An AI training proposal is not priced by the hours in the room. It is priced by one thing the document rarely states outright: whether anyone at your firm will still be using AI 90 days later. JLL’s 2025 Global Real Estate Technology Survey found roughly nine in ten commercial real estate firms piloting AI and only 5% hitting all of their program goals. A training proposal is your firm’s bet against that gap, so read it for the line items that produce lasting use, not the ones that produce a pleasant afternoon. This guide walks a principal at a small CRE firm through such a proposal line by line: what belongs in it, which deliverables carry outcomes and which are theater, the red flags to watch, and what a right-sized scope and price look like for a 4–20 person shop.
What an AI training proposal contains
Strip a good proposal to its parts and six things should be present: objectives tied to your business, learning outcomes stated as capabilities, deliverables, a format and agenda, a measurement plan, and a price with what drives it. Most proposals cover the first four. The two that separate a real program from a paid seminar are the outcome statements and the measurement plan.
Watch the language in the outcomes section. A weak proposal lists topics: “prompt engineering fundamentals,” “an overview of generative AI,” “responsible use.” A strong one lists capabilities your people will hold afterward: a broker producing a first-pass letter of intent from deal terms in under fifteen minutes, a property manager extracting and verifying key lease terms from a PDF in under thirty. Topics describe what the trainer will say. Capabilities describe what your firm can do on Monday, and only the second kind survives a busy week.
Deloitte’s 2026 Commercial Real Estate Outlook recommends treating AI literacy as a companywide learning program tied to role expectations, with tracked outcomes rather than a one-time event. That is enterprise language for a small-firm truth: the proposal you want reads like a plan to change how work gets done, not one to fill a calendar slot. The 90-day CRE training playbook lays out the drills and milestones a good program should follow, and a vendor proposal is worth reading against that standard.
Outcome line items versus vanity deliverables
This is the read that matters most. Every proposal is a list of deliverables, and each one either changes how your firm works or decorates the invoice. Sort them into two columns first.
Outcome line items produce artifacts your team keeps using. A shared prompt library built on your own documents (your LOI format, your lease-summary template, your market write-up structure) is the single most valuable thing a workshop can leave behind, because it becomes the closest thing a small firm has to documented process. Role drills on live work, a verification checklist for catching invented numbers, a follow-up session a few weeks out, and a re-measured before-and-after baseline all belong here. They are the parts that make usage stick.
Vanity deliverables photograph well and change nothing: a completion certificate, a generic slide deck nobody reopens, a 90-minute keynote about “the future of AI,” a vendor-branded PDF of prompts written for no firm in particular. None of these are scams. They are simply not what you are buying, and a proposal weighted toward them is selling an event, not a capability.
| Deliverable | Column | Why |
|---|---|---|
| Prompt library built on your firm’s documents | Outcome | Becomes reusable process; outlives the session |
| Role-based drills on live LOIs, leases, write-ups | Outcome | Builds the habit where the work happens |
| Verification checklist for AI output | Outcome | Trains the eye for your document types’ failure modes |
| Follow-up office hours weeks after the session | Outcome | Follow-up is the strongest predictor that usage sticks |
| Before-and-after time baseline on real tasks | Outcome | Gives you your own proof, not a vendor benchmark |
| Attendance or completion certificate | Vanity | Records that people sat down, nothing else |
| Generic “future of AI” keynote | Vanity | Motivation fades by the next deal cycle |
| Vendor-branded generic prompt PDF | Vanity | Not written for your documents; rarely opened twice |
| Tool-landscape comparison chart | Vanity | Interesting, but not a capability your team gains |
If most of the price sits in the right column, you are paying for theater. Ask the vendor to re-price around the left column and watch how they respond.
Red flags hiding in a CRE training proposal
A few patterns should slow you down. None is automatically disqualifying, but two or three together usually mean the program will not stick.
“Contact us for pricing,” then a number that scales to your firm’s size. Mature vendors explain what drives cost (team size, session count, whether follow-up is included) and quote a range. A price that seems tuned to how prosperous your office looks is a sign the vendor is anchoring on you, not on the work.
Generic examples instead of your documents. If the sample exercises use a made-up SaaS company or a stock marketing brief rather than a lease, an offering memorandum, or an LOI, the training teaches prompting in the abstract and leaves your team to bridge the gap to real work alone. That bridge is where most people give up. Insist that drills run on your document types, redacted where needed.
No verification discipline anywhere in the outline. Language models invent comps, cap rates, square footage, and clause numbers with total confidence. A program that does not build checking-against-source into every exercise is teaching your team to trust output they should be auditing. In the workshops we run, the verification habit is the part senior brokers respect most: it answers the “my name goes on this” objection directly.
Silence on confidential data. A proposal that never mentions how deal data is handled during and after training has not thought about your business. More on that below.
Scope creep dressed as generosity. A “training” proposal that quietly bundles a tool-stack audit, custom-GPT builds, or a written AI-policy document is mixing two engagements. A fluency workshop is prompting applied to your real work; audits, custom builds, and policy drafting are separate projects with their own timelines and prices, and folding them into a training line item inflates the number while diluting the training. Whether your firm even needs an outside trainer is worth settling first, a call covered in who should upskill your brokerage.
No follow-up and no references. One-and-done sessions look cheaper and produce the weakest adoption. A vendor who offers logos but not a phone number for a principal at a comparable firm is asking you to trust a slide; ask for the reference.
What a right-sized scope looks like for a small firm
A proposal built for a 10-person brokerage should look nothing like one built for a national franchise. The right scope for a 4–20 person shop is narrow on purpose.
The content is LLM fluency on your actual work: prompting ChatGPT, Claude, Gemini, or Microsoft Copilot to draft LOIs, summarize leases, turn comps and call notes into market write-ups, and handle routine email in your firm’s voice. That is the whole substance. A proposal that also promises to build you custom tools, audit your software stack, or write your AI policy is describing separate projects to price separately if you want them at all.
The format fits a small firm’s reality. Under about eight people, everyone in the room together, because watching a respected colleague produce real work fast is what converts skeptics. Larger than that, a pilot of three or four for two weeks, then expansion using their wins. Either way the sessions run on live deals, not tutorials, and end with a shared library your firm owns. This is the small-firm advantage the manifesto for how 4–20 person shops out-operate institutional giants argues throughout: you can retrain your entire company in a quarter, a thing no national brokerage can say.
The measurement is light enough that it happens: a day-one time audit on a handful of recurring tasks, repeated at the end, plus a weekly count of who is using the tool on real work. A proposal that promises a dozen dashboards for a ten-person firm has misjudged the client. Before comparing two scopes on price, make sure they are scoped to the same depth; the workshop checklist of things to demand before you book is a useful side-by-side.
How the proposal should treat your confidential deal data
Your firm handles rent rolls, tenant financials, and deal terms under NDA, and a training proposal that ignores that has skipped the question that keeps principals up at night. Three things should appear in the document.
First, a recommendation to train on business-tier accounts rather than personal free ones, since the major providers state that inputs on their business plans are not used for model training by default (verify the specific plan’s terms at purchase). Second, a classification habit taught during the sessions: what can be pasted as-is (public listing details, your own draft language) versus what gets anonymized or stays out (anything under NDA or containing tenant personal information). Third, naming one person who owns the “can this go in the tool?” question.
Notice what this is not: a forty-page data processing agreement or a governance framework written by outside counsel. A small firm without an IT department needs three usable rules taught in an afternoon, and a proposal that offers a heavyweight policy document as a training deliverable has, again, bundled a separate project into the workshop.
Reading the price
Market pricing gives you a sanity band. Hands-on AI team workshops generally run roughly $2K–15K depending on length, depth, team size, and whether follow-up is included. Multi-session programs with office hours cost more up front and produce meaningfully better adoption, usually the better buy for a firm that wants the training to stick.
An unusually cheap quote is worth decoding rather than celebrating. Low per-seat pricing usually means a public seminar seat or a self-paced course, which trains an individual without changing how your firm operates or running on your own documents. A high quote usually means build work has been folded in. Either way, know which you are buying. The guide on what AI training for a CRE team costs in 2026 walks through the ranges by format.
The honest comparison the proposal rarely names is the billable hours your principals would spend designing the program themselves, plus the deals lost to the firm across town that got fluent first. Read that way, a right-sized proposal on your own documents, with follow-up and a library you keep, is not an expense line. It is the cheapest capacity your firm will buy this year.
FAQ
What should an AI training proposal include?
Six parts: objectives tied to your business, learning outcomes stated as capabilities (not topics), specific deliverables, a format and agenda, a measurement plan, and a transparent price with its cost drivers. For a small CRE firm the two that matter most are the outcome statements (“a broker can draft a first-pass LOI in fifteen minutes”) and the measurement plan, because those are what separate a real program from a paid seminar.
How do I know if an AI training proposal is worth the price?
Sort the deliverables into two columns: things your team will keep using (a prompt library on your documents, role drills, a verification checklist, follow-up office hours) and things that only look good (certificates, a generic keynote, a branded prompt PDF). If most of the price sits in the first column, it is worth it; if it sits in the second, you are paying for an event. Ask the vendor to re-price around the outcome column and judge by how they respond.
What are the red flags in a CRE training proposal?
Opaque pricing tuned to your firm’s size, generic examples instead of your own leases and LOIs, no verification discipline in the outline, silence on confidential data, build or policy work dressed as training, no follow-up, and references offered as logos rather than a phone number. One is a question to ask. Three together usually mean the program will not stick.
How much should AI training for a small real estate team cost?
Hands-on team workshops generally run about $2K–15K in the current market, depending on length, depth, team size, and whether follow-up is included. Programs with office hours weeks later cost more and produce better adoption. A per-seat price in the low hundreds usually signals a public seminar or self-paced course rather than training on your firm’s work. Scope drives price, so an assessment call gives a truer number than a rate card.
Should the training use our own documents or generic examples?
Your own documents, redacted where needed. Prompting taught on a made-up company or a stock brief leaves your team to bridge the gap to real leases, LOIs, and offering memoranda alone, and that gap is where adoption dies. Training on your actual document types builds the habit where the work happens and produces a prompt library you can reuse the next day.
How should the proposal handle our confidential deal data?
It should recommend business-tier accounts (where inputs are not used for training by default, verified at purchase), teach a simple classify-before-you-paste rule for NDA material, and name one person who owns the “can this go in the tool?” question. What it should not do is deliver a forty-page governance document as a training output. A firm without an IT department needs three usable rules taught in an afternoon, not a policy binder.
What if the proposal bundles a tool audit or custom-GPT build?
Treat those as separate engagements, because they are. A fluency workshop is prompting applied to your real work. A software-stack audit, a set of custom-GPT builds, or a written AI policy are distinct projects with their own scope and price. Bundling them into a training line item tends to raise the total while diluting the actual training. Buy them separately if you decide you want them.
How do I measure whether the training worked?
Two numbers on one page. First, task time on a fixed set of recurring jobs, measured before training and again a month later. Second, weekly active use: how many people used the tool on real work this week, which should trend toward everyone. If usage is not near the whole team a month out, the program stalled, and the fix is usually social buy-in, not more training.
Key takeaways
- A training proposal is priced by whether usage sticks at 90 days, not by hours in the room; read it for outcomes, not agenda.
- Sort every deliverable into outcome line items (a prompt library on your documents, role drills, a verification checklist, follow-up, a re-measured baseline) versus vanity deliverables (certificates, keynotes, generic prompt PDFs), and check where the price sits.
- Watch for the CRE-specific red flags: generic examples instead of your leases and LOIs, no verification discipline, silence on confidential data, and build or policy work bundled into a training line item.
- A right-sized scope for a 4–20 person firm is narrow on purpose: LLM fluency on your real work, the firm in the room, a library you keep, and light measurement.
- The proposal should handle deal data with three usable rules (business-tier accounts, classify-before-paste, a named owner), not a governance binder.
- Market workshops run roughly $2K–15K; a suspiciously cheap quote usually buys a seminar seat, and a high one usually folds in build work.
If you want a right-sized scope and an honest number for your own firm rather than a rate card, the fastest first step is a read on where your team stands. Book your free AI-readiness assessment →
Arthur Wandzel
