Prompt Chaining: Turn Raw Project Data Into a Full Reporting Package in One Session

What This Builds

A repeatable four-step prompt sequence that takes raw project data and produces four distinct outputs in a single session: (1) a full weekly status report, (2) a one-page executive summary, (3) a risk update for the steering committee, and (4) a team-facing Slack message. Each step feeds into the next. You run the chain once, and your entire Monday reporting package is done in under 20 minutes — instead of assembling four separate documents from scratch.

Prerequisites

• Comfortable with Level 1 AI prompting for individual PM documents
• ChatGPT Plus ($20/mo) or Claude Pro ($20/mo) — needed for longer context and better instruction-following
• A structured way to collect your weekly project data before you start (5 minutes of prep the Friday before works well)

The Concept

A prompt chain is an assembly line for your brain. Imagine a factory where the first station stamps out a steel frame, the second station welds on the parts, and the third station paints the finished product. Each station does one specific job and passes the output to the next. A prompt chain works the same way — instead of asking the AI to do everything at once (which produces mediocre results), you ask it to do one focused task, take the result, and hand it to the next prompt as input. The AI builds on what it just produced rather than starting over. The output gets more refined and more tailored with each step.

Build It Step by Step

Part 1: Prepare Your Data Template (Do This Once)

Before you run the chain, you need your raw data in a consistent format. Create a simple text template — keep it in Notion, OneNote, or a recurring Friday calendar block — and fill it out each week in 5–10 minutes:

## Weekly Project Data — [Project Name] — Week of [Date]

**Project overview (for context):**
- Project: [Name and one-sentence description]
- Client: [Client name]
- Phase: [Current phase]
- Overall timeline: [Start → End, current status]

**This week's completions:**
- [Task/milestone completed]
- [Task/milestone completed]
- [Task/milestone completed]

**In progress:**
- [Task, owner, % complete]
- [Task, owner, % complete]

**Blocked or at risk:**
- [Item, reason, who needs to act]
- [Item, reason, who needs to act]

**Upcoming milestones (next 2 weeks):**
- [Milestone, date]
- [Milestone, date]

**New risks or issues this week:**
- [Describe]

**Budget this week:**
- Spent to date: $[X] of $[Y] total ([Z]%)
- Timeline completion: [A]% of project complete

**Key decisions made:**
- [Decision, who made it, date]

**Open decisions needed:**
- [Decision, owner, deadline]

Part 2: Run the Four-Step Chain

Open a new conversation in Claude or ChatGPT. Run these prompts in sequence, in the same conversation window. Do not start a new conversation between steps — the AI needs the full history to build each output on the last.

Prompt 1 — Full Status Report

Paste your completed data template, then add:

You are an experienced project manager writing a weekly status report. Using the project data above, write a complete weekly status report for a non-technical executive audience. Format:

**Overall Status:** [Red / Amber / Green] — [one-sentence justification]

**Accomplishments This Week:**
- [3–4 bullets]

**Upcoming Milestones:**
- [3 bullets with dates]

**Risks and Issues:**
| # | Description | Probability | Impact | Status | Owner | Mitigation |
|---|---|---|---|---|---|---|
[fill in 3–5 rows]

**Budget Summary:**
- Spend: $X of $Y ([Z]%)
- Schedule: [on track / behind / ahead]
- Commentary: [one sentence]

**Decisions Needed:**
- [item, owner, by when]

Write in first-person plural ("we"). Tone: confident, direct, solutions-oriented. Audience: client VP-level and above.

Wait for the output. Read it. If any section is wrong, say "Revise the risks section — the data migration risk should be Red, not Amber, because it has a hard dependency on the go-live date." Fix it before moving on.

Prompt 2 — Executive One-Pager

Now take the status report you just wrote and create a separate, shorter executive summary. This version is for a C-suite audience who will spend 60 seconds reading it. Format:

**[Project Name] — Week of [Date]**
**Overall: [Red/Amber/Green]**

**Bottom line:** [Two sentences maximum — what's the overall health and what's the one thing they need to know]

**3 things going well:**
- [bullet]
- [bullet]
- [bullet]

**1 thing that needs attention:**
[One paragraph, max 4 sentences: what's the issue, why it matters, what you're doing about it, what decision (if any) is needed from them]

**Next milestone:** [Name, date]

Prompt 3 — Steering Committee Risk Briefing

Using the risks from the full status report, create a risk briefing slide narrative for the steering committee. This is not a slide deck — it's the speaker notes and talking points a PM would use when presenting the risk section. Format:

**Risk Briefing — [Project Name] — [Date]**

For each risk (Red and Amber only):
- **[Risk name]** ([Red/Amber])
  - *What it is:* [one sentence]
  - *Why it matters now:* [one sentence — what happens if unresolved]
  - *What we're doing:* [one sentence]
  - *What we need from this group:* [specific ask or "no action needed — for awareness"]

After all risks: **Summary ask:** [One sentence summarizing what you need the steering committee to decide or approve today]

Prompt 4 — Team Slack Message

Finally, write a brief team-facing project update for Slack. This is internal — for the delivery team, not the client. Tone: friendly, direct, motivating. Format:

:wave: **Weekly Update — [Project Name]**

**Status:** [Green/Amber/Red emoji + one word]
**Week in review:** [2–3 sentences — what the team accomplished, honest but positive]
**Watch-outs for next week:** [2 bullets — what the team needs to stay focused on]
**Shoutout:** [Optional: call out one person or team for something specific this week]
**Next milestone:** [Date and what it is]

Keep it under 150 words. This should feel like a PM talking to a team, not a corporate memo.

Part 3: Review and Distribute

After running all four prompts, you have four drafts in the same conversation. Review each:

• Status report: add any political context the AI can't know (e.g., "the client is sensitive about the budget number — soften the language in the budget section")
• Executive one-pager: check that the "bottom line" is accurate and matches your verbal message to the client
• Steering committee notes: confirm the asks are actually what you need
• Slack message: personalize the shoutout and check the tone matches your team culture

Copy each output to its destination (email, SharePoint, Confluence, Slack). Total review time: 10–15 minutes.

Real Example: Infrastructure Migration Project

Setup: 8-month cloud migration for a regional bank, week 19 of 34, one PM running three workstreams.

Input data (filled template): Network migration 80% complete, security config review blocked pending client firewall approval (Day 12 of block), UAT environment stood up successfully, two new P2 bugs found in payment gateway testing, go-live date at risk if firewall approval doesn't come by Friday.

Prompt 1 output: Full RAG status report — Amber, 5-row risk table, budget at 54% spend / 56% timeline (tracking well).

Prompt 2 output: C-suite one-pager — "The project is on track in all workstreams except one: firewall approval. If the bank's security team does not approve the configuration by this Friday, the go-live date moves. We need your help escalating internally."

Prompt 3 output: Steering committee briefing with two items — firewall approval (Red, escalation ask) and payment gateway bugs (Amber, dev team handling, no action needed).

Prompt 4 output: Slack message calling out the network team for hitting their milestone and flagging the firewall as the one thing to watch.

Total session time: 22 minutes including review and distribution. Previous manual approach: ~2.5 hours assembling four separate documents.

What to Do When It Breaks

• Prompt 2 contradicts Prompt 1 → The AI sometimes drifts if the conversation is long. Say "In your one-pager, you said the budget was on track, but the status report said we're 3% over plan. Fix the one-pager to be consistent with the status report." Consistency checks are your job — the AI doesn't always self-monitor.
• The chain loses context after Prompt 3 → This happens with very long conversations. Fix: keep your data template short (under 600 words). Paste only the essentials — you can always add detail with a follow-up message.
• The risk table in Prompt 1 has the wrong columns or format → Edit the prompt to match your firm's risk register format. If your RAID log uses different headings, change the table headers in Prompt 1 exactly as you want them.
• Tone is too formal for the Slack message → Add to Prompt 4: "Our team Slack culture is casual — we use humor and colloquial language. Make this sound like a human, not a corporate memo." You may need to add an example: "Here's a message I wrote two weeks ago: [paste example]."

Variations

• Simpler version: Run only Prompts 1 and 4 — the full status report and the Slack message. Skip the executive one-pager and steering committee notes until you have a client meeting. Two outputs, 12 minutes.
• Extended version: Add a fifth prompt after Prompt 3: "Based on the risks we've discussed, generate 5 questions a steering committee member is likely to ask me, and draft my answers to each." Use these to prep for the meeting.
• Multi-project version: Run Prompts 1 through 4 for each project separately, then add a final prompt: "I run three projects. Here are their status reports: [paste all three]. Write a portfolio-level summary for my director showing cross-project health, shared risks, and resource conflicts."

What to Do Next

• This week: Run the full four-step chain on your current project. Time yourself. Compare to how long the manual version takes.
• This month: Refine each prompt based on what the AI gets wrong in the first 2–3 runs. Keep your final prompt versions in a text file or Notion page — this is your personal PM prompt library.
• Advanced: Store your finalized prompts in a Claude Project (see guide: "Claude Project: Build a Persistent PM Assistant That Knows Your Project"). The Project holds your project context; these prompts handle the weekly reporting cadence. Combine them for maximum efficiency.

Advanced guide for project manager professionals. These techniques use more sophisticated AI features that may require paid subscriptions.