tami@mprodigitaledge.com
Password: 0952
Welcome back. Let us warm up first.
Last week you learned what generative AIGENERATIVE AIA fluent word predictor trained on human writing. It looks at what you typed, plus everything it has already written, and picks the next most probable word. Then does it again. Then again. That is the whole trick. It is genuinely good at drafting, summarizing, structuring, and explaining. It reliably fails at hallucination (making things up), agreement with wrong framing, current data, math, and legal specifics. Multifamily example: When ChatGPT drafts a renewal letter, it is not "thinking" about the resident. It is predicting what words most commonly follow a request like yours. That is why the pattern (a warm renewal letter) comes out right and the specifics (rent amount, move-in date) will confidently come out wrong if you did not paste them in. Full deep dive: Week 08 (What AI Is in Plain English). Anchor: RETTC Principle 3 (Uphold Transparency & Explainability). actually is. Three quick recap questions to bring it forward. Tap any answer to reveal the coaching, no wrong tries counted.
Skip the warmup anytime. Click Begin Reading when you are ready.
AI vs Search vs Database
Last week you learned what generative AI actually is. Now here is the question that comes next. When someone asks you a question, which of the three tools you have on your desk should answer it? A search engine? Your Property Management SystemPROPERTY MANAGEMENT SYSTEM (PMS)The database of record for your property. Holds resident names, apartment numbers, rent amounts, lease dates, ledger history, work orders, and vacancy data. Examples in multifamily: RealPage, Yardi, Entrata, ResMan, AppFolio, Rent Manager. Multifamily example: when a resident asks "what is my balance," the answer lives in the PMS. It is not on Google. It is not in ChatGPT. AI cannot make it up correctly. The PMS is the single source of truth for anything about a specific resident or apartment. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity).? Or AI? Answering that correctly, every time, is what Week 09 is about.
When you close this workshop, you will never again ask AI a question that a database should answer. Or trust AI for a fact that lives on Google. That is the whole promise.
Beginner-friendly · Read at your own pace · Star slides, add notes, print, or email yourself anytime
Agenda.
Six short parts. About twenty minutes total. No timer. Pause whenever you need. Star any slide you want to come back to. Add your own notes on any slide. Everything you type stays in your browser.
Read at your own pace. Star any slide you want to bookmark. Add notes on any slide. Everything you type stays in your browser and is yours to keep.
Two stories from real properties.
This week is short because the mistake it prevents is simple, common, and costly. Two stories.
A property manager gets a supervisor's question by email late Friday afternoon: "How does our current delinquency compare year-over-year?" She does exactly what a smart property manager does. She pulls the year-to-date delinquency files from her Property Management SystemPROPERTY MANAGEMENT SYSTEM (PMS)The database of record for your property. Holds resident names, apartment numbers, rent amounts, lease dates, ledger history, work orders, and vacancy data. Examples in multifamily: RealPage, Yardi, Entrata, ResMan, AppFolio, Rent Manager. Multifamily example: when a supervisor asks "what is our current delinquency," the numbers live in the PMS. That is the source of truth. AI cannot make those numbers up correctly and cannot draw conclusions the numbers do not contain. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity).: current period, prior period, prior year same period. Three clean reports. Real numbers from the source of truth.
She uploads all three files to her approved AI tool and asks: "Compare these three periods. What are the key differences?" The AI produces a confident analysis: "Year-over-year delinquency is up approximately 12.4% from the same period last year. The largest driver is a seasonal pattern that historically shows in late Q3, compounded by continued market softening in the mid-tier apartment segment. Prior period comparison shows delinquency rising from 2.1% to 2.6%, a proportional increase of 23.8%."
Clean. Analytical. Professional. She forwards it to her supervisor with a note: "Full analysis attached. Summary below."
Monday morning her supervisor asks two questions. First: "Where did 12.4% come from? I did the math from your files and I get 8.7%." AI had the real data. AI got the math wrong. A comparison paragraph needs a percentage to sound complete, so the model produced one. Second: "What seasonal pattern? What market softening? The files you sent don't say anything about causes." They do not. AI fabricated the reasoning. A paragraph about drivers needs drivers, so the model invented them.
She used the right tool for the data. She used the wrong tool for the analysis. A probabilisticPROBABILISTICSame input can produce different outputs. AI is probabilistic. Ask it the same question ten times, you may get ten slightly different answers. Compared to deterministic systems (databases and arithmetic), where the same input always produces the same output. Multifamily example: ask an LLM "draft a warm renewal letter for a resident" ten times, you will get ten different letters. Ask your PMS "what is the current rent for apartment 214" ten times, you will get the same rent amount ten times. Two very different tools. Two very different jobs. Anchor: RETTC Principle 3 (Uphold Transparency & Explainability). tool cannot do the arithmetic a deterministicDETERMINISTICSame input, same output, every time. Math is deterministic. Databases are deterministic. Compared to probabilistic systems (AI), where the same prompt can produce different answers each time. Multifamily example: your PMS is deterministic. The math on a delinquency comparison is deterministic. AI producing a plausible-sounding percentage from the same files is probabilistic. Never let a probabilistic tool answer a question a deterministic tool owns. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity). tool owns. And a probabilistic tool cannot invent the causes a data set does not contain. Two jobs. Two failures. One smart workflow that still landed her in the wrong place.
That is what today prevents.
A property manager needed to draft a renewal letter with a warm tone. She was not sure how to start. So she opened Google. She searched "how to write a renewal letter with a warm tone." She read four blog posts. She copied phrases from two of them. She rewrote them into a draft. Twenty minutes gone.
Meanwhile, an AI drafting tool at her disposal would have produced a solid first draft in thirty seconds, ready for her edits. That is a job where AI shines: drafting.
She used the wrong tool. It cost her twenty minutes she did not have. That is what today prevents on the other side.
Both mistakes cost you. Both are avoidable once you can tell the three tools apart. That is what the next fifteen minutes fix.
Meet the three tools on your desk.
You already use all three every day. You may not have named them clearly. Once you do, the "which tool answers which question" call becomes automatic.
Search Engine
Google, Bing, DuckDuckGo, HUDHUD (U.S. DEPARTMENT OF HOUSING AND URBAN DEVELOPMENT)The federal agency that oversees Fair Housing enforcement, publishes Fair Market Rent (FMR) limits, and administers HUD-assisted housing programs. HUD.gov is the authoritative source for current federal housing guidance. Multifamily example: when you need current Fair Housing guidance or current FMR limits for a metro area, search hud.gov directly. Do not ask AI, which may quote outdated numbers or invent guidance. Anchor: RETTC Principle 2 (Promote Fairness & Consumer Protection). (U.S. Department of Housing and Urban Development) at hud.gov, your ILSILS (INTERNET LISTING SERVICE)Paid apartment listing platforms where prospects search for rentals. This is what multifamily uses instead of an MLS (Multiple Listing Service, which is single-family real estate). Multifamily example: Apartments.com, Zillow Rentals, Zumper, ApartmentGuide, Rent.com. An ILS is a searchable index of apartment listings, where the platform decides which listings show up first. When a leasing team wants to check competitor pricing, they can search their ILS. When a prospect wants to compare properties, they use an ILS. Anchor: RETTC Principle 3 (Uphold Transparency & Explainability). (Internet Listing Service) platforms like Apartments.com or Zillow Rentals.
What it does: Finds public information that already exists somewhere on the internet. Same query, same top results.
Search finds.
Database
Your PMS, your ledger, your rent roll, your CRMCRM (CUSTOMER RELATIONSHIP MANAGEMENT SYSTEM)The database of record for your prospects and leasing pipeline. Holds prospect inquiries, tour requests, follow-up history, applications-in-progress, which marketing source brought each prospect in, and how far each prospect has moved through the leasing process. Different from your Property Management System: the PMS handles current residents; the CRM handles people BEFORE they become residents. Multifamily example: Knock, LeaseHawk, Funnel, and Anyone Home are widely used CRM platforms in multifamily. When a prospect asks "when can I tour," the answer lives in the CRM. When a resident asks "what is my balance," the answer lives in the PMS. Both are deterministic databases. Both are sources of truth. Never let a probabilistic tool answer a question either of them owns. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity). (Customer Relationship Management system, like Knock), your compliance trackerCOMPLIANCE TRACKERA system that tracks completion of legally required or company-required tasks: Fair Housing training, background checks, insurance renewals, license renewals, safety inspections. Can be a standalone platform or a module inside your PMS or HR system. Multifamily example: when a regional manager asks "when was our team's last Fair Housing training," the answer lives in the compliance tracker. Like your PMS and CRM, it is a deterministic database. Ask it, do not ask AI. Anchor: RETTC Principle 2 (Promote Fairness & Consumer Protection)..
What it does: Retrieves specific facts your organization owns. Same query, same exact answer.
Database retrieves.
Generative AI
ChatGPT, Claude, Gemini, Copilot, and every leasing chatbot with an LLM under the hood.
What it does: Generates new text based on patterns in its training data. Same prompt can produce different answers.
AI generates.
Three jobs. Three different answers to "who owns this question." Write those three verbs down: Search finds. Database retrieves. AI generates. That is the whole workshop in seven words.
What a search engineSEARCH ENGINEA software system that finds public web pages (or documents in an index) matching keywords you type. Ranks them by relevance and popularity. Examples: Google, Bing, DuckDuckGo, plus site-specific search on HUD.gov, your ILS, or your compliance library. Multifamily example: when you Google "current HUD Fair Market Rent limits for Denver," Google is not "thinking." It is running your keywords against an index of the web and ranking the pages that best match. If HUD updates their limits, Google can find the new page. If the page does not exist yet, Google cannot invent it. Compared to AI, which will confidently invent limits from its training data if you ask. Anchor: RETTC Principle 3 (Uphold Transparency & Explainability). actually does.
A search engine is a keyword-matching system that indexes public web pages and ranks them for your query. That is the whole thing. It does not "know" anything. It does not reason. It looks in its index for pages that match your keywords, sorts them, and shows you the top results.
Three things a search engine does well
Three things a search engine cannot do
What a databaseDATABASEA structured storage system that holds specific facts in specific fields so software can retrieve them precisely. Deterministic: same query, same answer, every time. In multifamily, your Property Management System (PMS), your accounting ledger, your CRM, your compliance tracker, your maintenance ticketing system, and your vacancy report are all databases. Multifamily example: when the ledger says "$1,247.00 owed," that is a database returning a stored value. It is not an opinion. It is not a guess. It is the single fact the organization has agreed to treat as true. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity). actually does.
A database is a structured storage system. Facts live in specific fields. Software retrieves them precisely. Same query, same answer, every time. In your world you actually have two big databases. Your Property Management System (PMS) for current residents: RealPage, Yardi, Entrata, ResMan, AppFolio, Rent Manager. And your CRMCRM (CUSTOMER RELATIONSHIP MANAGEMENT SYSTEM)The database of record for your prospects and leasing pipeline. Holds prospect inquiries, tour requests, follow-up history, applications-in-progress, which marketing source brought each prospect in, and how far each prospect has moved through the leasing process. Different from your Property Management System: the PMS handles current residents; the CRM handles people BEFORE they become residents. Multifamily example: Knock, LeaseHawk, Funnel, and Anyone Home are widely used CRM platforms in multifamily. When a prospect asks "when can I tour," the answer lives in the CRM. When a resident asks "what is my balance," the answer lives in the PMS. Both are deterministic databases. Both are sources of truth. Never let a probabilistic tool answer a question either of them owns. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity). for prospects: Knock, LeaseHawk, Funnel, Anyone Home. Same behavior. Different data. The PMS is the one you touch most, so this slide focuses there. CRM callout follows.
Your PMS is the source of truthSOURCE OF TRUTHThe single system your organization has agreed to treat as authoritative for a given piece of information. When two systems disagree, you check the source of truth. When you have a question, you ask the source of truth first. Multifamily example: for anything about a specific resident, apartment, ledger balance, lease term, or vacancy status, your PMS is the source of truth. For anything about a prospect or leasing pipeline, your CRM is the source of truth. If ChatGPT says one number and your PMS or CRM says another, the database wins every time. Establishing a source of truth for each data category is Council-level work. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity). for residents. Categories it owns.
Your Customer Relationship Management (CRM) system holds prospect inquiries, tour requests, follow-ups, application-in-progress data, which marketing source brought each prospect in, and how far each has moved through the leasing process. Knock is the widely used example, alongside LeaseHawk, Funnel, and Anyone Home. Same rule as your PMS: deterministic source of truth. When a prospect asks "when can I tour," the answer lives in your CRM. Never ask AI. Never rely on search. Ask your CRM.
Why deterministic beats probabilistic here
Ask your PMS "what is the current rent for apartment 214." You get the same number ten times in a row. That is deterministic. Ask ChatGPT the same question. If it answers at all, it will fabricate a number, and the number may differ each time. That is probabilistic. For anything the organization has agreed is a fact, deterministic wins. Every time. No exceptions.
What generative AI actually does (quick recap from Week 08).
Week 08 covered this in depth. Fast recap here so the three-way distinction lands clean. Generative AI is a probabilistic word predictor trained on human writing. When you send it a prompt, it does not look anything up. It does not query a database. It generates a response one token at a time, based on what usually follows in the kind of writing it has seen.
The three ways AI is different from the other two tools
Where AI is genuinely the right tool
When the task is drafting, summarizing, structuring, explaining, or translating, and the specific facts required are ones you can paste in or verify after. That is the shine zone from last week, unchanged this week. Nothing about the three-tool distinction takes anything away from AI. It just tells you when to use it and when to reach for a different tool.
If Week 08's teaching stuck: AI is the fluent word predictor. It shines at drafting, structuring, summarizing, translating. It fails at facts, math, current data, and legal specifics. Week 09 adds one thing: those failures happen because AI is the wrong tool for those jobs. Search or your database is the right tool.
The one question that picks the tool.
Before you type a word into any of the three tools, ask yourself one question. The answer picks the tool automatically.
"Is the answer to this question already a fact, and if so, whose fact is it?"
The three paths that question opens
Notice: the question does not ask "which tool is best." It asks "whose fact is it?" That is the question that picks the tool. Trust me. Write it down.
Three anchor scenarios. Then the one rule.
Three scenarios you will hit this week. Walk through each. Notice which tool the framework picks.
This is our fact. Specific apartment. Specific number the organization owns. Deterministic answer. Go to the PMS. Do not ask AI. AI does not know 214 exists and will confidently make up a number.
This is a public fact. It exists on the internet. Rent comps sites, industry reports, competitor property listings, National Apartment Association surveys. Go to search. Look at three sources. The variance is data too. AI may give you a number, but you cannot tell if it is current or invented.
This is new writing. The output is a letter that did not exist before. AI shines. Pull the specific facts (Jane's name, move-in date, current rent) from the PMS. Follow your Council's approved workflow for what data you paste in. Two cautions: redaction alone is not enough (dates plus rents plus apartment number can still identify a specific resident), and even when AI has real data pasted in, it can still fabricate math or reasoning (Story One on Slide 3 shows exactly how). Human reviews every number and every claim before the letter goes out.
The one rule
The one bright lineBRIGHT-LINE RULEA term borrowed from law: a clear, unambiguous rule with no gray area. Something you either did or you did not. In multifamily governance we use this kind of rule for the places where interpretation gets you in trouble. Multifamily example: "Never let a probabilistic tool answer a question a deterministic tool owns." There is no partial version of that. Either the ledger answered it, or ChatGPT made something up. Week 07 uses one for the AI Council (no autonomous decisions that affect residents). Week 08 uses one for verification (every AI output gets human review before it ships). This week uses one for tool selection (probabilistic tools do not answer deterministic questions). Three clear rules. Zero interpretation. for picking your tool: never let a probabilistic tool answer a question a deterministic tool owns. Ledgers, rent amounts, resident balances, specific apartment vacancy: these live in the Property Management System. Not in AI. Not in a search engine. In the PMS. End of story.
There is no partial version of this rule. Either you pulled the number from the PMS, or you did not. Either you searched the actual HUD site, or you asked AI. Either the AI generated fresh text, or you asked it to state a fact only your database owns. Three tools. Three jobs. One rule separating them.
The Week 09 default rule.
One question. Three answers. Three tools. Here is the whole framework on one card. Print this if you print nothing else this week.
Questions people ask.
Tap any question to expand the answer.
No. When AI "searches the web," it is running a separate search under the hood and then summarizing what it finds. If the underlying source is wrong, or the AI misreads it, the answer is confidently wrong. For anything where the exact source matters (Fair Housing guidance, HUD updates, state law, vendor documentation), go directly to the source. AI is a good starting point. It is not the final answer.
Yes. And that is exactly why it is your single source of truth for anything about a specific resident, apartment, or ledger. A database is deterministicDETERMINISTICSame input, same output, every time. Ask a database "what is the current rent for apartment 214" ten times, you get the same answer ten times. Compared to probabilistic systems (AI), where the same prompt can produce a different answer each time. Multifamily example: your PMS is deterministic. Your ledger is deterministic. AI drafting a renewal letter about that resident is probabilistic. Never let a probabilistic tool answer a question a deterministic tool owns. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity).. AI is probabilistic. Never let a probabilistic tool answer a question a deterministic tool owns.
Not without an integration you can point to and the AI Council has approved. Consumer AI tools (ChatGPT, Claude, Gemini) do not have access to your PMS. If you paste PMS data into them, you have just moved Tier 2 data into a Tier 1 tool. That is a governance failure. Some enterprise AI platforms integrate directly with major PMS vendors under contract. Those are a different conversation, and they route through the AI Council for approval.
Your PMS. It is the system your organization has already contracted with, secured, and audited under a Business Associate Agreement or equivalent. Public search engines never touch resident data (there is no PII to search on Google). Public AI tools do touch resident data if you paste it in, and that is where most of the risk lives. Rule of thumb: resident-specific data belongs in the PMS. Nowhere else without the AI Council's blessing.
Sample prompt: "which tool should I use here?"
The Power Preview walks through a decision-support prompt you can use as a coworker or teammate hits you with a request they do not know how to route. Same template you can adapt for any request. Uses the RTF prompt structure (Role, Task, Format). Full RTF treatment: Week 14.
The RTF prompt
Task: Read the request I paste below. Then tell me: (1) which of the three tools is right, (2) why in one sentence, (3) one warning about the WRONG tool that the person might reach for by accident. If more than one tool is involved (which is common), name each tool and what it does in the workflow. Do not hedge. Do not add general advice.
Format: Three short sections labeled "Tool," "Why," "Watch out." Under 120 words total. No bullets. No preamble. If any resident data is involved, remind me to follow my property's data classification rules from Week 01.
The request: [Paste the actual request here.]
Next two slides: what happens when you use a weak prompt versus the RTF version. Same request. Two very different outputs.
What happens with just "help me route this."
Same AI. Vague prompt. Real request pasted in. Watch what comes back.
Great question. Pet rent varies widely across the multifamily industry depending on factors such as location, property class, and pet type. Typically, competitive properties charge somewhere between $25 and $75 per month for a small dog, though some luxury properties charge more. Best practices suggest researching your market thoroughly and considering both monthly pet rent and one-time pet fees. I would recommend reaching out to your regional manager and pulling comparable data from your local market. Additionally, it is worth considering...
What went wrong
You would still have to figure out on your own that the answer is: search the competitor's website. That is what a bad prompt costs. And notice AI never once said "I do not know what your specific competitor charges." That is the pattern from Week 02: hallucinationHALLUCINATIONWhen AI confidently produces false information as if it were fact. Not on purpose. AI is a word predictor: when it does not know the answer, it generates a plausible-sounding one anyway. The number, the date, the citation may look right and be completely made up. Multifamily example: asking AI "what is the current pet rent at my closest competitor property" and getting a confident dollar figure that was invented on the spot. Full treatment: Week 02 (Hallucinations and Sycophancy). Anchor: RETTC Principle 5 (Maintain Outcome-Based Accountability & Human Involvement). as confident guessing.
Same request. RTF prompt. Clean routing.
Task: name the right tool, one-sentence why, one warning about the wrong tool.
Format: three sections (Tool, Why, Watch out). Under 120 words.
Request: "A leasing pro asked what is the current pet rent our closest competitor charges for a small dog?"
Tool. Search engine. Go directly to the competitor property's own website or its listing on a major rentals site. Cross-check with two additional local competitors for range.
Why. Pet rent is a public fact the competitor publishes. Search takes you to the authoritative source that already exists.
Watch out. Do not ask AI directly. It will confidently produce a number from training data that is either outdated or invented. And do not ask your PMS. Your PMS holds your property's pet rent, not the competitor's.
What changed
Same AI. Different prompt. Different world. The prompt did the work. Full RTF treatment: Week 14.
Your turn. Ten minutes today.
Pick one request that came at you this week that you were not sure how to route. Copy the RTF prompt from Slide 12. Paste the request at the bottom. Send it to your approved enterprise AI toolENTERPRISE AI TOOLAn AI product with a business-grade data agreement that meets your governance requirements. Different from consumer ChatGPT or free-tier consumer AI in three ways: (1) contractual data protection (your prompts are not used to train the vendor's models), (2) admin controls (audit trail, user management, retention policies), and (3) Council-approved for Tier 2 data with proper redaction. Multifamily example: ChatGPT Enterprise, Claude for Work, or Microsoft Copilot with an enterprise agreement are enterprise AI tools. Consumer ChatGPT (free or Plus tier) is NOT. A leasing professional can paste properly redacted renewal-letter content into ChatGPT Enterprise but not into consumer ChatGPT. If you are not sure which tool your property has approved, ask your AI Council before pasting anything sensitive. Anchor: RETTC Principle 4 (Protect Privacy & Promote Data Integrity).. Read the output. Verify the tool recommendation against your own judgment. Then use whichever tool the framework picked.
If AI recommends its own use for a specific-resident question, that is your sycophancySYCOPHANCYWhen AI agrees with your framing even when your framing is wrong. AI is trained to be helpful, and "helpful" in training often meant "agreeing with the user." So when you ask a question with a flawed assumption, AI often accepts the assumption instead of pushing back. Multifamily example: asking AI "draft this eviction notice for late payment" when the situation actually calls for a payment plan first: AI drafts the eviction without questioning the framing. Full treatment: Week 02 (Hallucinations and Sycophancy). Anchor: RETTC Principle 5 (Maintain Outcome-Based Accountability & Human Involvement). alarm from Week 02. AI wants to be helpful. Sometimes helpful looks like agreeing to do a job it should not do. Route to the PMS anyway.
NIST AI Risk Management Framework and the RETTC 8 Principles are built into every tool. Try what you learned today in the workspace built for it.
One person. One tool decision. One week. That is how the "AI is unreliable" story on your team stops being about AI and starts being about tool selection.
The AI Vocabulary Cheat Sheet.
The core vocabulary for talking about AI in a multifamily context, updated for Week 09. All 21 terms. Every one shows up in vendor pitches, board conversations, and industry press. Keep this next to your desk.
The 21 core terms across Weeks 08 and 09. Same list appears on Slide 21 and in the downloadable PDF.
Two-page printable PDF: all 21 terms above, plus the three-way tool matrix (Slide 17) and the Week 09 default rule (Slide 10). Print and keep next to your desk.
↓ Download the Cheat Sheet (PDF)The three tools side by side.
A one-page reference. Bookmark or print. Bring to your next team meeting when someone reaches for the wrong tool.
The one-question decision walkthrough
Three questions. Any request gets routed correctly in under thirty seconds.
NIST AI Risk Management FrameworkNIST AI RMF (NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY AI RISK MANAGEMENT FRAMEWORK)A voluntary US government framework for managing risks associated with AI systems. Published January 2023 by the National Institute of Standards and Technology, a bureau of the US Department of Commerce. Not a law. Not regulation. A structured way to think about, measure, and manage AI risk that has become the de facto US standard for AI governance. The four functions: GOVERN (culture, policies, and accountability structures for AI risk). MAP (understanding context and categorizing what your AI system is). MEASURE (assessing and tracking AI risks in a specific implementation). MANAGE (prioritizing and responding to AI risks based on impact). Multifamily example: When your AI Council writes a policy that says "no autonomous decisions affect residents," that is GOVERN. When you categorize a leasing chatbot as "LLM with system prompt, Tier 2 data allowed," that is MAP. When you audit AI-drafted renewal letters quarterly for Fair Housing risk, that is MEASURE. When you decide which use cases get approved with guardrails vs. refused, that is MANAGE. Four functions. One rhythm. Link: https://www.nist.gov/itl/ai-risk-management-framework and RETTC AI Governance Framework.
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 09 concept lands somewhere in one or both.
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.
This week bridged to the frameworks
Source documents
Spot the tier: 8 scenarios.
Each scenario names a proposed workflow. Tap Green (safe to proceed), Yellow (Council review required), or Red (do not proceed as described). Feedback shows immediately. No penalty for wrong answers, just coaching. Reset all with the button at the bottom. All eight scenarios exercise this week's three-way distinction.
Five questions. Anonymous. No wrong tries counted.
Answers submit anonymously to Tami for curriculum improvement. No name, no property, no attribution. Reset with the button at the bottom.
The Week 09 action pack.
Four accordion sections. Each is a copyable, printable, forwardable resource that lives in this workshop and travels with you. Below them: an email to send to your supervisor with the vocabulary cheat sheet and a note about the tool decision you want to talk through.
The full 21 canonical terms. Same list as Slide 16 and the downloadable PDF. Copy any row you need to send to a coworker.
Never let a probabilistic tool answer a question a deterministic tool owns. Ledgers, rent amounts, resident balances, specific apartment vacancy: these live in the Property Management System. Not in AI. Not in a search engine. In the PMS.
"We use three different tools for three different jobs. Our Property Management System (PMS) and CRM are our sources of truth for anything about a specific resident, apartment, or prospect. Public search engines are our source for public facts and current information. Generative AI drafts, summarizes, and structures work our people then verify. We never let a probabilistic tool answer a question a deterministic tool owns. Anchored in NIST AI RMF and RETTC AI Governance Framework."
Download and send it up your chain
Two-page printable PDF: all 21 terms, the three-way matrix, the default rule, the board paragraph.
↓ Download the Cheat Sheet (PDF)Email a copy to your supervisor with a note about one tool-selection decision at your property you want to talk through. Fill in your supervisor's email address before sending. Never pre-fill.
Fill in your supervisor's email in the To field before sending. The email opens with the message pre-drafted for you to edit.
You know which tool answers which question now.
What you walked away with
Download the AI Vocabulary Cheat Sheet
All 21 terms across Weeks 08 and 09, the three-way tool matrix, and the Week 09 default rule. Two pages. Print and keep next to your desk.
↓ Download the Cheat Sheet (PDF)For your board · one paragraph
"We use three different tools for three different jobs. Our Property Management System (PMS) and CRM are our sources of truth for anything about a specific resident, apartment, or prospect. Public search engines are our source for public facts and current information. Generative AI drafts, summarizes, and structures work our people then verify. We never let a probabilistic tool answer a question a deterministic tool owns. Anchored in NIST AI RMF and RETTC AI Governance Framework."
Where to practice this safely
mPro Digital Edge is the enterprise AI workspace for multifamily. Every role in the organization, from on-site leasing to the C-suite, works inside a governed environment where the NIST AI Risk Management Framework and the RETTC 8 Principles are built into every tool.
One platform where every person in the organization gets measurably better at their job with every interaction.
Good job. See you next 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.
