AI-assisted workflows, prompts, and templates for TPM productivity and program management.
This is not a prompt dump. It is a working library for the parts of TPM work that are repetitive, messy, and easy to make worse with vague language: weekly status, executive updates, meeting notes, Jira hygiene, risk reviews, escalation drafts, launch readiness, customer signals, and program reviews.
The point is not to replace judgment. The point is to remove some of the mechanical work so TPMs have more room for the work that actually requires judgment: clarifying ambiguity, surfacing risk, aligning teams, making tradeoffs, and helping leaders make better decisions.
AI can help structure the first draft. The TPM still owns the facts, risk call, decision ask, and final artifact.
This repo contains practical AI workflows for TPMs working across Google Workspace, Slack, Salesforce, Jira, and normal meeting-note chaos.
The bar for inclusion is higher than having a prompt that worked once. A prompt that only works for one person in one meeting is not a tool yet. A tool is documented, reusable, safe to adapt, and clear enough that another TPM can use it without needing the backstory.
Some files are more mature than others. Treat this as a working library: useful now, but still subject to hard editing as prompts survive real inputs.
This repo is currently focused on AI-assisted TPM workflows for:
- Weekly status synthesis
- Executive program updates
- Meeting notes to action items
- Jira, Slack, Salesforce, and Google Workspace cleanup
- Risk and dependency reviews
- Escalation drafts
- Launch readiness reviews
- Customer-signal synthesis
- Director-level artifact critique
The strongest materials are the weekly status workflow, executive program update prompt, meeting-notes workflow, artifact critique prompt, safety guidance, and prompt quality rubric.
This repo is not trying to be a general AI prompt library. Prompts belong here only if they help a TPM make a better decision, reduce execution risk, clarify ownership, improve reporting, or turn scattered inputs into a usable program artifact.
Read these first:
- Safety and data handling
- Operating principles
- Standard context block
- Weekly status workflow
- Executive program update prompt
- Director artifact critique prompt
- Prompt quality rubric
The safety file should come first. Bad inputs create bad outputs, and sometimes they create data handling problems no one needed.
Start with these. They are the clearest examples of how this repo is meant to be used.
| Workflow | Use it for | Why it matters |
|---|---|---|
| Weekly status workflow | Turning Jira, Slack, meeting notes, and RAID updates into a weekly status draft | Keeps status focused on what changed, what is blocked, and what decision is needed |
| Executive program update | Preparing director or VP-level updates from sanitized program context | Separates facts, assumptions, risks, decisions, and asks |
| Meeting notes to actions | Converting messy meeting notes into decisions, owners, due dates, and open questions | Fixes the handoff problem after the meeting ends |
| Artifact critique | Reviewing TPM artifacts before leadership review | Makes the human review bar explicit before the artifact goes upstairs |
The first release focuses on common TPM workflows.
- Weekly status reporting
- Meeting notes to action items
- Jira hygiene reviews
- Slack thread summaries
- RAID log creation and cleanup
- Standup prep
- Acceptance criteria cleanup
- Executive program updates
- Decision memos
- Escalation drafts
- QBR summaries
- Director-level artifact reviews
- Dependency mapping
- Risk burndown planning
- Technical tradeoff analysis
- Launch readiness reviews
- Compliance intake workflows
- Customer escalation workflows
- Program strategy reviews
- Operating model design
- Influence without authority planning
- Metrics and instrumentation planning
- Blameless retrospectives
TPMs spend a lot of time turning scattered inputs into usable program artifacts.
A single weekly update might pull signal from Jira epics, Slack threads, Google Docs, Sheets, Salesforce cases, meeting notes, dashboards, and a few hallway conversations that never quite made it into a system of record.
The writing is not always the hard part. The hard part is figuring out what changed, what is real, what is assumed, what is blocked, and what decision needs to happen next.
This repo is meant to help with that middle step.
AI can get a first pass into shape quickly. The TPM still owns the judgment. That is where the job is.
A TPM needs a weekly executive update.
The raw material is spread across:
- Jira epic summaries
- A Slack blocker thread
- Google Doc meeting notes
- Salesforce customer escalation notes
- The current RAID log
The TPM uses:
The output should make these things clear:
- Current program health
- What changed this week
- What risk increased or decreased
- What is blocked
- What decision is needed
- Who owns the next action
The prompt does not decide the status. It helps organize the evidence so the TPM can make the call cleanly.
ai-automations/
README.md
operating-principles.md
safety-and-data-handling.md
prompt-quality-rubric.md
REPO_MANIFEST.md
prompts/
README.md
standard-context-block.md
junior-tpm/
README.md
weekly-status-update.md
meeting-notes-to-actions.md
jira-hygiene-review.md
slack-thread-summary.md
raid-log-builder.md
standup-prep.md
acceptance-criteria-builder.md
senior-tpm/
README.md
executive-program-update.md
decision-memo.md
escalation-draft.md
dependency-map.md
risk-burndown-plan.md
technical-tradeoff-analysis.md
launch-readiness-review.md
qbr-summary.md
director-review/
README.md
artifact-critique.md
program-strategy-review.md
what-am-i-missing.md
director-level-program-review.md
influence-without-authority-plan.md
operating-model-design.md
metrics-and-instrumentation-plan.md
blameless-retrospective.md
tools/
README.md
google-workspace.md
slack.md
salesforce.md
jira.md
workflows/
README.md
weekly-status-workflow.md
customer-escalation-workflow.md
launch-readiness-workflow.md
compliance-intake-workflow.md
executive-review-workflow.md
templates/
README.md
weekly-status-input-template.md
salesforce-signal-intake-template.md
jira-export-field-list.md
examples/
README.md
sanitized-weekly-status-example.md
sanitized-escalation-example.md
sanitized-decision-memo-example.md
sanitized-launch-readiness-example.md
Start with the standard context block. Then paste the specific prompt below it.
A useful TPM prompt usually needs:
- Program objective
- Customer or business impact
- Audience
- Current phase
- Milestone or target date
- Jira context
- Slack context
- Google Doc, Sheet, Slide, or meeting-note context
- Salesforce customer signal context, when relevant and approved
- Known risks
- Known blockers
- Decision needed
- Desired output format
Do not use these prompts as a replacement for validation. Use them to get to a cleaner first draft, identify gaps, and make the next conversation sharper.
1. Gather the relevant source material.
2. Remove or anonymize sensitive information.
3. Add the standard context block.
4. Run the relevant prompt.
5. Review the output for accuracy, tone, missing context, risk language, and ownership.
6. Convert the output into the final artifact.
7. Publish in the right system of record.
8. Update Jira, Slack, Salesforce, or Google Workspace links as needed.
A draft is not done until the system of record is updated. The prettiest status update in the world does not help much if Jira still says the blocked work is green.
Good output does not just sound polished. It helps the reader act.
A good TPM artifact should answer:
- What changed?
- Why does it matter?
- What is at risk?
- Who owns the next step?
- What decision is needed?
- What happens if we do nothing?
That last question matters. A lot of program risk hides in silence.
Do not paste sensitive data into unapproved tools.
That includes customer names, contract terms, renewal details, account notes, secrets, certificates, private keys, internal architecture details, vulnerability details, source code, production logs, legal content, restricted incident details, employee data, personal data, health data, payment data, and unreleased roadmap commitments.
When in doubt, reduce the input to the minimum useful shape:
Customer A in the financial services segment has a renewal risk tied to delayed delivery of Capability X. The exact account name, revenue figure, and contract language have been removed.
That is usually enough for a useful program summary without carrying data you do not need.
This repo breaks when the prompts are treated as authority instead of drafting aids.
Common failure modes:
- The source data is stale, incomplete, or political.
- Slack opinions get treated like confirmed facts.
- Jira says Green because no one updated the ticket.
- A polished executive summary hides a missing decision.
- The model invents certainty because the prompt did not force assumptions into the open.
- A TPM ships the draft without checking owners, dates, metrics, and risks.
The dangerous output is not the obviously bad one. The dangerous output is clean, confident, and wrong.
This repo sits next to the broader TPM materials:
tpm-templatesfor reusable program artifactsprogram-reporting-frameworksfor executive reporting patternssecurity-program-playbooksfor security and infrastructure program executionlearning-notesfor technical TPM background and system design concepts
This repo is the AI layer. It helps turn messy inputs into better TPM artifacts faster.
A new prompt or workflow should be added only if it meets this standard:
- It solves a real TPM problem.
- It explains when to use it.
- It names the inputs needed.
- It includes the prompt or workflow steps.
- It includes example input and output, or links to a sanitized example.
- It explains where the prompt fails.
- It includes human review checks.
- It has data handling notes.
- It helps someone make a decision or reduce risk.
If a prompt only makes something sound nicer, it probably does not belong here yet. Nice is fine. Useful is the bar.