An AI consultant, done right for a small commercial real estate firm, does three things: finds the workflows where AI actually saves hours, decides for you whether to buy an off-the-shelf tool or build something custom, and then either gets your team fluent or ships the automation. That is the whole job. Not reselling software with a markup, not writing code you will never understand, not a slide deck about the future of proptech. The word actually in the question is doing a lot of work, because the field is crowded with people who charge for AI advice that amounts to a ChatGPT login and a pep talk. A 12-person brokerage does not need a data-transformation program; it needs someone to answer a short list of concrete questions and leave behind something the team can use on Monday. This is what a real engagement contains, phase by phase, so you can price and judge any proposal that lands on your desk.
Before the engagement types, one bit of orientation. The single most consequential thing an AI consultant decides for a small firm is whether to buy or build, and the full version of that decision (when an off-the-shelf proptech subscription beats a custom project, and when it does not) lives in our buy-vs-build playbook for CRE. The broader case for why a lean shop can out-operate an institution with the right handful of tools sits in the small-firm CRE manifesto. This piece is the practical layer under both: what you are buying when you hire someone to help.
The Core Job in One Sentence
An AI consultant translates “we should probably be using AI” into a short, ordered list of decisions a non-technical owner can make. That translation is the entire value, and it is harder than it sounds, because the gap between AI activity and AI results is enormous at small firms. NAR’s 2025 Technology Survey found 68% of Realtors now use AI tools, yet only 17% report a significant positive impact on their business. A good consultant exists to close that 51-point gap for one specific firm.
The job breaks into three moves. First, diagnosis: which of your recurring, hour-eating workflows (lease abstraction, LOI drafting, market write-ups, inbox triage) a general-purpose assistant can genuinely help with, and which it cannot. Second, the buy-vs-build call: whether the answer is a tool you can subscribe to this week, a training program so your team does it themselves, or a custom automation worth commissioning. Third, delivery: getting the team fluent or the automation shipped, then handing it off so the firm owns it.
Everything else a consultant might sell you is either a subset of these three moves or a distraction from them. If a proposal cannot be mapped back to diagnose, decide, deliver, be suspicious.
The Four Engagement Types
Most AI consulting for small CRE firms takes one of four shapes. They differ in cost, duration, and how much the firm changes at the end. A trustworthy consultant will tell you which one you need, and it is usually not the most expensive.
| Engagement | What it is | Typical duration | What changes |
|---|---|---|---|
| AI-readiness assessment | A diagnostic working session that maps your highest-cost workflows and hands back a prioritized recommendation | Days | You know exactly where AI helps and where it does not |
| LLM-fluency training | Hands-on workshops that make your team genuinely capable with tools like ChatGPT, Claude, Gemini, or Microsoft Copilot on your own documents | Weeks | The team can produce usable lease summaries, LOIs, and write-ups themselves |
| Custom automation build | A purpose-built system that runs a high-volume workflow (extracting rent-roll data, screening deals) without hand-prompting | Weeks to months | One workflow moves from manual to automated |
| Ongoing advisory | A retained relationship for vendor selection, new-workflow reviews, and keeping current as tools change | Continuous | The firm has a standing answer for “should we use this?” |
The important pattern: the first two are cheap and solve most small-firm problems, and the last two are only worth it once you have done the first two. A firm that jumps straight to a custom build without a diagnosis is buying a solution to a problem it has not defined. That is the most common way engagements waste money, and a consultant worth hiring will stop you from doing it.
What a Typical Engagement Looks Like, Phase by Phase
Whatever the type, a well-run engagement moves through four phases. Knowing them lets you read any proposal and spot where a firm is trying to skip the unglamorous parts.
Phase 1: Discovery
The consultant learns how your firm works, not how a generic brokerage works. Where do the hours go? Which tasks are repetitive and text-heavy? What data is confidential and what is public? This is the phase lazy providers skip, and skipping it is why so much AI advice is generic. Good discovery produces a ranked list of candidate workflows and a clear read on your constraints, including the fact that you have no IT department and run on Excel, Outlook, and PDFs.
Phase 2: Recommendation
The consultant returns with the buy-vs-build call and a sequence. This is where a real one earns their fee by narrowing, not expanding: three workflows to start with, a recommendation to buy a specific category of tool for one and train the team for another, and an honest “leave this one manual for now.” The recommendation should be legible to a non-technical owner and tied to a rough return, which is exactly the kind of simple model our ROI-on-automation guide lays out so you can sanity-check the math yourself.
Phase 3: Delivery
The work happens. For training, that means hands-on sessions on your real leases and LOIs, not vendor sample files, because fluency only transfers when people practice on the documents they handle daily. For a build, it means shipping the smallest version that proves the workflow, closer to a minimum viable product than a finished platform. Our MVP explainer unpacks that distinction for owners deciding how much to build up front.
Phase 4: Handoff
The engagement ends with the firm owning the result, not depending on the consultant forever. For training, that means a documented prompt library and a named internal owner. For a build, it means documentation, access, and a plain-language runbook. If the handoff leaves you unable to operate without the consultant, the engagement created a dependency, not a capability.
What You Should Physically Get
Vague engagements produce vague deliverables. A concrete one leaves physical artifacts on your desk. Insist on knowing which of these you will get before you sign:
- A workflow map ranking your recurring tasks by hours spent and AI suitability.
- A one-page confidential-data rule stating which tools are approved and what deal data may never be pasted into them.
- A buy-vs-build recommendation naming, for each priority workflow, whether to subscribe, train, or build.
- A prompt library (the handful of prompts that reliably produce a good lease summary or first-draft LOI) if training is part of the scope.
- A simple ROI model you can update yourself as usage grows.
- A runbook and access if anything was built, so the firm can operate it without the consultant.
If a proposal cannot tell you which of these lands at the end, it is selling activity, not outcomes.
The Buy-vs-Build Honesty Test
The clearest signal that you have hired a good AI consultant rather than a disguised salesperson is this: they will often tell you not to build anything. For a large share of small-firm workflows, an off-the-shelf proptech subscription already does the job well enough, and commissioning custom software would be slower, costlier, and harder to maintain. The categories are mature: document and lease tools, deal and pipeline platforms, marketing and listing tools. A consultant who never recommends buying is one whose incentives point at billing you for a build.
The honest calculus is simple. Buy when a tool fits your process closely enough and the subscription is a fraction of a build. Build only when a workflow is high-volume, genuinely specific to how your firm operates, and no product bends to it without breaking. Most firms discover, after an honest diagnosis, that they need to buy two tools and train their team, and that the custom build they imagined is a year away at best. Firms that skipped the diagnosis and jumped straight to modernizing tend to tell a different story, the kind of “we finally stopped doing it the old way” account collected in our piece on digitization at small firms. The common thread is that the win came from a small, well-chosen change, not a sweeping rebuild.
This is not a minor point of etiquette. JLL, surveying more than 1,500 CRE decision-makers, found 88% of firms piloting AI but only 5% achieving their program goals, with the average firm running five separate pilots. That is the signature of buying access repeatedly without a diagnosis first. A consultant’s real product is the discipline to pick one or two right things instead of five wrong ones.
What It Costs
Pricing depends on which engagement type you need, and honest market ranges look like this. An AI-readiness assessment is often free or low-cost, because it is also how a good consultant qualifies whether they can help you. Structured LLM-fluency training for a small team generally runs from the low thousands to the low tens of thousands of dollars, depending on team size and how many sessions. Custom workflow automation is a different order of magnitude, typically a low-six-figure engagement depending on scope, and an off-the-shelf proptech subscription that avoids a build entirely can be a few hundred dollars a month.
The sequencing is what protects the spend. Because the assessment and training are cheap and the build is not, doing them in order means you only ever commission the expensive work once you know precisely which workflow to automate and what good output looks like. Any firm quoting you a six-figure build before a diagnosis is asking you to pay for the answer before the question has been defined.
Red Flags: What a Good Consultant Will Not Do
A trustworthy engagement is as much about what a consultant declines to do as what they deliver. Watch for these:
- Recommending a custom build before diagnosis. Nobody can specify what to build before the team has run the workflow with AI and can say what good output is.
- Reselling one vendor for everything. A consultant tied to a single platform is a channel partner, not an advisor.
- Describing an enterprise transformation to a 12-person firm. Data lakes, governance boards, and multi-quarter roadmaps are answers to problems you do not have.
- Promising a fixed percentage of time or cost saved. Honest ranges and hypotheses are fine; precise guarantees pulled from thin air are not.
- Leaving no handoff. If the plan makes you permanently dependent on the consultant, it built a dependency, not a capability.
Deloitte’s 2026 Commercial Real Estate Outlook, surveying more than 850 executives, still found 27% of firms blocked by implementation challenges tied to expertise and change resistance, and those were firms with real technology budgets. A small firm faces the same barriers with fewer resources, which is exactly why the value of a good consultant is judgment and sequencing, not scale.
Frequently Asked Questions
What does an AI consultant actually do for a small firm?
For a 4–20 person firm, an AI consultant diagnoses which of your recurring workflows AI can genuinely help with, decides whether to buy an off-the-shelf tool or build something custom, and then delivers, either by training your team to be fluent with tools like ChatGPT, Claude, Gemini, or Microsoft Copilot, or by shipping a specific automation. The output is a short list of decisions a non-technical owner can act on, plus physical deliverables like a workflow map, a data-handling rule, and a prompt library. It is not reselling software or writing code you cannot maintain.
Do I need an AI consultant or can I just use ChatGPT?
Many small firms can start with a subscription and self-teach, and a good consultant will tell you if that is your situation. The value of hiring help is speed and avoided mistakes: an outside diagnosis stops you from buying five tools you do not need, tells you which two workflows to start with, and writes the confidential-data rule before someone pastes a live rent roll into a free consumer tool. If your team is already producing usable work with AI on real documents, you may not need one. If access has not turned into results, that gap is exactly what a consultant closes.
How long does an AI consulting engagement take?
It depends on the type. A readiness assessment takes days. LLM-fluency training runs over a few weeks, because turning a first success into a durable habit takes reinforcement, not a single workshop. A custom automation build runs from a few weeks to a few months depending on scope. Ongoing advisory is continuous by design. Anyone promising firm-wide transformation from one afternoon session is selling attendance, not capability.
What is the difference between an AI consultant and a proptech vendor?
A proptech vendor sells you their product; an AI consultant helps you decide whether you need any product at all, and if so, which one. The vendor’s incentive is a subscription; the consultant’s should be your outcome. The practical tell is that a good consultant will sometimes recommend a competitor’s tool, or no tool, or training instead of software, recommendations no vendor will ever make. If a consultant only ever points you at one platform, they are functioning as a reseller.
Will a consultant make me build custom software?
A good one usually will not. For most small-firm workflows, an off-the-shelf proptech tool already does the job, and a custom build would be slower and more expensive to commission and maintain. Building is worth it only when a workflow is high-volume, specific to how your firm works, and not served by any existing product. A consultant who recommends a build before diagnosing your workflows, or who never recommends simply buying a tool, has incentives pointed at their invoice rather than your result.
How much does an AI consultant cost?
As a market orientation, a readiness assessment is often free or low-cost, LLM-fluency training for a small team generally runs from the low thousands to the low tens of thousands of dollars, and custom automation is typically a low-six-figure engagement depending on scope. An off-the-shelf subscription that avoids a build can be a few hundred dollars a month. The order matters: doing the cheap diagnosis and training first means you only pay for an expensive build once you know exactly which workflow to automate.
How do I keep confidential deal data safe during an engagement?
The consultant should produce a one-page rule early: which tools are approved, what data is fine to paste (public listings, hypotheticals), and what never leaves the firm (client names tied to financials, unexecuted terms, anything under NDA). Default to business-tier accounts rather than personal free logins, since the business tiers of the major assistants offer settings that keep your inputs out of model training. Confirm the current terms for whichever tool you approve, because these settings change. Any engagement that starts putting real deals into tools before this rule exists is doing it backwards.
What should I get at the end of an engagement?
Physical artifacts, not just advice. Expect a workflow map ranking your tasks by hours and AI suitability, a one-page confidential-data rule, a buy-vs-build recommendation for each priority workflow, a prompt library if training was in scope, a simple ROI model you can update yourself, and a runbook plus access if anything was built. If a proposal cannot name which of these you will receive, it is selling activity rather than a result you can hold.
How do I know if the engagement worked?
Measure behavior on real work, not attendance. After training, track whether the team reaches for AI on live tasks and produces output they are comfortable sending: lease summaries, LOI drafts, market write-ups. Time a fixed set of real jobs before and after. For a build, watch throughput on the automated workflow. Logins, completed courses, and a nice exit survey can all look fine while nothing about how the firm operates has changed. If the behavioral numbers move, the engagement worked.
Where to Start
The first step is the cheapest and the most revealing: a diagnosis of where your firm’s hours go and which of those workflows a trained team or a well-chosen tool would genuinely change. That is precisely what a free AI-readiness assessment produces: a working session that maps your highest-cost workflows, flags the confidential-data rules you need before anyone starts, and tells you honestly whether you need a full workshop, a lighter course, an off-the-shelf tool, or just clearer rules for what you already have. It is, in miniature, the discovery phase of any real engagement, so you can sample how a good consultant thinks before committing a dollar. Book a free AI-readiness assessment and you will leave with the first phase already done and a clear read on whether hiring further help is worth it at your firm.
Arthur Wandzel
