---
title: "Prompt: lessons learned from an AI initiative"
slug: lecciones-aprendidas-iniciativa-ia
lang: en
updated: 2026-10-19
---

# Prompt: lessons learned from an AI initiative

How to use it: copy the block, paste it into ChatGPT, Claude or Gemini and replace the fields in brackets. It works best if you also paste the original proposal and the closing report or tracking data. Use it to prepare the team session, not to replace it: the people who were there know which signal was ignored. Do not paste confidential data unless your company has approved it.

```text
Act as a project director with experience closing technology initiatives, who separates data from interpretation, does not look for culprits, and turns every review into a verifiable rule for the next one.

Context I am giving you:
- Company and area: [e.g. insurer, 800 people, customer service]
- Initiative and status: [name · close / pause / scale]
- What the original proposal promised, with number and deadline: [copy word for word]
- What happened, with figures and dates: [ ]
- Milestones or tracking data I have: [week by week, if available]
- Original proposal and closing report: [paste text or write "none"]

What I need: a lessons-learned review with exactly these six fields, in this order:
1. What we expected: the promised number, unrewritten.
2. What happened: the actual number with a date, no adjectives.
3. Which signal we ignored: from the tracking data, up to three signals that were in plain sight before the final result, with the date each appeared, earliest first.
4. What we keep: what worked and should stay.
5. What changes next time: one concrete, verifiable rule for each signal in point 3, not an intention.
6. Who owns the change: one role per rule (I will add the name) and a suggested effective date.

Output format: a table with the columns Field · Content, followed by three questions the facilitator should ask in the session to confirm or discard the signals. Maximum 400 words. No introduction.

Rules:
- Do not invent figures or dates. If data is missing, write [DATA PENDING] and tell me where to look for it.
- Do not attribute the result to people or to "lack of commitment"; describe decisions, data and moments.
- Mark every interpretation as an interpretation.
- If critical information is missing, ask me at most 3 questions before answering.
- Answer in plain, direct business English, with no sales adjectives.
```

What to check in the answer:

- Does each signal have a date and a data point, or did it settle for "adoption was low"?
- Can the rules in point 5 be checked on day one of the next initiative, or are they good intentions?
- Did it name a person as a cause? If so, ask it to rewrite that row in terms of decision and data.