Week-14-of-52: RTF: the workhorse prompt structure that makes your first prompt land closer to the target.
AI Tuesdays · Week 14 · RTF: Role, Task, Format
Multifamily NEXT · AI Tuesdays
Week 14 · RTF
Self-guided · ~25 min · Phase 2
Warmup
The Workshop
Power Preview
Resources & Tools
Take Action
Week 14 of 52 · Self-Guided · ~20 min
0 starred0 notes
Warmup · Week 13 Recap
Welcome back. Let us warm up first.
Last week (Week 13) was about iteration: why your first prompt is a draft, and how a deliberate second prompt is almost always better than the first. This week is the flip side. Instead of getting the first prompt closer by reading the answer and refining, what if your first prompt were already good? That is what RTF does. Role, Task, Format. Three parts, in one line, that make the first prompt land much closer to the target.
New here?
Coming to AI Tuesdays for the first time? Start with Week 13 first. This workshop assumes you already know that your first prompt is a draft you refine. Week 13 teaches iteration in about 25 minutes. Come back here after.
Question 1 of 3
Week 13 recap: the four-step loop
Fill in the blanks. The Week 13 loop is: Ask, ____, ____, Stop.
Read, Refine.
Ask a good-enough first prompt. Read the answer to see what is missing. Refine with one deliberate follow-up in the same chat. Stop when it is good enough to finish by hand. That was Week 13.
Question 2 of 3
Week 13 recap: the three refinement moves
Which set of three is correct?
ADD, CORRECT, SHOW.
Add what was missing. Correct what was wrong. Show it an example of right. Hold onto these because RTF is a way to bake ADD and part of CORRECT into the FIRST prompt, so you need fewer rounds.
Question 3 of 3
Bridge into Week 14
You could refine a prompt for ten rounds. Or you could write a better first prompt. What is the case for a better first prompt?
A structured first prompt gets you closer sooner.
Iteration is not going away. But if you give the first prompt a real structure, it lands closer to the target, so you need one or two rounds instead of five. That structure is RTF: Role, Task, Format. This week.
Multifamily NEXT · AI Tuesdays · Week 14
Role. Task. Format.
RTF: the workhorse prompt structure that makes your first prompt land closer to the target.
Most AI prompts fail for one reason. You told AI what to do, but you did not tell it who is doing it, what shape the answer should take, and who is on the other end. RTF fixes that in one line. Give it a Role (who it is being right now). Give it a Task (what it is doing, specifically). Give it a Format (what the answer should look like). That is the whole framework. Three parts. One line. And the first prompt gets you 80 percent of the way there instead of 30 percent.
When you close this workshop, you will write prompts that give you what you actually needed, more of the time, the first time you ask.
Last Tuesday · Week 13
Iteration: Your Second Prompt Is Almost Always Better
You learned to refine after the fact. This week: how to make the first prompt land closer so you need fewer refinements.
Localize · Which of these is you?
Which role are you reading this in? Your answer sets up the examples you will see next.
Beginner-friendly · Read at your own pace · Star slides, add notes, print, or email yourself anytime
The Workshop · Agenda
Six parts. About 25 minutes. No timer.
Pause whenever you need. Star anything you want to come back to. Add your own notes as you go. This is designed to teach you RTF as a habit, not race you through it.
Part 1 · ~4 min
R · Give AI a Role
Telling AI who it is being changes what it knows to draw on, and how it talks. One line up front.
Part 2 · ~4 min
T · Name the Task
What the AI is actually doing. Not the topic. The verb plus the object plus who it is for.
Part 3 · ~4 min
F · Specify the Format
What shape you want the answer in. Length, structure, tone. The step almost everyone skips.
Part 4 · ~3 min
Assembled RTF
All three parts, in one line. Short, ordered, repeatable. The habit worth building.
Part 5 · ~5 min
Power Preview: bare prompt vs RTF prompt
Same task, twice. First a plain request. Then the RTF version. The gap between them is what RTF buys you.
Part 6 · ~5 min
Try This Today · Reference · Session Complete
One RTF drill for this afternoon. Updated cheat sheet. Board paragraph. Where to practice safely.
No timer. Pause whenever you need. The estimate above counts reading plus every tap and reflection.
The Workshop · Why You Should Care
The cost of skipping structure.
Two things happen when your first prompt has no structure. Both are quiet. Both are expensive. You have probably done both this week.
Scenario one
The generic wall
A property manager asks AI to "write a rent increase notice." The answer is a bland corporate paragraph that could be from any landlord in America. She spends fifteen minutes rewriting it in her own voice before she can send it.
The prompt did not tell AI who it was being (an experienced property manager), what exactly it was doing (a 60-day rent increase notice for a good long-term resident), or what shape the answer should take (warm, under 150 words, one clear next step). All three things she had in her head. None of them went into the prompt.
Scenario two
The wrong shape
A marketing coordinator asks AI for "ideas for a summer social media campaign." He gets a wall of prose, three paragraphs describing themes, when what he actually wanted was a table with five posts, each with a caption and a hashtag.
The answer was not wrong. It was the wrong shape. He never told AI what the output should look like. That is the F in RTF, and almost nobody thinks to include it.
Both are structure problems. RTF fixes both in one line, before you hit send.
The Workshop · Part 1 of 6
R · Give AI a Role.
The first line of an RTF prompt names who AI is being right now. That single line changes what words AI reaches for and who it thinks is on the other end. "You are a leasing professional" and "You are a copywriter" produce different answers to the same next sentence, because they draw on different corners of what AI was trained on.
Without a role
"Write a note about the pool being closed for maintenance."
AI defaults to the average voice of every pool-closure notice ever written. Corporate. Bland. Reads like a legal disclaimer.
With a role
"You are a friendly on-site property manager. Write a note about the pool being closed for maintenance."
Warm. Neighborly. Talks to residents like people who live there. Anticipates the questions a real manager would.
How to write the role
Start with "You are a" then name the role plus one qualifier.
"You are a leasing professional at a mid-priced garden-style property" beats "You are a leasing pro" because the qualifier tells AI which corner of the training data to lean on. It reads a Class B garden community differently than a Class A luxury tower.
"You are a maintenance supervisor with fifteen years of experience" beats "You are a maintenance person" because seniority changes tone: more direct, more decisive, less hedging.
One line. Six or seven words. Every time.
Try one now
Write the role line for a prompt you are about to send. Not the whole prompt, just the "You are a ___." Read it back. Does it name a real corner of expertise, or is it generic?
The Workshop · Part 2 of 6
T · Name the Task.
The T is the verb. What is AI actually doing? Writing? Rewriting? Summarizing? Comparing? Ranking? Explaining to a beginner? Most bad prompts skip this or blur it. "Help me with the newsletter" is not a task. "Rewrite this newsletter opener to be 40 percent shorter and end with a question" is a task.
Three parts of a good task line
1. The verb. Write. Rewrite. Summarize. List. Compare. Draft. Explain. Convert. Not "help with" or "do something about." A verb tells AI what shape of output to reach for.
2. The object. What is the verb acting on? A renewal notice for a 24-month resident. A summary of the last three inspection reports. Three subject-line options for a re-engagement email.
3. Who it is for. A resident who has been here two years. A regional manager reviewing 40 properties. A first-time apartment hunter comparing three listings. The audience changes the answer.
Weak task
"Help me with a rent increase."
No verb. No object. No audience. AI has to guess whether you want the letter, the math, the policy, or the conversation.
Strong task
"Write a 60-day rent increase notice for a two-year resident whose rent is going from $1,450 to $1,495."
Verb: write. Object: 60-day notice with the delta named. Audience: two-year resident. AI now knows what shape and tone to reach for.
The T is where you stop being polite and get specific. AI does not need you to be nice. It needs you to be clear.
Try one now
Write the task line for the same prompt you started above. Just T. Verb, object, audience, in one sentence.
The Workshop · Part 3 of 6
F · Specify the Format.
The F is the step almost everyone skips. What shape should the answer be? A five-line email? A three-column table? A bulleted list of ten items with one supporting sentence each? A short paragraph followed by three options?
If you do not tell AI the shape, it picks one for you, and the shape it picks is usually a wall of prose. That is not because AI is lazy. That is because "prose" is the average of everything it was trained on. Named a format = you got what you wanted. Skipped format = you got what the internet averages to.
What to name in the format
Length. "Under 120 words." "Three sentences." "One paragraph plus three bullets."
Structure. "As a table with three columns." "As an email with subject line, greeting, body, close." "As five bullets."
Tone. "Warm and neighborly." "Direct and unemotional." "Professional but not corporate."
Constraints. "Do not invent any dates or amounts." "Do not mention the pool." "Include one clear next step."
No format
You asked for "summer campaign ideas"
You got four paragraphs of prose describing themes. You wanted a table with five posts, one caption each, one hashtag each. You never told it that.
With format
You asked for the same, plus "as a five-row table: post number, caption, hashtag."
You got the table. Copy-paste into your content calendar. Done in two minutes.
Try one now
Write the format line for your prompt. Length, structure, tone, and one constraint. In one line.
The Workshop · Part 4 of 6
The full RTF prompt.
Put the three lines together in order. Role, then Task, then Format. That is the whole structure. It works because each part clears the way for the next: the role sets the voice, the task sets the direction, the format sets the shape.
Sample RTF prompt
Role: You are a friendly on-site property manager at a mid-priced garden-style community.
Task: Write a 60-day rent increase notice for a two-year resident whose rent is going from $1,450 to $1,495.
Format: Warm, under 150 words, one clear next step at the end. Do not invent any dates or amounts. Use brackets for anything the manager needs to fill in.
The pattern to memorize
"You are a [role plus qualifier]. [Task in one sentence with verb, object, audience]. [Format: length, structure, tone, one constraint]."
Three sentences. Twenty seconds to write. The gap between this and "write me a rent increase notice" is enormous.
Try one now
Put your three lines from the last three slides together. Read it out loud. If it does not sound like something you would actually send to AI, tighten it.
The Workshop · Part 5 of 6
RTF meets iteration.
This is where Week 13 and Week 14 come together. RTF makes your first prompt land closer to the target. Iteration handles what is left. They are not competing methods. They are two halves of the same habit.
The two work together
Without RTF: your first prompt lands 30 percent of the way there. You need four or five refinements to reach a usable draft.
With RTF: your first prompt lands 80 percent of the way there. You need one refinement, maybe two, to reach the same draft.
RTF does not replace Week 13. It shortens the loop.
When RTF is not enough
RTF gets you a shape. It does not give AI facts it does not have. If your prompt needs specific numbers (a resident's balance, a property's vacancy rate, this year's renewal cadence), you still have to supply them. RTF is structure, not omniscience.
When the answer comes back and the shape is right but the facts are missing, that is not an RTF failure. That is the moment to iterate using an ADD move from Week 13: give it the fact, keep the RTF structure, send again.
The habit worth building
1. Write your first prompt as RTF. Role, task, format.
2. Read what came back. If the shape is right but a fact is off, do a one-line ADD or CORRECT.
3. Two rounds should be enough for most work tasks. If you are on round five, RTF was probably too vague or the task was too big.
If AI keeps missing the same thing after two RTF refinements, start fresh with a tighter RTF instead of adding a sixth follow-up.
The Workshop · Where AI Goes Wrong
The bright line, again.
A better prompt structure never changes what data is allowed in. This was the Week 13 bright line. It is the Week 14 bright line too. It will be the bright line every week of Phase 3 while we teach prompt structures.
The bright line
Structuring a prompt improves the words. It never changes what data is allowed in.
A well-written RTF prompt is easier to read, easier to send, and easier to reuse. It is not a permission slip. If a resident's real name and balance should not go into a public AI tool as a plain prompt, they should not go in as a nicely-formatted RTF prompt either.
Right way
RTF with brackets
Write the RTF prompt. Where sensitive data would go, use [Resident name], [balance], [move-in date]. AI drafts the template. You fill the sensitive parts in yourself, by hand. Same speed. Zero data exposure.
Wrong way
Sensitive data pasted in
Same RTF prompt, but you paste the resident's real name, real balance, real address so AI can "personalize" it. Structure did not change the rule. The rule is: real identifiers plus real financials stays out of a general AI tool.
Full data classification treatment: Week 1. Practice on Slide 19 of this workshop.
The Workshop · Default Rule
The default: RTF every prompt worth sending.
Not every prompt needs RTF. "What is the capital of Nebraska" needs zero structure. But every prompt worth thinking about for more than twenty seconds is worth thirty more seconds to structure. That is the trade RTF asks you to make, and it pays off almost every time.
The Week 14 default rule
1. If it is a real work task, use RTF as your first prompt. Role, task, format. In order.
2. Read the answer. If it is not right, refine once using the Week 13 moves (ADD, CORRECT, SHOW).
3. Stop when the draft is good enough to finish by hand. Do not chase perfect.
4. If the data is sensitive, write RTF with brackets. Fill in the private parts yourself.
Trust me
Try RTF on three real tasks this week. Pay attention to how many refinements each one needs. You will feel the difference in the first one. Two weeks from now, RTF will be automatic and you will forget you used to skip it.
Guess Green, get seen. Choose Yellow, stay mellow. Red? Use your head.
Structuring your prompts does not change the tier of your data. If you are unsure, treat it as Yellow.
The Workshop · Q & A
Questions people ask about RTF.
Tap any question to expand the answer. These are the honest questions that come up when RTF stops being a slide and starts being a habit.
RTF has three parts. That is why we teach it first in Phase 3. Any prompt framework with more than four parts becomes a memorization drill instead of a habit. RTF fits in your head after two prompts. If a framework needs a card taped to your monitor, it is not going to survive a real Tuesday.
Both work. Labels help you when you are learning, because they force you to think about each part separately. Once RTF is a habit, most people drop the labels and write it as one flowing sentence: "You are a friendly property manager. Write a 60-day rent increase notice for a two-year resident going from $1,450 to $1,495. Keep it warm, under 150 words, with one clear next step." AI does not care about the labels. It cares that the three things are there.
Write what you would say if a new hire asked you to describe your job in one sentence. "You are a regional manager overseeing 12 apartment properties in three states. You care about NOI and resident retention equally." The role is not a job title. It is a stance. Name the stance you want AI to take, and it will.
Long is fine if every line pulls its weight. "Under 150 words, warm tone, one next step, do not invent numbers, use brackets for anything the manager fills in" is five constraints in one line and every one matters. What you want to avoid is padding: repeating the same idea in different words, adding constraints that contradict each other, or asking for "professional but casual but friendly but authoritative." Pick a lane.
Role-Task-Format is the order because it matches how a prompt reads: set the voice, then the direction, then the shape. Format-Task-Role technically works but is harder to write, because you do not know what shape you want until you know what the task is. Stick with R-T-F. Muscle memory is worth more than optimization.
No. Structure never changes what data is allowed in. This is the whole point of Slide 9. RTF makes a better prompt. It does not make a general AI tool a safe place to paste sensitive data. Use brackets for the sensitive parts, get the template, fill in the details yourself.
Power Preview · Setup
Watch RTF do the work.
The next two slides each show the same task twice. Bare prompt first. RTF prompt second. Same AI, same session, only the prompt changes. This is where the framework earns its keep.
Slide 13
Example one: the renewal email (same task from Week 13)
Last week you saw a bare prompt and then refined it round by round. This week you see the same task, this time started with RTF. Notice how much closer the FIRST answer is to what you actually needed. That is the RTF payoff before iteration ever kicks in.
Slide 14
Example two: a monthly owner's report
A harder task where format matters more than voice. The bare prompt gives you a wall of prose. The RTF prompt gives you a scannable one-pager an owner can read in twenty seconds. Sample only, not real portfolio data.
Reminder before you scroll
Neither example uses a real resident name, real balance, real property numbers, or a real owner. They use bracketed placeholders. The bright line from Slide 9 holds here too.
Power Preview · Example 1 · Renewal Email
Same task, twice.
A 60-day rent increase notice for a two-year resident, $1,450 to $1,495. First the bare prompt. Then the RTF prompt. Nothing sensitive was pasted.
Bare prompt · the whole thing you sent
Write a rent increase notice for a resident.
What AI sent back
Subject: Important Notice Regarding Your Lease Renewal
Dear Valued Resident,
We are writing to inform you of an upcoming adjustment to your monthly rental rate. Effective at the conclusion of your current lease term, your monthly rental obligation will be adjusted in accordance with current market conditions and operational considerations. This decision has been made after careful review of comparable properties in the surrounding area.
Please review the attached documentation and reach out to our office at your earliest convenience should you have any questions or concerns regarding this matter.
Sincerely, Property Management
Corporate. Cold. No amounts. No dates. Reads like a bank rejection letter. Would take fifteen minutes to rewrite.
RTF prompt · the whole thing you sent
Role: You are a friendly on-site property manager at a mid-priced garden-style community.
Task: Write a 60-day rent increase notice for a two-year resident whose rent is going from $1,450 to $1,495.
Format: Warm, under 150 words, one clear next step at the end. Use brackets for anything the manager needs to fill in. Do not invent dates.
What AI sent back
Subject: A quick heads-up about your renewal
Hi [Resident First Name],
It has been great having you at [Community Name] for the past two years, and I wanted to give you plenty of notice before your renewal date.
Starting [New Rent Effective Date], your monthly rent will go from $1,450 to $1,495. That is a $45 change and reflects what similar apartments in the area are renting for right now. Nothing else about your lease is changing.
If you would like to talk it through or start the renewal paperwork, just reply to this email or stop by the office. I am happy to walk through it with you.
Thanks for making [Community Name] home, [Manager Name]
Warm. Specific. The delta is named. One next step. Brackets protect sensitive fields. Ten seconds of edits and it ships.
The difference is not that AI got smarter. The difference is that you told it who it was, what to do, and what shape to use.
Power Preview · Example 2 · Monthly Owner's Report
When format matters more than voice.
A monthly owner's report for one property in a large third-party management company's portfolio. The kind of report a portfolio operations analyst prepares monthly for an owner-client. Sample only, all numbers bracketed.
Bare prompt · the whole thing you sent
Write a monthly owner's report for one of our properties.
What AI sent back
Monthly Property Report
The property continues to perform in line with expectations for the current reporting period. Occupancy has remained relatively stable and management continues to focus on resident retention initiatives and operational efficiency. Collections activity has proceeded as anticipated with ongoing follow-up on outstanding balances by the site team.
Financial performance is generally aligned with budgeted expectations, with certain line items showing minor variances. The maintenance team addressed several service requests during the month and continues to maintain the property in good condition. Leasing activity remains an area of focus, with the team working on prospective residents and ongoing renewal conversations.
We look forward to continued positive performance and welcome any questions the ownership may have regarding the property or its operations.
Vague. Zero numbers. Zero structure. An owner reading this learns nothing. Would be rejected on submission.
RTF prompt · the whole thing you sent
Role: You are a portfolio operations analyst at a large third-party management company that manages 1,000+ apartment communities for institutional owners.
Task: Draft a one-page monthly owner's report summary for one property, written for an institutional owner who reads dozens of these each month.
Format: Six labeled sections in this order: Occupancy, Delinquency, NOI vs Budget, Maintenance Highlights, Leasing Pipeline, Upcoming Lease Expirations. One lead metric per section, one sentence of context. Close with a two-sentence "What changed this month" paragraph. Use bracketed placeholders for all numbers. Under 200 words.
Upcoming Lease Expirations: [XX] leases expiring in the next 60 days. Renewal offers sent on [XX] of them; verbal commitments on [X].
What changed this month: [Two-sentence narrative from the site team on the single most important operational story of the month.]
Scannable in twenty seconds. Six sections, one metric each. Every bracket is a slot the site team fills in from the property management system. Zero real data pasted.
This is what the F does at scale. When an owner reads forty of these a month, structure IS the report.
Power Preview · Try This Today
One RTF drill you can run this afternoon.
Fifteen minutes. One task. Bare prompt first, then RTF. You will feel the difference before you finish reading the second answer.
The drill
1. Pick a real work task. Something you would actually send today. A resident email, a policy summary, a maintenance write-up. Nothing with sensitive data.
2. Send the bare version first. One sentence. What you would have sent last week. Read what comes back. Notice how much you would have to fix.
3. Send the RTF version second. Same task. Role, Task, Format, in order. Same AI, new chat.
4. Compare the two answers side by side. How much closer did the RTF version land? How many fewer edits would it need? Write it down.
Good job
Every leasing professional, property manager, and regional I have run this drill with has said the same thing: "I cannot believe how much closer the RTF one was." That is not because RTF is magic. That is because you have been asking the wrong questions.
Where to practice safely
If you are running this drill on real property data, do it inside mPro Digital Edge. The platform is built for multifamily, the guardrails are in place, and the AI Council rules travel with the tool. If you do not have access, use bracketed placeholders and a public AI tool. Same drill, safer inputs.
Log the drill
Once you run the two prompts, come back and note: What was the task? How different were the two answers? What surprised you?
Reference · Bookmark this slide
Vocabulary cheat sheet: 35 core terms.
Every AI term this curriculum has introduced so far, in one place. Plain-language definitions. This is the same term list on the PDF download at the bottom of this slide and on Slide 21.
Foundational (from Weeks 8-10)
Generative AI
AI that creates new content (text, images, audio, code) based on training data patterns.
Large Language Model (LLM)
The kind of AI behind ChatGPT, Claude, Gemini, and Copilot. Trained on vast text and predicts one word (token) at a time.
Token
A small chunk of text an LLM reads and writes. Roughly 0.75 words per token; 1,000 tokens is about 750 words.
Training Data
The vast collection of text an LLM learned patterns from. Not a searchable database, a compressed pattern archive.
Prediction
How AI decides what to say next: the most likely word based on the pattern of what came before, one word at a time.
Hallucination
AI producing a confident, fluent answer that is factually made up. Especially common when it does not know something you did not tell it.
Sycophancy
AI agreeing with what you just said instead of pushing back. Feels helpful, is often wrong.
Prompt anatomy (from Week 11)
Prompt
What you type to AI. Your input. The one part of the container you fully control.
System Prompt
A hidden instruction the AI tool sends the model before your prompt. Shapes tone, role, and constraints. You do not write it and cannot see it.
Conversation History
Every earlier message in the current chat, yours and AI's. AI re-reads all of it every turn. This is how AI "remembers."
Context Window
The ceiling on how much text AI can read in one conversation. When it fills up, the oldest content silently drops off.
Prompt Anatomy
The five parts of a well-formed prompt: instruction, context, input data, output format, examples.
Output Format
The shape you want the answer in: length, structure, tone. One line in your prompt saves you rewriting the whole reply.
Prompt quality (from Week 12)
Specificity
The single biggest lever on prompt quality. Naming the audience, the format, and the constraints so AI is not guessing.
Ambiguity
What makes AI fall back on average defaults. The parts of a prompt AI has to guess. Iteration cannot rescue what specificity should have prevented.
Iteration (from Week 13)
Iteration
Refining a prompt after reading the answer, in the same chat. Your first prompt is a draft. Your second, written on purpose, is almost always better.
Refinement
The act of sending a targeted follow-up: add what was missing, correct what was wrong, or show an example of right. One deliberate change beats a full rewrite.
The Three Moves
ADD (give it the context it did not have), CORRECT (tell it what to stop doing), SHOW (give it an example of right). Every refinement is one of these three.
Diminishing Returns
When each new round of refinement barely changes the answer. Your signal to stop prompting and finish the piece by hand.
Loop
The Ask → Read → Refine → Stop cycle. The four-step rhythm that turns first-answer disappointment into a usable draft.
This week · RTF (Week 14) *NEW
Prompt Structure
A repeatable pattern for writing prompts. RTF is a prompt structure. Seven of them are taught in Phase 3 (Weeks 14 through 20), each for a different kind of job.
RTF
Role, Task, Format. The workhorse prompt structure. Three parts in one line that make the first prompt land closer to what you actually needed.
Role (in RTF)
The first line of an RTF prompt. Tells AI who it is being right now. "You are a friendly on-site property manager." Changes what AI draws on and how it talks.
Task (in RTF)
The second line. What AI is actually doing. A verb, an object, and enough detail that two people could not interpret it differently. Not a topic.
Format (in RTF)
The third line. What shape the answer should take. Length, structure, tone. The part almost everyone skips, and the reason so many AI answers feel "close but not quite."
Governance and data (running through the curriculum)
Data Classification
Sorting a piece of information by which AI tool it is allowed to enter. Tier 1 (safe for any AI), Tier 2 (enterprise only), Tier 3 (never in AI). Full treatment: Week 1.
Tier 1 · Safe
Public or fully de-identified information. Any approved AI tool is fine.
Tier 2 · Enterprise Only
Sensitive but useful information. Only in an approved enterprise AI workspace with data protections in place.
Tier 3 · Never
Information that never enters an AI tool. Real names tied to sensitive details, financial identifiers, protected-class data, and screening or eviction information.
Use Case
A specific way you plan to use AI: who is doing it, what task, what data goes in, what comes out. The Council approves use cases, not tools.
Guardrail
A rule inside your prompt or your workspace that limits what AI can do. "Do not invent numbers" is a guardrail. Full treatment: Week 17.
Human-in-the-Loop
A person reviews AI output before it goes to a resident or a decision. The RETTC principle behind almost every safe use case.
AI Council
The group in your organization that approves use cases, reviews new tools, and owns the response when something goes wrong. Full treatment: Week 7.
NIST AI RMF
National Institute of Standards and Technology AI Risk Management Framework. The four functions: GOVERN, MAP, MEASURE, MANAGE.
RETTC AI Governance Framework
Real Estate Technology and Transformation Center. Eight principles specific to rental housing. Detailed on Slide 18.
Read down the left column when you are writing a prompt. Read down the right when you want to see it in action.
The Template
R · ROLE
You are a [role] at [qualifier]. One role. Include a qualifier. Do not stack.
T · TASK
[Verb] [object] for [audience], [one or two narrowing details]. A verb, an object, enough detail to remove ambiguity.
F · FORMAT
[Length]. [Structure]. [Tone]. Length. Shape. Voice. The line almost everyone skips.
Filled In
Role: You are a friendly on-site property manager at a 200-apartment garden-style community.
Task: Write a 60-day renewal email to a resident who has renewed twice, offering a 12-month renewal at a 3 percent increase.
Format: Under 150 words. Email with subject line and greeting. Warm but professional. Use [Resident name].
Watch for these traps
Stacked roles. One role per prompt. Topic instead of task. A verb is not optional. Missing format. AI defaults to 300 words of corporate prose. Real resident data. Better structure never buys permission. Use brackets.
NIST AI Risk Management FrameworkNIST AI RMF (NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY, AI RISK MANAGEMENT FRAMEWORK)
The US federal framework for managing risk in AI systems. Four functions: GOVERN, MAP, MEASURE, MANAGE. Voluntary but widely adopted.
Full document at nist.gov/itl/ai-risk-management-framework & RETTC 8 PrinciplesRETTC AI GOVERNANCE FRAMEWORK (REAL ESTATE TECHNOLOGY AND TRANSFORMATION CENTER)
Eight principles for AI governance in rental housing. Anchored to Fair Housing, resident privacy, and human accountability. Published at rettc.org/frameworks/ai-governance.
Two frameworks. The NIST AI Risk Management Framework (AI RMF) from the National Institute of Standards and Technology, and the AI Governance Framework from RETTC (Real Estate Technology and Transformation Center, at rettc.org). US-anchored. Together they cover the risk side (NIST) and the operating side (RETTC) of governing AI in rental housing. Every Week 14 concept about RTF lands somewhere in one or both.
Week 14 · RTF anchors
Week 14 concept
NIST AI RMF anchor
RETTC AI Governance anchor
Role: telling AI who it is being
MAP function: understanding the context in which AI is being used, before the risk emerges.
Principle 2 · Transparency & Explainability: users should understand what role AI is playing.
Task: naming what AI is producing
GOVERN function: use cases must be defined before AI is put to work on them.
Principle 8 · Purpose Limitation: AI is used only for the tasks it was approved for.
Format: specifying the output shape
MEASURE function: quality of AI output depends on being able to evaluate whether it met the spec.
Principle 4 · Accountability: humans review AI output against a defined standard before acting on it.
RTF with brackets (sensitive data)
MANAGE function: risks are actively mitigated, not just identified.
Principle 6 · Privacy Protection: resident data stays out of general AI tools regardless of how well the prompt is written.
The bright line (data allowed in does not change with structure)
GOVERN function: policy sets what data is permitted in AI, not the sophistication of the prompt.
Principle 3 · Fair Housing Compliance: no prompt structure changes what protected-class or resident data can enter AI.
The RETTC AI Governance Framework · all 8 principles
Official titles as published at rettc.org/frameworks/ai-governance. These are the 8 principles the Framework asks every rental housing provider and technology partner to adopt.
#
Principle
1
Establish an Organizational Philosophy on AI Use and Oversight. Adopt a clear philosophy that defines how AI will be used, governed, and monitored across operations.
2
Promote Fairness & Consumer Protection. Ensure AI systems comply with the Fair Housing Act, Fair Credit Reporting Act, Americans with Disabilities Act, and state and local law.
3
Uphold Transparency & Explainability. Promote a general understanding of when AI influences key decisions, and provide clear explanations to renters and applicants of decisions affecting them as required by law.
4
Protect Privacy & Promote Data Integrity. Adhere to federal, state, and local consumer privacy laws, with an emphasis on data minimization, security, and informed consent.
5
Maintain Outcome-Based Accountability & Human Involvement. Engage humans in significant decisions, establish processes to review model outcomes, provide escalation channels for complaints and appeals, and educate staff on AI use and limitations.
6
Promote Positive Renter Experiences and Outcomes. Elevate technologies that enhance affordability, access, and housing stability, and use AI to improve resident experiences.
7
Strive for Innovation with Responsibility. Encourage development of AI tools that benefit residents, prospects, guests, and operators, without compromising trust.
8
Encourage Third-Party Partner Accountability and Due Diligence. Incorporate robust due diligence and risk management controls for third-party AI tools. Invite transparency from vendors regarding data sources, model limitations, and known risks.
1. RTF is not the only prompt structure. It is the workhorse. Other jobs need other structures. Chain of Thought (Week 15) is for reasoning problems. Few-Shot (Week 16) is for pattern-matching. RTF is the default, not the universal solution.
2. RTF cannot rescue a task you do not understand. If you cannot describe the task clearly to a colleague, RTF will not save you. Think first.
3. Role prompting has diminishing returns for very simple tasks. Asking AI "you are a math teacher" to solve 2+2 does not help. Save RTF for the tasks worth structuring.
4. We are confident everywhere. Same note as every week: workshop writing is confident by design. If any specific claim struck you as too confident, that instinct is right. Verify it.
Reference · Spot the Tier
RTF is a structure. Classification is a rule.
Eight RTF prompts a real person might write this week. Your job: read each one, then tap the tier that matches what the person actually put into the prompt. Structure never changes the tier. Only the data does.
Tier 1 · Green: general information, templates, no PII. Tier 2 · Yellow: real business context but no resident identifiers or financials. Tier 3 · Red: real resident names, real balances, real medical, real applicant status, real portfolio financials.
Scenario 1
"You are an on-site property manager. Write a 60-day rent increase notice for a two-year resident whose rent is going from $1,450 to $1,495. Use brackets for the resident's name and address."
Green. The RTF prompt uses bracketed placeholders for identifying information and a generic delta ($1,450 to $1,495) that could be any resident. Nothing here identifies a real person or account.
Not quite. The brackets keep the identifying data OUT of the prompt. Amounts alone are not identifiers. Green.
Scenario 2
"You are a leasing professional. Draft a follow-up to a prospect who toured yesterday. Prospect name: Maria Alvarez. Cell: 555-0198. She loved unit 4C but was worried about the pet fee."
Red. Real name, real cell phone, real prospect status, real unit number. Excellent RTF structure. Wrong data classification. Bracket the name and phone, keep the RTF, get a template, fill in the details yourself.
Look again. There is a real name plus a real phone number plus prospect status plus a unit. RTF cannot save this. Red.
Scenario 3
"You are a maintenance supervisor. Write a standard operating procedure for how a technician should respond to a leaking dishwasher call. Format: five numbered steps, each under 20 words, plus a one-line safety note."
Green. Generic SOP. No property named, no resident, no financials. Standard-of-care documentation is safe to draft with a public AI tool.
Not quite. This is a generic SOP with no real people or numbers. Green.
Scenario 4
"You are a portfolio operations analyst. Summarize this month's NOI variance for [Property Name]. Actual: $[XXX,XXX], Budget: $[XXX,XXX]. Format: three bullets, the largest driver first, use bracketed placeholders throughout."
Green. Fully bracketed. No property name, no real actuals, no real budget. AI drafts the shape, the analyst pastes real numbers in the final version by hand. Textbook RTF-plus-brackets.
Look again. Every sensitive slot is bracketed. Nothing real was put in. Green.
Scenario 5
"You are a property manager. Write a reply to a resident who is asking about the noise from the construction. Resident is Sam Chen, unit 312, has lived here 4 years, has never complained before. Include an offer for a rent credit if noise continues past next Friday."
Red. Real name, real unit, real tenure, and a proposed financial concession. Any one of those makes it Yellow. Combined, plus the rent credit, this is a resident record. Bracket everything. Then re-send.
Real resident details plus a financial offer. The RTF is fine. The data is not. Red.
Scenario 6
"You are a leasing professional. Explain the difference between a co-signer and a guarantor to a first-time renter. Format: two short paragraphs, plain English, no legal jargon."
Green. Pure educational content. No specific resident, no specific application. Note: never treat AI output as legal advice, but drafting a plain-English explanation is safe.
Not quite. Generic education. No real applicant or property. Green.
Scenario 7
"You are a regional manager. Compare the delinquency trends across our three Florida properties: [Property A], [Property B], [Property C]. Include specific unit numbers where residents are 60+ days late."
Red. Property names are bracketed but the request itself asks AI to receive specific unit numbers of residents 60+ days late. That is a list of at-risk households by address. Yellow if you strip the unit numbers. Red as written.
Look at what the prompt is asking to include: specific unit numbers of delinquent residents. That is a list of at-risk households. Red.
Scenario 8
"You are a marketing coordinator. Draft five social media captions for our summer amenity refresh at [Community Name], focused on the renovated pool deck. Format: table with caption, hashtag, and best time to post."
Green. Marketing content, generic amenity, bracketed community name. Zero residents mentioned. Zero financials. RTF plus brackets in action.
Not quite. Public marketing content with a bracketed community name. Green.
Take Action · Anonymous Quiz
Five questions. Anonymous. Just for the pattern.
Nothing here identifies you. Your answers help us see where Week 14 landed and where to reinforce next Tuesday. Tap one answer per question, then submit at the bottom.
Question 1 of 5
What do the three letters in RTF stand for?
Question 2 of 5
Which is the strongest ROLE line?
Question 3 of 5
Which is the step that almost everyone skips when writing a prompt?
Question 4 of 5
A well-structured RTF prompt with resident names, balances, and unit numbers pasted in is:
Question 5 of 5
Your RTF first prompt comes back with the right shape but a wrong fact. What do you do?
Take Action · Policy & Download
Take one thing back to your team.
The RTF Template Quick Reference is a one-page PDF built for your monitor or your break room. Print it. Send it to your supervisor. Talk about RTF at your next team huddle. That is how a workshop becomes a habit.
This week's downloads
RTF Template Quick Reference. One page. R-T-F on the left, worked example on the right. Bracket-first data-safety note at the bottom.
AI Vocabulary Cheat Sheet (35 terms). Every term this curriculum has introduced, in one place. Same list as Slide 16.
Give your supervisor a preview of what you learned this week. One paragraph, one link. Takes ten seconds. Opens the door to a real conversation at your next one-on-one.
Opens your email client with the message pre-filled. You add your supervisor's email address before sending. Nothing is sent for you.
Take Action · Session Complete
Good job. Session Complete.
What you walked away with
1. The three parts of RTF and why they work in that order.
2. A pattern to memorize: "You are a [role plus qualifier]. [Verb] [object] for [audience]. [Length. Structure. Tone. One constraint]."
3. Two worked examples showing what the F does at scale, from a renewal email to a monthly owner's report.
4. The bright line, again: structure never changes what data is allowed in. Brackets stay in the toolkit.
5. A drill you can run this afternoon on any real task, plus a one-page reference for your monitor.
The one-paragraph board version
This week we introduced RTF, the workhorse prompt structure. Three parts, one line: a role that tells AI who it is being, a task that names the verb and object and audience, and a format that specifies length, shape, and tone. The workshop paired RTF with the previous week's iteration habit to show how a structured first prompt gets us 80 percent of the way to a usable answer instead of the 30 percent a bare prompt returns. The bright line held: structure improves the words but never changes what data is allowed in. Bracketed placeholders stay in every prompt with sensitive slots.
Where to practice safely
If your team is ready to move from "let us try RTF in the free tool" to "let us do this on real property data with the guardrails in place," that is what mPro Digital Edge is built for. Multifamily-specific AI, AI Council rules travel with the tool, sensitive data stays inside your organization.
Next Tuesday · Week 15
Chain of Thought: how to make AI show its work.
This week you told AI the shape you wanted the answer in. Next week you tell it the shape you want the reasoning in. Chain of Thought is what you use when the task is complex, the stakes are higher, and you need to see how AI got to its answer before you trust it. RTF plus Chain of Thought is a serious combination.
