AI Tuesdays · Week 12 · The Right Way to Talk to AI
Multifamily NEXT · AI TUESDAYS
Week 12 · The Right Way to Talk to AI
Self-guided · ~25 min · Phase 2
Warmup
The Workshop
Power Preview
Resources & Tools
Take Action
Week 12 of 52 · Self-Guided · ~20 min
0 starred0 notes
Warmup · Week 11 Recap
Welcome back. Warm up first.
Last week you learned the parts of a prompt: what AI actually receives when you hit send, and the pieces a well-formed prompt is built from. This week goes one layer in. You can have every part in place and still get weak output, because the words inside those parts are vague. Specificity is the fix. Three quick questions first.
New here?
Coming to AI Tuesdays for the first time? Read Week 11 first. This week assumes you already know the parts of a prompt. Week 11 teaches that in about 25 minutes. Come back here after.
Question 1 of 3
Week 11 recap: what AI reads
When you hit send, what does AI actually read?
All three.
Your prompt, the hidden system prompt the tool adds, and the conversation history. If a detail is not in one of those three places, AI does not have it. Week 11 covered this. Hold onto it, because specificity is about what you pack into that first place: your prompt.
Question 2 of 3
Week 11 recap: why a prompt fails
You asked for a summary and got a wall of text instead of the table you pictured. Which part of the prompt was missing?
The output format.
AI cannot guess the shape in your head. If you do not name it (a table, five bullets, two sentences), AI picks one for you. Naming the format is one kind of specificity. This week names the rest.
Question 3 of 3
Bridge into Week 12
Every part of your prompt is present. The instruction reads: "write something professional." What is still wrong?
"Professional" is vague.
AI knows the word. It just cannot tell which professional you mean: warm and personal, or formal and legal. So it guesses. Vague in, vague out. This week you learn to say it so AI cannot guess wrong.
Multifamily NEXT · AI Tuesdays · Week 12
The Right Way to Talk to AI
Vague in, vague out. Specific in, specific out.
Week 11 gave you the parts of a prompt. This week is about the words you put inside those parts. Here is the whole idea in one line: AI matches the specificity of your request. Ask for something vague and AI hands you something vague, fluently and confidently. Ask for something specific and measurable and AI hands you that. The skill that separates a beginner from a builder is not a secret phrase. It is the habit of saying exactly what you want, in terms AI can actually check.
When you close this workshop, you will stop asking AI to be "professional" or "engaging" and start telling it exactly what would make you say "yes, that is it."
Recorded this week
Tami is on the Multifamily NEXT tour in Chicago, so this session is pre-recorded. Same workshop, same set-your-own-pace format. Read it whenever your Tuesday allows.
Last Tuesday · Week 11
The Anatomy of a Prompt
You learned what AI receives and the parts a prompt is built from. This week: the words that go inside those parts, and how to make them specific.
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 keep. Add your own notes as you go. This is built to give you a habit you can use this afternoon, not to race you through it.
Part 1 · ~3 min
What "specific" actually means to AI
Three plain tests: can AI measure it, can AI verify it, can AI decide it without guessing. A word that fails all three is vague.
Part 2 · ~4 min
The five dimensions of specificity
Audience, scope, tone, output format, constraints. Five places you can tighten any prompt. Most weak prompts are loose in three of them.
Part 3 · ~3 min
The eight vague words
Professional. Engaging. Modern. Good. And four more. The words that feel like instructions but tell AI nothing it can act on.
Part 4 · ~3 min
The swap
Every vague word, replaced with something AI can check. A working table you can keep next to your desk.
Part 5 · ~7 min
Power Preview: same task, two prompts
One vague prompt. One specific prompt. Same request. Read what AI produces from each. The gap is the whole lesson.
Part 6 · ~5 min
Try This Today · Reference · Session Complete
One five-minute specificity audit you can run on a real prompt today. Updated vocabulary sheet. A one-line version for your board. And where to practice this safely.
Time estimate is honest: reading time plus interaction time plus a little reflection. Move at your pace.
Why You Should Care
Two requests you have probably made.
Two composite situations, built from the pattern we see constantly. In both, a capable professional gave AI a reasonable-sounding request and got back something useless. In both, the AI tool got blamed for a problem that was sitting in the request.
Scenario one
The email that reads like a legal notice
A leasing professional needs to tell a resident their renewal offer changed. She wants it to land warmly. She opens ChatGPT and types: "make this email sound professional."
AI delivers "professional": stiff, formal, distant. "Please be advised that your lease terms are subject to the following modifications." Technically professional. Completely wrong for a four-year resident she wants to keep. She softens the greeting and sends it. The resident still reads it as cold.
AI did exactly what she asked. "Professional" has a dozen meanings and AI picked one. She meant warm and human. She never said so. Vague in, vague out.
Scenario two
The description that says nothing
A marketing coordinator needs a property description for a new listing. She types: "write an engaging description of our property."
AI produces something engaging-sounding: "Nestled in a vibrant neighborhood, our property offers luxury living at its finest." It could describe any property in the country. It names nothing real: not the courtyard, not the walk to the train, not the renovated kitchens. It is fluent and empty.
"Engaging" told AI to sound a certain way, not to say anything specific. AI had no facts to work with because she gave it none. Engaging in, empty out.
Both are specificity problems. Both are avoidable. Neither is AI's fault, but both look like AI's fault to the person who asked for a feeling instead of a result.
Part 1 · Teaching A
What "specific" means to AI.
You have heard "be specific" your whole life. With AI it comes down to one thing. A coworker fills in what you leave out. AI cannot. It guesses. So being specific is simple: leave AI nothing to guess.
Before you hit send, ask one thing: could AI guess wrong here?
You wrote "short." AI guesses how short. Say "under 80 words".
You wrote "friendly." AI guesses what that looks like. Say "open with their first name, end by offering to help".
You wrote "professional." AI guesses: warm, formal, or legal? Say "warm, no legal wording".
The whole idea
That is all "specific" means. Not longer. Not fancier. Just fewer blanks for AI to fill in.
Gut check · 15 seconds
Which one leaves AI nothing to guess?
The second one.
It says the length, the tone, and how to end, so AI is not guessing at any of it. The other two are all feeling and no target. AI will still write something, but it will be AI's guess, not yours.
Part 2 · Teaching B
Five places you can be specific.
Any prompt can be tightened in five places. Most weak prompts are loose in three or four of them. You do not need all five every time, but naming them lets you spot instantly which one you left open.
Dimension 1
Audience · Who is this for?
A four-year resident you want to keep reads differently than a brand-new applicant. "For a resident" is vague. "For a four-year resident renewing at a small increase who we want to retain" is specific.
Dimension 2
Scope · How much, and what to leave out?
"Tell me about our pet policy" invites a page. "List only the three pet rules a new resident must know before signing, one line each" is scoped. Telling AI what to leave out is as useful as telling it what to include.
Dimension 3
Tone · The exact voice, not a vibe word.
Not "professional." Instead: "warm, plain-spoken, no legal phrasing, the way you would explain it to a friend over the counter." A described voice beats a one-word label every time.
Dimension 4
Output format · The shape of the answer.
Length, structure, what to include or omit. "Two short paragraphs, under 80 words each, no bullet points, no header." If you do not name the shape, AI picks one, usually longer than you wanted.
Dimension 5
Constraints · The hard rules and the do-nots.
"Do not mention the dollar amount of the increase. Do not promise anything about future rent. Stay under 150 words total." The do-nots are where most specific prompts do their real work.
Before you paste · Specific is not permission to overshare
Week 1 taught the three data tiers. Making a prompt specific tempts you to paste the resident's whole file so AI has more to work with. Do not. The details AI needs to be specific (a first name, a move-in year, a tone) are almost never the sensitive ones (full legal name plus account number, application details, health or disability information). Consumer ChatGPT, Claude, and Gemini free tiers are Tier 1 only. Give AI specific context with safe placeholders, not real Tier 2 or Tier 3 data. You will practice spotting this on Slide 19.
Gut check · 15 seconds
A prompt reads: "Write an engaging description of our property." Which dimensions would fix it fastest?
Tone and scope.
"Engaging" is a vibe word (it fails the tone test), and the prompt names no real features for AI to use (no scope). Fix both: "In a warm, plain voice, write two sentences that name our three best-known features: the courtyard, the renovated kitchens, and the ten-minute walk to the train." Now AI has a voice and facts instead of a feeling.
Part 3 · Teaching C
Eight words that quietly wreck prompts.
These eight feel like instructions. They are not. Each one asks AI for a feeling without giving it anything to aim at, so AI fills the gap with its best guess. When you catch yourself typing one, that is your signal to stop and say what you actually mean.
Vague word 1
Professional
Could mean warm, formal, or legal. AI guesses. Say the actual voice you want.
Vague word 2
Engaging
A feeling, not a fact. AI adds fluff instead of substance. Name what to include.
Vague word 3
Modern
Modern to whom? Say the real style: "short sentences, no corporate jargon" or "clean and minimal."
Vague word 4
Compelling
Persuasive how? Name the one action you want the reader to take.
Vague word 5
Good
The emptiest word in prompting. Good by whose standard? State the standard.
Vague word 6
Great
Same problem as good, louder. Replace it with a measurable target.
Vague word 7
Nice
Nice is not a spec. Warm? Polite? Brief? Say which one you mean.
Vague word 8
Impactful
Impact on what? Name the outcome: "makes the reader reply within a day."
Not banned, just unfinished
None of these words is forbidden. Each one is a placeholder for a real instruction you have not written yet. The next slide gives you the swap: every vague word, traded for something AI can check.
Gut check · 15 seconds
You wrote: "Make this email more professional and engaging." What is the fastest fix?
Replace the words with what you mean.
"Professional and engaging" gave AI two feelings and zero targets. "Warmer and more personal, and name the one action you want the resident to take" gives AI a voice and a job. Same length prompt. Completely different output.
Part 4 · Teaching D
Every vague word has a specific replacement.
This is the swap. On the left, the vague word you reach for. On the right, what to say instead, phrased so AI can check it. Keep this where you write prompts. In a week or two the swaps become automatic and you stop reaching for the vague word at all.
Vague word
Say this instead, so AI can check it
"Professional"
"Warm and personal, no legal phrasing" (or "formal and brief" - name which one)
"Engaging"
"Name the three real features to include" (facts, not fluff)
"Modern"
"Short sentences, no corporate jargon" or "clean and minimal"
"Good" / "Great"
"Meets this standard:" then name the standard out loud
"Short"
"Under 80 words" or "two sentences" or "one short paragraph"
"Detailed"
"Cover these three points, one line each" (name the points)
"Make it better"
"Change these two things:" then name the two changes
Reflect · 30 seconds
Think of one prompt you sent this week that used a vague word from this table. Which word did you use, and what did you actually mean by it?
Just for you. Naming the gap once makes it far easier to catch the next time.
Saved
Part 5 · Teaching E
Three moves that make any prompt specific.
You do not have to rewrite a prompt from scratch. Take the vague prompt you already have and run it up three rungs. Circle, replace, add. About sixty seconds, and a vague prompt becomes a specific one.
Rung 1 · Circle
Circle every adjective and vibe word.
Professional, engaging, good, nice, modern, compelling. Each circle marks a spot where AI is about to guess what you meant.
Rung 2 · Replace
Replace each with something measurable.
"Professional" becomes "warm, no legal phrasing." "Short" becomes "under 80 words." "Good" becomes the actual standard. Use the swap table from the last slide.
Rung 3 · Add
Add the output format at the end.
The one thing almost always missing: the shape you want. Length, structure, what to include or leave out. One line at the very end of the prompt.
Watch a prompt climb.
Vague: "Write a professional, engaging renewal letter that is short and good."
Climbed: "Draft a renewal letter for a four-year resident renewing at a small increase we want to keep. Warm and personal, no legal phrasing. Two short paragraphs, under 80 words each, no bullet points. Do not name the dollar amount of the increase."
Gut check · 15 seconds
Prompt: "Give me a nice, modern welcome email for new residents." You climb the ladder. What does Rung 1 (circle) catch?
"Nice" and "modern."
Both are vibe words. Rung 2 replaces them ("warm and welcoming, short sentences, no corporate jargon"), and Rung 3 adds the format ("under 120 words: one short paragraph, then three bullet points for move-in steps"). Circle, replace, add. Done in a minute.
Part 6 · Teaching F
The bright line for every prompt.
Everything in this workshop points to one rule. Say it out loud. Write it on a sticky note. Put it on your monitor.
The bright line
If AI has to guess what you meant, you were not specific enough.
That is the whole test. Not "was the prompt long enough." Not "did I use the right words." Just: is there anything left for AI to guess? A property name you did not give. A tone you assumed. A format you pictured but never named. Every guess is a place your output can go sideways.
This is not a complaint that AI is unreliable. AI is reliable at working with what you give it. This is a warning against assuming AI knows what you did not say. Two different things.
The recovery move
When the output is off, do not start over. Ask which guess AI got wrong, then add that one thing: "Redo that, but make it warmer and do not mention the price." Most bad outputs are one missing detail away from right.
Next: the default rule that turns this into a habit. Then, on Slide 12, the vague-versus-specific runs you can compare side by side.
The Workshop · Default Rule
The rule of Week 12.
If you take one thing from this workshop, take this.
The default rule
AI matches your specificity. Say exactly what you want, in terms AI can check.
Vague in, vague out. Specific in, specific out. Every time. When the output disappoints, your first move is not to blame the tool. It is to ask what you left vague, and make that one thing specific.
Your synthesis.
Write down three things you will do differently in your next AI conversation, based on what you just learned. Not aspirational. Actual. Specific to your role.
Commitment card · Three things
The Workshop · Q&A
Questions readers are already asking.
These are the kinds of questions multifamily professionals ask while working through specificity. Tap to expand.
Length is not the goal, specificity is. A specific prompt is often longer than a vague one, but not always. "Under 80 words, no legal phrasing" is more specific AND shorter than a rambling paragraph. Add words only where they remove a guess. If a word is not making the request more measurable, verifiable, or decidable, cut it.
No. Most everyday prompts need two or three, usually tone, output format, and one constraint. The five dimensions are a checklist for when the output keeps missing, not a form to fill out every time. Start with the dimension AI keeps getting wrong.
That is normal and it is fine. Being specific does not mean having the perfect answer in your head first. It means naming what you DO know: the audience, the tone, the length, the one thing to avoid. AI gives you a draft, you see what is off, and you make the next prompt more specific. That back-and-forth is Week 13 (Iteration). Specificity just gives you a better starting point so the iteration is faster.
Close, with one difference. A coworker fills gaps from context you share: they know your property, your residents, your style. AI shares none of that unless you put it in the prompt. So you have to say out loud the things you would let a coworker assume. That is the whole shift.
Your open question.
What is one prompt you want to make more specific (or the vague word you keep reaching for)? Type it here. Saved locally to you.
Power Preview · Setup
Same task. Two prompts. Watch the gap.
One task: draft a renewal letter for a real resident. On the next two slides you will see two prompts for it. First, the vague version most people type. Second, the specific version you now know how to write. Same AI, same task. The distance between the two outputs is the whole point of this workshop.
The task, both runs
Maria Alvarez is a first-year resident in apartment 214 at Riverside on Broad, renewing at a small increase. She has never filed a work order and rated her move-in 5 of 5. Your property's voice is warm and personal. You want her to stay.
Note before the runs
The outputs on the next two slides are composite examples of what these prompts typically produce. Actual AI output varies run to run (temperature, from Week 10). The point is the gap in quality, not any specific word choice.
Power Preview · Run 1 · Vague
Run 1: The vague prompt.
The prompt (as most people type it)
Write a professional, engaging renewal letter for Maria. Make it good.
What AI produces (typical)
Dear Maria,
We are pleased to inform you that your lease at [Property Name] is coming up for renewal. As a valued resident, we would love for you to continue calling [Property Name] home for another year.
Please review the enclosed renewal terms and reach out to our office at your earliest convenience to discuss your options. We appreciate your residency and look forward to another wonderful year together.
Sincerely, [Property Management Team]
What is wrong with this.
"Professional" and "engaging" gave AI a feeling, not a target. It defaulted to corporate.
Nothing about her first year or her 5-of-5 survey. Because none of it was in the prompt.
Placeholder brackets everywhere. [Property Name], [Property Management Team]. The property was never named.
It reads like every renewal letter in the country. Because "good" has no owner and no standard.
None of this is AI failing. AI matched the specificity it was given, which was almost none. Vague in, vague out.
Power Preview · Run 2 · Specific
Run 2: The specific prompt.
The prompt (specific on every dimension that matters)
Draft a renewal letter for Maria Alvarez, apartment 214 at Riverside on Broad. She is a first-year resident renewing at a small increase, quiet, no work orders, and rated her move-in 5 of 5. Tone: warm and personal, no legal phrasing, the way you would talk to her at the counter. Two short paragraphs, under 80 words each. No header, no bullets. Address her by first name and reference her survey naturally. Do not mention the dollar amount of the increase.
What AI produces (typical)
Hi Maria,
Your first year with us at Riverside on Broad has been a pleasure. When you rated your move-in a perfect 5, our whole team took note. It is exactly the experience we want every resident to have, and knowing we got it right for you means a lot.
We would love to have you stay another year. Your renewal is ready for review below. Take a look and let me know what questions you have. I am here whenever you need me.
Warmly, The Riverside on Broad team
What changed. Only the prompt.
Her name and property are here. Because you named them (audience and scope).
Her real 5-of-5 experience shows up. Because you included it.
Warm, personal tone. Because you described the voice instead of typing "professional."
No dollar amount. Because you set that constraint.
Shorter, no headers. Because you named the format.
Same AI. Same task. The gap is the prompt.
The lesson
The specific prompt is not longer for its own sake. Every extra line removed one guess. The 90 seconds you spend being specific saves the 15 minutes you would spend editing a generic letter into something you would actually send.
Power Preview · Try This Today
One exercise you can run this afternoon.
You have 20 minutes today. Do this once. It will change how you write every prompt after.
The exercise · Specificity audit
Step 1. Pull up your last three AI conversations. Any tool.
Step 2. In each opening prompt, circle every vague word: professional, engaging, good, nice, modern, short, detailed.
Step 3. Count them. Most first-time auditors find two to four vague words per prompt, and no output format named at all.
Step 4. Pick one prompt. Replace each vague word using the swap, and add one line naming the output format. Rerun it.
Step 5. Compare the two outputs. Notice how much less editing the specific one needs. That gap is your permanent return on being specific.
A warning about what sounds good.
When your specific prompt works, AI produces fluent, confident text. Week 10 taught you that fluent and confident does not mean correct. Still true. Read what AI gives you and verify anything factual. A more specific prompt often makes AI sound even more like an expert, which is exactly when the risk of believing a wrong answer is highest.
Where to practice safely
The specificity audit is safer to run in a governed workspace.
If your organization is not sure where your team should practice prompt-writing on real resident data, this is the environment designed for it. Data classification, human review, and the RETTC AI Governance Framework are built in. Learn more from Multifamily NEXT at tami@mprodigitaledge.com. mPro Digital Edge: the AI workspace built for using AI with governance, not against it.
Practice this on us
You just spent 25 minutes learning to be specific and to test confident AI. Test this one too. Ask us for a demo. Ask what "governance built in" actually looks like in practice. Ask for third-party reviews. Any AI workspace that survives those questions is worth considering. Any that does not is worth walking away from. Multifamily NEXT wants you practicing that scrutiny on every AI product, including this one.
Share this workshop with a coworker
If Week 12 helped, someone on your team probably needs it too. Send them the link.
Resources & Tools · Vocabulary
Reference · Bookmark this slide
The AI vocabulary, through Week 12.
Every term you have met so far, in plain language. Bookmark this slide. The same list is in the printable cheat sheet and on the download slide, so you always have it.
Term
Plain-language definition
Prompt
Everything you send AI in one message: your instruction plus any context, data, and formatting you include.
Specificity
How precisely a request is stated. AI matches it: vague in, vague out; specific in, specific out.
Vague qualifier
A feeling word that sounds like an instruction but gives AI no target: professional, engaging, good, nice, modern.
Instruction
The verb-plus-object core of a prompt: what you want AI to do. Strongest when it comes first.
Context
The background AI cannot see unless you provide it: who it is for, the situation, the tone, what matters.
Output format
The shape you want back: length, structure, what to include or leave out. Name it in one line.
Constraint
A hard rule or a do-not: word limits, facts to avoid, things not to promise.
Audience
Who the output is for. A four-year resident and a new applicant need different letters.
Scope
How much to cover and what to leave out. Narrowing scope is a form of being specific.
Tone
The described voice you want, not a one-word label. 'Warm, no legal phrasing' beats 'professional.'
System prompt
A hidden instruction the AI tool adds to every chat before your message. You do not see it, but it shapes replies.
Context window
The ceiling on how much AI can hold in view at once. In long chats, the earliest messages drop off.
Token
The small piece of text AI reads and writes. Words are one or more tokens each; the window is measured in tokens.
Temperature
A setting that controls how varied AI output is. Higher means more variety, so the same prompt can differ run to run.
Hallucination
A confident, fluent AI answer that is factually wrong. Fluent is not the same as correct.
Sycophancy
AI's tendency to agree with or flatter you rather than push back. Ask it to challenge your prompt.
Human-in-the-Loop
A person reviews AI output before it is used or sent. Required for anything that reaches a resident.
Custom instructions
Standing context you save once on your AI account so every chat starts with it (ChatGPT), a Project (Claude), or a Gem (Gemini).
LLM
Large Language Model. The kind of AI behind ChatGPT, Claude, and Gemini: it predicts text from your prompt.
Safe placeholder
A stand-in you use instead of real sensitive data, so a prompt can be specific without exposing Tier 2 or Tier 3 information.
Data tier
Week 1's classification of what data may go where: Tier 1 safe for any AI, Tier 2 enterprise only, Tier 3 never.
Iteration
Improving output through a short back-and-forth: your second prompt, sharpened by the first result. Week 13's topic.
Before you hit send on any prompt, run down this list. If you can answer all five, your prompt is specific. If you cannot, that is where AI will guess.
1
Did I remove the vague words? No professional, engaging, good, nice, modern left unexplained.
2
Did I name the audience? AI knows who this is for.
3
Did I describe the tone? A real voice, not a label.
4
Did I name the output format? Length and shape, in one line at the end.
5
Did I set at least one constraint? A word limit, a fact to avoid, a do-not.
Two or three of these cover most everyday prompts. Use all five when the output keeps missing. This is the same rule from Slide 9, in checklist form.
Resources & Tools · Standards Library
Reference · Bookmark this slide
The two frameworks behind everything here.
Every week of AI Tuesdays is anchored to two published US frameworks: the NIST AI Risk Management Framework and the RETTC AI Governance Framework. This is the permanent reference. Bookmark it.
NIST AI Risk Management Framework
The National Institute of Standards and Technology (NIST) publishes the AI Risk Management Framework (AI RMF): the operational structure for managing AI risk, organized around four functions.
Govern
Set the rules and accountability for how AI is used.
Map
Understand the context and the risks of each AI use.
The Real Estate Technology and Transformation Center (RETTC) published the AI Governance Framework for rental housing: the industry-specific values that sit alongside NIST. All eight Pillars, in full:
Pillar 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, and designate accountable ownership for AI oversight.
Pillar 2. Promote Fairness & Consumer Protection
Ensure AI systems comply with applicable provisions of the Fair Housing Act, the Fair Credit Reporting Act, the Americans with Disabilities Act, and relevant state and local data regulations.
Pillar 3. Uphold Transparency & Explainability
Promote a general understanding of when AI influences key decisions in the customer journey.
Provide renters and applicants clear, understandable explanations of decisions affecting them as required by law.
Pillar 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.
Pillar 5. Maintain Outcome-Based Accountability & Human Involvement
Engage humans in significant decisions in the customer journey.
Establish processes to review model outcomes and provide escalation channels for complaints and appeals.
Educate staff on the effective use and limitations of AI tools.
Develop testing and ongoing evaluation to ensure AI tools remain aligned with applicable laws and with business goals.
Encourage evaluation of AI tools based on their ability to maintain consistent and accurate data flows across the renter journey.
Pillar 6. Promote Positive Renter Experiences and Outcomes
Elevate technologies that enhance affordability, access, and housing stability.
Explore AI tools that improve resident experiences by streamlining communication, personalizing services, and supporting inclusive resident engagement.
Consider engaging renters and resident groups in conversations about how AI enhances their housing experience.
Pillar 7. Strive for Innovation with Responsibility
Encourage development of AI tools that benefit residents, prospects, guests, on-site teams, and operators.
Educate all stakeholders on how AI enhances renter opportunity and housing affordability.
Pillar 8. Encourage Third-Party Partner Accountability and Due Diligence
Promote the responsible development of AI tools across external partners.
Invite transparency from third parties regarding data sources, model limitations, and known risks.
Specificity is a governance skill, not just a productivity one. Pillar 5 (human involvement) and Pillar 3 (transparency and explainability) both depend on you being able to say exactly what you asked AI to do and check that it did it. A vague prompt cannot be governed, because no one, including you, can say what it should have produced.
Take Action · Spot the Gap
Eight specificity scenarios.
Pick the best answer for each. This is where Week 12 becomes a reflex. Tap Reset when you want to run them again.
Scenario 1 · Vague word
A leasing pro types: "Write a professional email to a resident about late rent." The email comes back cold and threatening. What went wrong?
"Professional" was the problem.
It has many meanings and AI picked the wrong one. Describe the voice: "firm but kind, assume it was an oversight, and offer a payment plan."
Scenario 2 · Missing audience and scope
A marketer types: "Write a description of our property." AI returns generic filler that could describe anywhere. Fastest fix?
Name the audience and the facts.
AI had no audience and no facts. "For prospects searching downtown, write two sentences naming our courtyard, our renovated kitchens, and the ten-minute walk to the train."
Scenario 3 · No output format
You ask AI to "summarize this 12-page policy" and get five dense paragraphs when you wanted a quick list. What was missing?
The output format.
Name the shape: "Summarize as five bullet points, one line each, in plain language." Output format is the most-skipped kind of specificity.
Scenario 4 · Specific is not oversharing
To make a collections email specific, you are about to paste the resident's full name, apartment, and complete payment ledger into free ChatGPT. Best move?
Give the facts, not the file.
Being specific does not mean oversharing. Free ChatGPT, Claude, and Gemini are Tier 1 only (Week 1). Provide the two relevant facts with a placeholder, or use a governed workspace for real resident data.
Scenario 5 · The recovery move
AI's draft is close but too formal. What is the fastest way to fix it?
Add the missing detail.
Most off outputs are one detail away from right. Name the guess AI got wrong and add it. That is the recovery move from Slide 9, and it is faster than starting over.
Scenario 6 · Measurable beats vague
Two prompts: "keep it short" versus "under 80 words." Which is specific, and why?
"Under 80 words."
"Short" fails the measurable test; AI has to guess how short. "Under 80 words" passes it. Put a number on it wherever you can.
Scenario 7 · The strongest prompt
Which prompt gets the most useful renewal letter?
The second one.
It names the audience, the tone, the format, and a constraint. The others just stack vague words. Specific in, specific out.
Scenario 8 · Specificity as governance
A vendor's AI writes resident letters. Your AI Council asks how you would confirm it is doing the job. What do you need first?
A checkable spec.
Pillar 5 (human involvement) and Pillar 3 (transparency) both need a spec you can check against. If you cannot say specifically what AI should produce, no one can verify that it did. Specificity is a governance tool, not just a writing one.
Take Action · Knowledge Check
Five questions. Anonymous.
Answer all five, then check. Nothing is stored with your name. The tip under each question is the most useful part, so read them even on the ones you get right.
1. You asked AI for something "professional" and got cold, formal output. Why?
Tip: "Professional" has a dozen meanings. Describe the actual voice instead: "firm but warm, assume good faith."
2. Which of these requests is specific?
Tip: "Under 80 words" is measurable. "Short," "professional," and "good" are all guesses. Put a number on it.
3. You wanted a quick list and got five dense paragraphs. What was missing from your prompt?
Tip: Output format is the most-skipped kind of specificity. Name the shape: "five bullets, one line each."
4. To make a prompt specific, should you paste a resident's full ledger into free ChatGPT?
Tip: Specific is not oversharing. Free ChatGPT, Claude, and Gemini are Tier 1 only. Use a placeholder, or a governed workspace for real data.
5. Fill in the blank: "AI matches your ____."
Tip: Vague in, vague out. Specific in, specific out. That is the whole workshop in one line.
Take Action · Download
Take one thing back to your team.
One page to keep, one email to send. Both ready right now.
The Specificity Checklist (PDF)
A one-page card: the five-second specificity check, the swap table of vague words, and the three-move ladder. Print it and keep it where you write prompts.
If your supervisor has not seen Week 12, send them this. The button opens your email client with a draft already written. The address line is blank on purpose: type in your supervisor and send.
Draft (edit before sending)
Subject: What I learned in AI Tuesdays Week 12
Short note that AI matches how specific your request is, plus the one habit you are committing to this week. You fill in your commitment before you send.
Take Action · Session Complete
You finished Week 12.
Twenty-five minutes ago, "make it professional" felt like an instruction. Now you know it is a guess waiting to happen. That is a real shift.
Session complete · What you walked away with
✔ The three tests for specific: measurable, verifiable, decidable.
✔ The five dimensions: audience, scope, tone, output format, constraints.
✔ The eight vague words, and the swap for each one.
✔ The three-move ladder: circle, replace, add.
✔ The recovery move: add the one missing detail instead of starting over (Slide 9).
✔ A one-page checklist to keep at your desk.
Here is the thing. The people who get the most out of AI are not the ones with secret prompts. They are the ones who say exactly what they want. You can do that now. Good job. I am proud of you.
For leaders · Where your team practices this safely
A specific prompt is also a governable one. When your team practices prompt-writing on real resident data, it needs an environment with data classification, human review, and the RETTC AI Governance Framework built in. That is what mPro Digital Edge is: the AI workspace built for using AI with governance, not against it. To see it, reach Multifamily NEXT at tami@mprodigitaledge.com.
Next Tuesday · Week 13
Iteration: your second prompt is almost always better.
This week you learned to write a strong first prompt. Next week you learn the back-and-forth that turns a good first draft into exactly what you needed, in two or three quick turns. Specific gets you started; iteration gets you finished. See you Tuesday.
Tami Siewruk
CEO, Multifamily NEXT and mPro Digital Edge
Built by Multifamily NEXT. Powered by mPro Digital Edge, the AI workspace for governed AI use in multifamily.
MID-YEAR 2026 EDITION - UPDATED JUNE 2026
AI Regulations and the Multifamily Industry
What changed in 2026 and what your communities need to do about it. Plain-language guides covering laws in effect, $141.8M in pricing software settlements, attorney general lawsuits by state, and your compliance playbook. Written for operators, not lawyers.
Why regulations moved beyond "AI" to "automated decision-making technology." Your screening, pricing, chatbots, renewal scoring, and hiring tools are all in scope. Plus: the end of the vendor shield defense.
Shadow AI and On-Site Risk
The risk most guides miss: on-site teams using unapproved AI tools to draft resident notices and process data outside your governed environment.
Federal Landscape and HUD Enforcement
Executive Order 14409, six federal bills moving through Congress, and HUD's position that Fair Housing applies whether a decision is made by a human or an algorithm.
State-by-State Laws in Effect
Colorado SB 26-189 (ADMT framework), California AB 325 (algorithmic pricing), Texas TRAIGA, Illinois HB 3773, NYC Local Law 144, and California CCPA rules. What each requires and what to do about it.
Pricing Software Lawsuits: $141.8 Million
The DOJ consent decree rules, the five requirements every pricing vendor must meet, 26 operator settlements, the Yardi litigation track, and private right of action risk.
AG Lawsuits by State
Enforcement actions in D.C. ($1.4M), Arizona ($1M), New Jersey, New York, and Florida's notable absence. Federal compliance does not guarantee state immunity.
Legislative Trends to Watch
Five categories of state legislation trending across more than two dozen states. Algorithmic pricing, ADMT disclosure, hiring AI, data privacy, and chatbot transparency.
Your Compliance Playbook
Three priority actions to start this quarter, 10 total action items, guidance for smaller operators, plus vendor MSA review, insurance coverage, and data purging protocols.
Download the Reports
Two comprehensive guides. One for U.S. operations, one for Canadian. Both in plain language. Both updated quarterly for NEXT Steps members.
UNITED STATES
AI Regulations Guide for Multifamily Housing
Fair Housing foundation, ADMT framework, operator liability, Shadow AI, HUD enforcement, EO 14409, six federal bills, state laws (CO, CA, TX, IL, NYC), $141.8M in settlements, AG lawsuits by state, legislative trends, and your prioritized compliance playbook.
25 pages · 7 parts · 6 state laws · 10 action items · Mid-Year 2026
Federal AIDA collapse and National Sprint, Quebec Law 25 (toughest in North America), Ontario Bill 194 and IPC-OHRC principles, Alberta/BC modernization, Competition Bureau investigation, CAN/DGSI 101 standard, and resident screening liability.
Comprehensive analysis · 3 provinces · Federal + provincial · Q1 2026
The question is no longer "are we using AI?" The question is "are any of our tools making decisions about people?" For every multifamily operator, the answer is yes.
From the Guide: Part 1, The ADMT Framework Shift
Who This Guide Is For
If you run, manage, or invest in apartment communities and you use any automated tools for screening, pricing, leasing, or hiring, this guide is for you.
C-Suite Executives
Boards and investors are asking about AI risk. The executive summary gives you the five things you need to know in two pages.
Regional Managers
Your portfolio spans multiple states. This tells you which laws apply where and what to prioritize first at your communities.
Marketing Directors
Your chatbots, automated ads, and AI-driven content tools are regulated. Know the disclosure rules before they create a compliance gap.
Compliance and Legal Teams
State-by-state penalties, AG enforcement actions, private right of action risk, vendor MSA language, and the NIST framework alignment.
These Guides Are One Piece of NEXT Steps
NEXT Steps is the continuing education program for multifamily professionals who are serious about building AI fluency. The quarterly AI Regulatory Reports for the U.S. and Canada are just two of the resources you get as a member.
Quarterly AI Regulatory Reports (U.S. + Canada)
Every quarter, you receive updated guides covering new laws, enforcement actions, legal precedents, and compliance deadlines for both the United States and Canada.
Monthly Micro-Learning
Focused lessons you can complete in 15 minutes. Practical AI skills your team can put to work the same week.
Live Sessions
Hands-on workshops and Q&A sessions with Tami Siewruk and the Multifamily NEXT team. Bring your real questions.
Community of Builders
Connect with multifamily professionals who chose to act, not just learn. Share what works, ask for help, grow together.
Tools and Templates
Governance checklists, audit trackers, vendor evaluation frameworks, and other resources built specifically for multifamily operations.
Stay Current Without the Work
The AI regulatory landscape is changing every quarter. We track it so you do not have to. You get the update. You stay informed. You stay compliant.
Get Both Reports and Every Quarterly Update After Them
NEXT Steps members receive updated U.S. and Canadian AI Regulatory Reports every quarter, along with monthly micro-learning, live sessions, and a community of multifamily professionals building AI fluency together.
