report layout
- Who receives the lessons list
What keeps you in control
It always asks you first
- Strategy leader approves the lessons list before sharing it
Hard limits
- Never assigns blame to a person
- Cites data source and date for every figure
It stops when
- Done: all due decisions reviewed and classed
- Stop: data missing for most decisions, report that the log needs better measures
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide
