Skill Hub — AI-Assisted Analysis

AI-Assisted Status Reporting

Draft a structured project status update using AI to synthesize inputs, then review and correct it before sending it to stakeholders.

What this skill is for

Status reports require aggregating data from tickets, chat threads, meeting notes, and planning tools into a coherent narrative that tells the truth about progress, risks, and blockers. AI can synthesize these inputs quickly, but it often softens bad news, uses outdated data, and hides blockers behind generic language. This skill provides a workflow where AI drafts the status update and a human reviews it for accuracy, honesty, and actionability before it is sent. The output is an AI-Drafted Status Update with human review marks that show what was corrected, what was added, and what was removed.

When to use this skill

A weekly or monthly status update is due and the inputs are scattered across multiple systems and conversations.
A steering committee or project sponsor needs a high-level status summary and the team wants to reduce drafting time.
A sprint or phase review requires a structured progress report with metrics, risks, and blockers.
The project has multiple workstreams and the status report must aggregate progress from each without losing critical detail.
An incident or support team needs a periodic status report on ticket volumes, resolution times, and recurring patterns.
The same status format is used repeatedly and the team wants to automate the structural synthesis while keeping human review.

Real work situations

SAP project has three workstreams and scattered updates

An SAP implementation has workstreams for data migration, integration, and custom development. Updates exist in Jira, Confluence, Slack threads, and meeting notes. AI synthesizes these inputs into a single status update. The human reviewer discovers that the AI used last week's ticket numbers because the current export was not included in the prompt, softened the integration delay from "two weeks behind" to "slightly behind schedule," and omitted the blocker that the sandbox environment is down. The reviewer updates the metrics, corrects the timeline language, and adds the blocker with the owner and expected resolution date.

AMS support team needs weekly ticket metrics

A SAP AMS support team must produce a weekly status report for the client showing ticket volume, average resolution time, reopened tickets, and escalations. AI generates a draft from the ticketing system export. The human reviewer finds that the AI counted duplicate tickets created by the auto-routing system as separate incidents, did not flag that one critical ticket has been open for eight days without assignment, and used a template that refers to a KPI that the client no longer cares about. The reviewer removes duplicates, adds the critical ticket as a red flag, and replaces the outdated KPI with the current client priority.

Integration project has red flags buried in chat threads

A middleware integration project is in testing. The team communicates in Slack. The AI draft of the status report extracts progress from the project management tool but does not read the chat threads. The human reviewer checks the threads and finds that a failed test on Tuesday was never logged as a ticket, that the target system's certificate expires in ten days, and that the business partner has not provided the test data they promised. The reviewer adds all three items to the status report, marks the certificate issue as a red flag, and assigns the data delay to the business relationship manager for escalation.

Inputs required

Ticket or task data — the current status of work items from the project management or ticketing system.
Meeting notes or action logs — decisions, blockers, and risks from recent meetings.
Chat or email threads — informal updates that may contain critical issues not logged in formal systems.
Previous status update — to check for continuity, follow-ups, and changes in trend.
Plan or milestone list — the baseline against which progress is measured.
Stakeholder expectations — what the audience cares about, what format they expect, and what level of detail is appropriate.

Questions to ask

What is complete, what is in progress, and what is blocked?
What has changed since the last status update, and are the trends improving or worsening?
What risks or blockers are new, and what happened to the risks from the previous update?
Does the AI use current data, or is it relying on outdated exports or old meeting notes?
Does the AI soften or hide bad news behind generic language such as "on track" or "minor delay"?
Are all metrics correct, and do they match the source systems?
Does the status update include specific next steps, or is it only a backward-looking summary?

Working method

Gather the inputs. Collect the latest ticket data, meeting notes, chat threads, and the previous status update. Verify that the data is current and complete.
Define the report scope. Decide what the audience needs to know, what format to use, and what level of detail is appropriate.
Run the AI synthesis. Prompt the AI to draft a status update that covers progress, risks, blockers, and next steps, using the gathered inputs.
Capture the AI draft. Save the AI output as a draft. Do not send it yet.
Check data accuracy. Verify every metric, date, and status against the source systems. Correct any outdated or incorrect data.
Check for honesty. Read the draft for generic language that hides bad news. Replace soft language with specific facts.
Check for completeness. Ensure that new blockers, risks, and escalations are included, and that follow-ups from the previous update are addressed.
Check for actionability. Ensure that the update includes specific next steps with owners and deadlines, not just a summary of what happened.
Produce the validated status update. Output the corrected status update with human review marks showing what was changed and why.
Send and archive. Distribute the update to the intended audience and archive it alongside the inputs and the review log.

Decision rules

If the AI uses outdated data, update it with the current source data before sending the report.
If the AI omits a blocker or risk, add it and mark it with the correct severity and owner.
If the AI softens bad news, replace the soft language with specific facts and timelines.
If the AI invents a metric or status that is not in the source data, remove it and log the error.
If a trend is worsening, state it clearly rather than using the same language as the previous update.
If the report is for a client or sponsor, apply a stricter honesty and accuracy standard than for internal team updates.
If the AI duplicates content from the previous update without showing progress, replace it with current information.

Deliverables

AI-Drafted Status Update — the raw AI output, preserved for traceability.
Validated Status Update — the corrected and honest status report with human review marks.
Review Log — a record of data corrections, tone corrections, and additions made by the human reviewer.
Archive Package — the inputs, the AI draft, the validated update, and the review log stored together.

Templates

AI-Drafted Status Update

---
project: [project name]
period: [start date] to [end date]
report_date: [YYYY-MM-DD]
ai_tool: [tool name]
reviewer: [name]
---

## Executive Summary
- **Overall status**: [Green / Yellow / Red] — [one-sentence reason]
- **AI draft**: [attached]
- **Human review**: [Completed / In progress]

## Workstream Progress
| Workstream | Status | Complete | In Progress | Blocked | Notes |
|--------------|--------|----------|-------------|---------|-------|
| [name] | [G/Y/R] | [n/m] | [n/m] | [n/m] | [human-reviewed note] |

## Key Metrics
| Metric | This Period | Last Period | Trend | Source | Verified |
|--------|-------------|-------------|-------|--------|----------|
| [name] | [value] | [value] | [up/down/flat] | [system] | [Yes / No] |

## Risks and Blockers
| # | Item | Severity | Owner | Impact | Resolution Target | Status |
|---|------|----------|-------|--------|-------------------|--------|
| 1 | [description] | [High / Med / Low] | [name] | [what happens if unresolved] | [date] | [Open / Mitigated] |

## Next Steps
| # | Action | Owner | Deadline | Dependency |
|---|--------|-------|----------|------------|
| 1 | [description] | [name] | [YYYY-MM-DD] | [none / item #] |

## Human Review Log
| # | AI Original | Human Correction | Reason |
|---|-------------|------------------|--------|
| 1 | [what AI wrote] | [what human changed] | [outdated / soft / invented / missing] |

Quality checklist

[ ] Every metric is verified against the source system and is current.
[ ] No bad news is hidden behind generic or softened language.
[ ] Every blocker and risk has a severity, an owner, and a resolution target.
[ ] Every follow-up from the previous status update is addressed or explicitly deferred.
[ ] The update includes specific next steps with owners and deadlines.
[ ] The human review log contains at least one entry or a note that no corrections were needed.
[ ] The update can be understood by someone who has not been following the project daily.

Common mistakes

Mistake: Sending the AI draft without checking the data currency. Consequence: The report shows last week's ticket count as current, creating a false impression of progress and eroding trust when the discrepancy is discovered.
Mistake: Allowing the AI to soften language about delays and blockers. Consequence: A "two weeks behind" delay is reported as "slightly behind schedule," and the sponsor does not provide the support needed to recover until the delay becomes critical.
Mistake: Omitting informal blockers that were only discussed in chat. Consequence: A certificate expiration, a missing test data commitment, or an unlogged test failure is not visible to leadership until it causes an outage.
Mistake: Not archiving the inputs and review log, so the same corrections are repeated every reporting cycle. Consequence: The team never learns which inputs the AI misses or how to improve the prompt, wasting time on repeated manual fixes.

Weak output vs Strong output

Weak output — bad AI usage

A project manager uses AI to draft a weekly status update for an SAP AMS client. The AI produces a report that says "ticket volume is stable, average resolution time is within SLA, and the project is on track." The human reviewer does not check the data and sends the report. In reality, the ticket volume has increased twenty percent, the average resolution time is at the SLA boundary, and one critical ticket has been unassigned for six days. The client notices the unassigned ticket independently and questions the accuracy of the entire report. The project manager has no review log to explain what happened and must rebuild credibility over the next two weeks.

Strong output — good AI usage

## Validated Status Update — SAP AMS Support, Week 24

### Key Metrics
| Metric | This Week | Last Week | Trend | Verified |
|--------|-----------|-----------|-------|----------|
| Tickets opened | 47 | 39 | Up 20% | Yes — ServiceNow export 2026-06-12 |
| Avg resolution time | 4.8h | 4.2h | Up 14% | Yes — SLA threshold is 5.0h |
| Critical tickets > 5 days | 1 | 0 | Up | Yes — INC-2847 unassigned since 2026-06-06 |

### Risks and Blockers
| # | Item | Severity | Owner | Resolution Target |
|---|------|----------|-------|-------------------|
| 1 | INC-2847 unassigned for 6 days | High | Support Lead | 2026-06-13 |
| 2 | Ticket volume trend up 20% for 2 weeks | Medium | Account Manager | 2026-06-20 |

### Human Review Log
| # | AI Original | Correction | Reason |
|---|-------------|------------|--------|
| 1 | "ticket volume is stable" | "ticket volume up 20%" | AI used outdated export |
| 2 | "project is on track" | added red flag INC-2847 | AI omitted critical unassigned ticket |
| 3 | "avg resolution time within SLA" | "avg resolution time at SLA boundary, 4.8h" | soft language corrected |

Agent instructions

AI Prompt Pattern

Role: You are a status reporting assistant.
Context: I need a status update for [project / team / period]. The inputs are attached: ticket data, meeting notes, chat threads, and the previous status update. The audience is [stakeholder type] and they care about [metrics / risks / blockers / next steps].
Tasks:
1. Synthesize the inputs into a structured status update with: executive summary, workstream progress, key metrics, risks and blockers, and next steps.
2. Use specific numbers and dates. Do not use generic phrases like "on track" or "progressing well" without supporting data.
3. Flag any metric that has changed by more than 10% from the previous period.
4. List every blocker and risk with a severity label and an owner if known.
5. Include next steps with owners and deadlines.
6. If the inputs are missing data for a section, state "data not available" rather than omitting the section or guessing.
Constraints: Do not soften bad news. Do not omit blockers because they are only mentioned in chat. Do not use data that is not in the inputs. Do not invent metrics or owners.

Agent dos

Ask for the latest source data and the previous status update before generating the draft.
Verify every metric against the source system before approving the report.
Replace soft language with specific facts and numbers.
Archive the inputs, the AI draft, and the review log together for audit and improvement.

Agent don'ts

Do not send the AI draft without verifying data currency and accuracy.
Do not allow generic language like "on track" or "minor delay" without specific supporting data.
Do not omit blockers that were discussed in informal channels.
Do not invent metrics, trends, or owners that are not in the source material.

Related skills

AI Output Review — general review workflow for any AI-generated output.
AI-Assisted Meeting Synthesis — extracting structured decisions and actions from meetings that feed into status reports.
AI Accountability — defining ownership and review gates for AI-generated reports.
Productivity and Execution Control — the broader skill group for daily execution, follow-up, and delivery tracking.

Related Atlas pages

SAP Process Audit — structured diagnostic frame that informs status reporting on process health.
Operational Memory for SAP AMS — building reviewable operational records from periodic status updates.

Verification status and limitations

This skill is a public working interpretation of AI-assisted status reporting. It is not official project management, SAP, or governance framework documentation. The accuracy of the status update depends on the currency and completeness of the inputs provided to the AI. AI tools may not have access to real-time system data or informal chat channels. The human review step is mandatory and must include a verification of data currency, metric accuracy, and tone honesty. Use this skill as a structured starting point, not as a replacement for project management judgment or stakeholder communication.