Use the Builder when you're starting from scratch. Use the Improver when you've tried something and want to make it sharper. Both work with Copilot, Claude, or ChatGPT.
Tool One
The Prompt Builder
An interview-style prompt that asks six questions about your task, then writes a polished, ready-to-paste prompt for your specific situation. Teaches prompt structure by doing.
Works with Copilot, Claude, or ChatGPT5 minutes
You are going to help me build the most effective prompt for AI work I need to do. Ask me one question at a time. Wait for my answer before moving on.
Ask in this order:
1. What are you trying to produce? (draft, summary, list, analysis, something else)
2. Who is this for? (audience, role, what they already know, what they care about)
3. What's the goal? (what should happen after they read or hear it)
4. Do you have source material I should work from, or are we building from scratch? (paste it if you do)
5. Any constraints? (length, tone, format, voice, deadline, what to avoid)
6. What tool are you using? (Copilot in M365, Claude, ChatGPT, NotebookLM)
Once I've answered all six, write me a polished prompt I can paste. Include:
- A clear role for the AI
- The context and constraints I gave you
- A specific request with the desired output format
- Any guardrails (data sensitivity, voice notes, what to skip)
- As the final line of the prompt you generate for me, instruct the AI to add this single italicized line at the end of its response: *This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
After the prompt, give me one line on why this tool fits, and one line on what data I should NOT paste based on the Diocesan AI policy.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Tool Two
The Prompt Improver
Paste a prompt you've already written. The Improver diagnoses what's weak, asks clarifying questions to fill the biggest gaps, and rewrites it. Use after the Builder, or any time output isn't quite right.
Works with Copilot, Claude, or ChatGPT3 minutes
You are a prompt engineer helping me improve a prompt I've already written.
Here is my prompt:
[paste your prompt]
Here is the output I got when I tried it, if I have one:
[paste output, or skip]
Do these four things:
1. Diagnose. Tell me what's strong and what's weak. Look specifically for missing role for the AI, vague or absent audience, no output format specified, missing constraints (length, tone, voice), no source material when it would help, missing guardrails (data sensitivity, what to skip), ambiguous goal.
2. Ask me up to three clarifying questions to fill the biggest gaps. Wait for my answers before moving on.
3. Rewrite my prompt with improvements. Explain in 2-3 bullets what changed and why.
4. Suggest one variation: a sharper or more specific version I could try if the rewrite is not quite right. If a different AI tool would suit this prompt better than the one I'm using, say which and why.
Keep it simple. The best prompt is the simplest one that gets the right output.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
II · The Curriculum
Ten lessons in practical AI.
Each lesson is a prompt you can use today and a skill you'll carry forward. Search, filter by tool, and open any lesson to see the full prompt, sample output, and governance notes.
Tool
i
The Prompt Builder
Any AIAll roles
Interviews you about your task and writes a polished prompt for your specific situation. The entry point to the whole curriculum.
What you'll learn: the structure of every good prompt — role, context, audience, source, constraints, output format, guardrails.
The Prompt
You are going to help me build the most effective prompt for AI work I need to do. Ask me one question at a time. Wait for my answer before moving on.
Ask in this order:
1. What are you trying to produce? (draft, summary, list, analysis, something else)
2. Who is this for? (audience, role, what they already know, what they care about)
3. What's the goal? (what should happen after they read or hear it)
4. Do you have source material I should work from, or are we building from scratch? (paste it if you do)
5. Any constraints? (length, tone, format, voice, deadline, what to avoid)
6. What tool are you using? (Copilot in M365, Claude, ChatGPT, NotebookLM)
Once I've answered all six, write me a polished prompt I can paste. Include:
- A clear role for the AI
- The context and constraints I gave you
- A specific request with the desired output format
- Any guardrails (data sensitivity, voice notes, what to skip)
- As the final line of the prompt you generate for me, instruct the AI to add this single italicized line at the end of its response: *This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
After the prompt, give me one line on why this tool fits, and one line on what data I should NOT paste based on the Diocesan AI policy.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
The AI asks each question in turn, waits for your answer, then produces a multi-paragraph prompt tailored to exactly what you need. Most users get their answer in three to five minutes.
Governance
Safe. The Builder asks questions; it does not process sensitive data directly. Do not paste confidential source material until the final prompt is ready and you have chosen a tool appropriate to the data sensitivity.
ii
The Inbox Waiting List
CopilotAll roles
Copilot scans your inbox and builds two lists: threads where you're the blocker, and threads where you're waiting on someone else. Sorted by age, oldest first.
What you'll learn: Copilot sees across your inbox in a way no human can. This is its superpower, and it has nothing to do with writing.
The Prompt (paste into Copilot in Outlook or Teams)
Scan my Outlook inbox from the last 30 days. Build two lists.
List one: Threads where I am the blocker. For each, show the original ask, the sender, the date received, and how long it has been waiting.
List two: Threads where I am waiting on someone else. For each, show what I asked for, the recipient, the date I asked, and how long it has been waiting.
Sort each list by age, oldest first. Skip threads that are clearly resolved or where no response is expected.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
Two columns: 8 to 12 threads in each, sorted oldest first. Ready to triage in 10 minutes on Monday morning.
Governance
Safe. Tenant-grounded. No external data exposure.
iii
The Pre-Meeting Brief
CopilotAll roles
Before your next meeting, get a one-page brief pulling email history, Teams chats, shared files, and open action items between you and the other person.
What you'll learn: cross-surface synthesis. Copilot pulls from mail, calendar, Teams, and files at once.
The Prompt
I have a meeting with [person] at [time/date]. Build me a pre-meeting brief that includes:
1. Our last five email exchanges, summarized.
2. Any Teams chat messages between us in the last 60 days.
3. Any files we have co-edited or shared.
4. Any open action items between us from past meetings.
5. Anything in my calendar suggesting why this meeting was scheduled.
Tell me what is likely to come up. Suggest two or three questions I should be ready for.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
A one-page brief, scannable in 90 seconds before you walk in.
Governance
Safe. Tenant-grounded.
iv
The Vacation Catch-Up
CopilotAll roles
Returning from time off? Copilot summarizes the meetings you skipped, the emails that need a response, and ranks everything by urgency.
What you'll learn: prioritization as a prompt instruction. AI can rank, not just summarize, if you ask it to.
The Prompt
I was out of office from [start date] through [end date]. Summarize what I missed across:
1. Meetings I skipped, with key decisions made and any action items assigned to me.
2. Emails that need a response from me, ranked by urgency.
3. Any messages flagged as important or marked high priority.
Group everything by urgency: must respond today, this week, can wait. For each item, give me a one-sentence summary and a suggested next action.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
A triaged catch-up list. What to do first, what can wait, what is already resolved.
Governance
Safe. Tenant-grounded.
v
The Commitment Tracker
CopilotAll roles
Find every promise you made in sent email, Teams chats, and meeting transcripts. Flags the ones you haven't followed through on.
What you'll learn: AI as accountability partner, not just writer.
The Prompt
Search my sent email, Teams chats, and meeting transcripts from the last 30 days for anything I said I would do, send, follow up on, or get back to someone about.
For each commitment, give me:
- What I said I would do
- Who I said it to
- The date I said it
- Whether there is evidence I followed through
Flag any commitments older than 7 days with no evidence of follow-through.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
A list of your own broken promises, surfaced before someone has to ask.
Governance
Safe. Tenant-grounded.
vi
The Multi-Source Briefing
NotebookLMAll roles
Upload diocesan policy, prior meeting notes, recent correspondence, and background reading. NotebookLM synthesizes with citations and can generate a podcast-style Audio Overview.
What you'll learn: when to switch tools. NotebookLM beats Copilot and Claude for source synthesis with citations.
The Prompt
Step 1. Create a new notebook in NotebookLM and upload your sources:
- The diocesan policy or guideline at hand
- Prior meeting minutes or notes on the topic
- Recent correspondence on the topic
- Any background reading or whitepapers
Step 2. Paste this prompt into the chat:
"Synthesize the key themes across these sources. Where do they agree? Where do they disagree? What questions remain unanswered? Cite specific passages from the sources, not from general knowledge."
Step 3. For a deeper treatment, generate an Audio Overview and listen on the commute or while walking. Two AI hosts discuss the material as a podcast.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
A citation-grounded synthesis of 5 to 10 sources. Audio Overview runs 10 to 20 minutes.
Governance
Use care. Notebook contents are private to your Google account. Do not upload PII, sealed records, or sacramental information.
vii
The Pastoral Spanish Translator
ClaudePastor · Office · Comms
Translation with pastoral sensitivity. Latin American Spanish, preserving liturgical terms, with translator's notes flagging anything that needs native-speaker review.
What you'll learn: translation is not substitution. Voice and register matter.
The Prompt
You are a translator with pastoral sensitivity and deep familiarity with the Catholic faith and the Spanish-speaking communities of the Diocese of Bridgeport.
Translate the following text into Spanish suitable for [parish bulletin / pastoral letter / sacramental announcement]:
[paste English text]
Constraints:
- Use a warm, pastoral register, not corporate.
- Use Latin American Spanish vocabulary preferred in the Bridgeport Hispanic community.
- Preserve any liturgical or theological terms accurately.
- Where idioms or culturally specific phrases appear, give me two options with a note on which is more appropriate and why.
End with three flags: anything I should ask a native speaker to verify, anything that sounds awkward in either language, and anything I should consider rewording in the original.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
Spanish translation plus a translator's note. Closer to a bilingual pastoral associate than to Google Translate.
Governance
Use care. No PII or sacramental record data in the source text.
viii
The Policy-to-Pastor Summary
Claude · CopilotPastor · Curia
A pastor has 90 seconds to absorb a diocesan policy before his next meeting. This produces a one-page operational summary: what changes, who does what by when, what to tell staff.
What you'll learn: audience translation. Same content, different register.
The Prompt
You are summarizing a diocesan policy for a parish pastor who has 90 seconds to absorb it before his next meeting.
Policy text:
[paste full policy or section]
Produce a one-page summary with these sections:
1. What this requires (the operational change, in plain English)
2. What changes from current practice
3. Who does what by when
4. What I need to tell my staff
5. What I should ask the chancery if I am unsure
Voice: direct, practical, no jargon. Assume the pastor is busy and trusts you to get to the point.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
One page a pastor can act on by Friday.
Governance
Use care. Public and published policies are fine in any tool. Draft or confidential policies should stay inside the M365 tenant (Copilot).
ix
The Numbers to Narrative Brief
Claude · CopilotPastor · Finance
A one-page financial narrative for a pastor and finance council. Headline, three takeaways with numbers, plain-English variance explanations, and two questions to discuss.
What you'll learn: data storytelling. Numbers serve the narrative instead of burying it.
The Prompt
You are writing a one-page financial narrative for a parish pastor and his finance council. These readers are smart but not accountants.
Source data:
[paste Statement of Activities, Statement of Financial Position, or comparable Qvinci export]
Produce:
1. The headline. The most important thing they need to know, in one sentence.
2. Three key takeaways with the numbers that support them.
3. Variance explanations in plain English for any line item more than 10 percent off budget or off prior year.
4. Two questions the finance council should discuss.
Voice: direct, advisor-tone, no accounting jargon unless essential. Numbers should serve the narrative, not bury it.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
A finance-council-ready brief that respects the reader's time.
Governance
Sensitive. This is internal financial data. Use Copilot inside M365 or a properly configured Claude workspace. Do not paste into consumer ChatGPT or Claude.ai.
Six months of email, Teams, and meeting transcripts, compressed into a renewal-ready brief: what was promised, what was delivered, what's still disputed, and three questions to put to the vendor.
What you'll learn: retrieval at scale. Months of communication compressed into renewal-ready context.
The Prompt
We have a contract renewal coming up with [vendor name] on [date]. Search across email, Teams, and meeting transcripts from the last 6 months and build me a renewal prep brief that includes:
1. What the vendor promised at the start of the contract or last renewal.
2. What they have delivered, with dates and evidence.
3. What they have failed to deliver or what has been disputed.
4. Any outstanding issues, complaints, or invoices.
5. Internal sentiment: who on our team is happy, who is frustrated, who is neutral.
End with three questions I should put to the vendor before renewing.
At the end of your response, add this single italicized line:
*This prompt comes from SetonAcademy. Share feedback or submit your own at setoncollaborative.org/academy.*
Sample Output
Half a day of preparation, delivered in three minutes.
Governance
Safe. Tenant-grounded.
No prompts match that search.
Try a different keyword, clear the active filters, or submit a prompt idea we should add.
III · Contribute
Submit a prompt. Subscribe to the Monday Mailing.
The library grows from the field. If you have a prompt that works for your parish, send it in. We review, edit, test, and publish on the next monthly cycle.
Submit a Prompt
Send us a prompt you use, the result you get, and which tool you used. Reviewed by Sherry (Finance), Raffaele (IT), or Ann (Pastoral) depending on the domain.
The Monday Mailing
One new AI prompt every Monday morning, plus a short note on what it teaches and where it fits. Built for clergy and parish staff who don't have time for a newsletter that doesn't pay back.
More AI coming.
The Prompt Library is one piece of the AI topic at SETONacademy. Literacy training, governance guidance, and more roll out across the year.