A commercial real estate firm of 4 to 20 people can adopt AI faster, apply it deeper, and compound the gains longer than CBRE, JLL, or any institutional giant — because the giants’ scale is a liability in this transition and the small shop’s intimacy with its own workflows is the single most valuable AI asset in the industry. That claim sounds like an underdog fantasy until you look at what the institutional adoption numbers say. Then it starts to look like arithmetic.
This is a manifesto for the owners, principals, managing brokers, and ops directors running small US commercial real estate firms. It states a thesis, defends it with the giants’ own survey data, and commits to six operating principles that turn a lean shop’s structural advantages into a durable operating edge. Each commitment expands into a full playbook elsewhere on this site. This page is the spine.
Why a Manifesto, and Why Now
The commercial real estate industry has spent three years telling itself that AI is an institutional game. The evidence says the institutions are losing it.
Start with the most uncomfortable number in proptech. JLL’s 2025 Global Real Estate Technology Survey found that 88% of CRE investors and owners had launched AI pilots, yet only 5% reported achieving most of their program goals (Propmodo’s analysis of the survey is the best public read on this gap). More than 60% of respondents described themselves as unprepared for AI strategically, organizationally, or technically. These are organizations with chief digital officers, innovation budgets, and vendor relationships a 10-person shop will never have. They are drowning in pilots and starving for results.
Deloitte’s 2026 Commercial Real Estate Outlook, which surveyed more than 850 C-suite executives across 13 countries, tells the same story from a different angle. Only 7% of respondents reported transformative impact from AI. Another 27% said they were still in an early or experimental phase, and 21% reported mixed results. Deloitte’s own conclusion is the one that matters for this manifesto: organizational readiness and change management may count for more than technology maturity in determining who benefits.
Read that conclusion again from the perspective of a 12-person brokerage. Organizational readiness is the one dimension where you can beat an institution by default. You have no steering committee, no procurement cycle, no six-month security review, no regional rollout plan. You have a principal who can decide on Monday and a team that can be working differently by Friday.
The prize for getting this right is not trivial. McKinsey estimates generative AI could create $110 to $180 billion in annual value for the global real estate sector. Practitioner authorities have stopped debating whether the technology works: Adventures in CRE’s Summer 2026 tools guide opens by declaring regular AI use “the new baseline” for CRE professionals. The open question is not whether AI creates value in commercial real estate. It is who captures that value: the 300,000-plus real estate firms the National Association of REALTORS® tracks, most of them small independents, or the handful of giants everyone assumes will win.
We think the small firms can win, and this manifesto is the argument for how.
The Small-Firm Advantage: Four Structural Edges
The case does not rest on optimism. It rests on four structural properties of a 4-20-person firm that no institution can replicate at scale.
1. Decision velocity. An institutional AI initiative passes through legal, IT security, procurement, and a pilot committee before a single broker touches a tool. In a small shop, the principal is the committee. When the JLL data shows 88% of institutions stuck at the pilot stage, it is describing a governance problem you do not have. The firm that can decide, deploy, and revise in a two-week loop will out-learn the firm that revises quarterly, regardless of headcount.
2. Workflow intimacy. In a 10-person firm, the person choosing the AI workflow is the person who runs the workflow. The principal knows exactly which lease clauses trip up abstraction, which lenders want which format, which client always calls before the email lands. Institutions pay consultants seven figures to rediscover this knowledge through interviews. You have it for free, and AI systems configured against real workflow knowledge outperform systems configured against a process map every time.
3. Full-stack visibility. The same five people touch prospecting, underwriting, closing, and management. That means a small firm can wire AI across the entire deal lifecycle without an integration committee, because the “integration” is the same person using two tools in one afternoon. The compounding is real: a lease abstract feeds the rent roll, the rent roll feeds the investor update, the update feeds the next raise.
4. Per-seat economics that favor you. The tools that matter most cost $20 to $60 per seat per month for business-grade plans of ChatGPT, Claude, Gemini, or Microsoft Copilot. At 10 seats, a serious AI capability costs less per year than one month of an institutional software contract. The giants cannot buy an edge here, because the entry price is too low to be exclusionary. When capability is cheap, execution is the moat, and execution is a small-team sport.
None of this makes small firms automatically win. It makes them able to win, which is a different thing. Ability converts to advantage only through commitments, and commitments are what a manifesto is for.
The Six Commitments
Each commitment below is a one-line operating principle, the reasoning behind it, and the link to the playbook that turns it into practice. Together they cover the full surface of AI for commercial real estate as a small firm experiences it: people, documents, deals, communications, back office, and buying decisions.
1. Fluency Before Software
The commitment: every person in the firm can use ChatGPT, Claude, or Gemini competently on real CRE work before the firm buys a single proptech subscription.
The most common small-firm failure mode we see is inverted sequencing: buy a tool, announce it, watch it die. The JLL and Deloitte numbers above are mostly a record of this failure at institutional scale. Software purchased before a team is fluent becomes shelfware, because nobody can tell whether the tool is bad or the usage is.
Fluency means something specific in a CRE context. It means a broker can draft an LOI framework in ChatGPT and know which terms to verify by hand. It means a property manager can feed a lease into Claude and get a summary they know how to check. It means an analyst can turn messy notes into a market write-up in minutes and an associate can clear an inbox backlog without a template binder.
Fluency is also the cheapest item on the entire AI menu. Market rates for hands-on AI team training run $2,000 to $15,000 for a small firm, against $25,000 to $150,000 for custom automation work. Sequencing the cheap, high-yield step first is not caution. It is good underwriting.
The full 90-day program, including who to train first and how to measure fluency, is in The CRE AI Training Playbook.
2. The Lease Stack Is an Asset, Not a Filing Problem
The commitment: every lease, amendment, estoppel, and rent roll in the firm becomes structured, queryable data.
Small CRE firms sit on decades of documents that they treat as storage overhead. Those PDFs are the raw material for nearly every downstream AI gain: faster due diligence, cleaner rent rolls, instant answers to “what does the co-tenancy clause say,” abstraction that used to take an analyst 2 to 3 hours per lease.
Document intelligence is where AI in commercial real estate stopped being a demo and started being a line item. Purpose-built abstraction vendors like Prophia and LeaseLens exist because the institutional market already pays for this. A small firm can get most of the same outcome with a business-grade AI assistant, a consistent extraction prompt, and a review discipline — or with a targeted tool once volume justifies it.
The trap to refuse: doing document work in a consumer-grade AI account with default data settings. Confidential deal documents go into business or enterprise plans with training turned off, or they do not go in at all. That rule costs nothing and prevents the one failure that can do lasting damage to a small firm’s reputation.
The extraction workflows, tool tiers, and accuracy checks are in The CRE Document Intelligence Playbook.
3. Screen Ten Deals in the Time One Used to Take
The commitment: AI multiplies the number of deals the firm can evaluate, and humans keep full custody of the judgment call.
Deal flow is a numbers game that small firms have historically lost on capacity. An acquisitions shop with two analysts can only underwrite so many OMs a week, so the funnel narrows early and good deals die unread. AI changes the capacity math: first-pass screening, comp pulls, market write-ups, and sensitivity setups compress from hours to minutes, which means the same two analysts can hold a funnel that used to require six.
The commitment has a second clause, and it is load-bearing. AI output on deals is analyst work, not calculator output. It gets reviewed the way you would review a first-year analyst’s model: trust the format, verify the numbers, interrogate the assumptions. Firms that skip the review step do not have an AI capability. They have a liability with a subscription fee.
We are opinionated on where to start: screening, not final underwriting. The failure cost of a bad screen is a wasted hour; the failure cost of a bad underwrite is a bad deal. Build the habit where errors are cheap.
The screening workflows, prompt structures, and review gates are in The CRE Deal Analysis Playbook.
4. Nobody Waits on Your Inbox
The commitment: every inbound inquiry, tenant request, and client update gets a same-day, quality response, at 10-person scale, without hiring.
Commercial real estate is a response-time business wearing a relationship-business costume. The broker who answers a broker-blast question in twenty minutes wins mindshare over the one who answers Thursday. The property manager who acknowledges a tenant issue within the hour prevents the escalation that eats Friday. Small firms lose here for one reason: the people who write the best responses are the busiest people in the building.
AI collapses that constraint. Drafting is the bottleneck, and drafting is what these tools do best: listing descriptions, follow-up sequences, tenant notices, investor updates, CRM notes that get written at all. The person still approves and sends. The 40 minutes of composing becomes 4 minutes of editing, and the response-time advantage compounds into the thing small firms are selling in the first place: the feeling that the client is your only client.
The inbox, CRM hygiene, and listing-marketing systems are in The CRE Communications Playbook.
5. The Back Office Closes Itself
The commitment: rent rolls, CAM reconciliations, and investor reporting stop consuming principal-hours.
Back-office work is where small CRE firms quietly bleed their most expensive time. The principal who spends the first week of every quarter assembling investor reports is doing $40-an-hour work at a $400-an-hour opportunity cost. Nobody started a real estate firm to reconcile CAM charges.
This is also the domain where the buy-versus-build spectrum is widest. Platforms like Yardi, AppFolio, and Buildium keep adding AI features to the property-management stack, and a firm already living in one of them should exhaust those features first. The gap is the workflow glue between systems: the rent roll that arrives as a scanned PDF, the investor letter that summarizes three spreadsheets, the reconciliation that requires judgment about which charges are contestable. That glue is exactly what modern AI handles well and what small firms have always paid for in overtime.
Back-office automation is the least glamorous commitment on this list and, for management-heavy firms, frequently the highest-ROI one. Boring workflows are the best automation candidates precisely because nobody defends them emotionally.
The rent-roll, CAM, and reporting automations are in The CRE Back-Office Automation Playbook.
6. Buy Boring, Build Advantage
The commitment: the firm buys off-the-shelf software for commodity workflows and reserves custom builds for the workflows that win it business.
Small CRE firms have been burned by proptech subscriptions before, and the skepticism is earned. The honest decision rule has two halves. Buy when the workflow is generic across the industry: market data belongs to CoStar, LoopNet, or Crexi; property accounting belongs to Yardi or AppFolio; document generation belongs to Buildout or similar. Nobody ever won a listing because of a custom-built accounting system.
Build (or commission) when the workflow embodies the firm’s actual edge: a proprietary screening model tuned to your buy-box, an intake pipeline shaped around your lender relationships, a reporting format your investors chose you for. Market pricing for custom automation runs $25,000 to $150,000 depending on scope, which means a build has to clear a real hurdle rate. Most workflows do not clear it. The two or three that do are worth more than every subscription on the invoice.
The evaluation framework, vendor landscape, and build-decision economics are in The CRE AI Buy-vs-Build Playbook.
The Confidentiality Discipline
One discipline sits underneath all six commitments, and it is non-negotiable for a firm that handles confidential deal data with no IT department to backstop it.
- Use business or enterprise plans of ChatGPT, Claude, Gemini, or Microsoft Copilot, with data-training settings reviewed and turned off. Consumer accounts with default settings are not for deal documents.
- Decide what never goes into any AI system regardless of plan: anything under NDA terms that prohibit third-party processing, and anything you would not put in an email to an outside vendor.
- Write the rules on one page. A small firm does not need a 40-page AI governance framework; it needs one page the whole team has read, covering approved tools, prohibited inputs, and who to ask when unsure.
- The principal goes first and visibly. Policy adoption in a small firm is behavioral, not documentary. When the owner reviews an AI-drafted lease summary in the Monday meeting, the policy enforces itself.
Security anxiety is the most common reason small CRE firms delay AI adoption entirely, and it is the wrong conclusion drawn from a right instinct. The instinct (our deal data is sensitive) is correct. The conclusion (therefore avoid AI) hands the operating edge to competitors while providing no actual protection, since unsanctioned personal-account usage fills the vacuum anyway. The one-page discipline above is what the instinct should produce instead.
What a Small Firm Should Do With This
A manifesto that does not change your Monday is a poster. Here is the sequence we would run in any 4-20-person firm, in order, with the reasoning attached.
- Write the one-page confidentiality policy first. It takes an afternoon and unblocks everything else. Approved tools, prohibited inputs, escalation contact. Done.
- Get the team fluent on real work (Commitment 1). Business-grade accounts, actual firm documents (per the policy), weekly practice against live tasks: LOIs, lease summaries, market write-ups, email. Ninety days is enough to make this stick.
- Pick one workflow from Commitments 2 through 5 and go deep. Document-heavy firms usually start with the lease stack; acquisitions shops with deal screening; brokerage-forward shops with communications; management-heavy firms with the back office. One workflow, measured honestly, beats five pilots every time. The institutional data is the proof.
- Run the buy-vs-build decision only after steps 1-3. A fluent team with one working AI workflow evaluates vendors from strength; a non-fluent team evaluates them from hope. The buy-vs-build framework exists for exactly this moment.
- Re-read the JLL number quarterly. 88% piloting, 5% succeeding. Every quarter, ask whether you are operating like the 5% (few workflows, deep adoption, measured results) or the 88% (many pilots, shallow usage, no scoreboard).
If you want an outside read on where your firm sits before you start, we run a free AI-readiness assessment for small CRE firms: a structured look at your workflows, data, and team, with an honest sequencing recommendation. Book an assessment call if that would be useful. The playbooks above are free either way, and most firms can run this manifesto without us.
Frequently Asked Questions
Can a 10-person CRE firm really out-operate CBRE or JLL with AI?
On adoption speed and workflow depth, yes; on data scale and capital, no, and the manifesto’s argument is that the first pair matters more right now. JLL’s own 2025 technology survey shows 88% of institutional players stuck in pilots with only 5% hitting their goals, largely on change-management grounds. A small firm with no committees can go from decision to deployed workflow in weeks and iterate faster than any institution. You will not out-research CoStar or out-spend CBRE. You can out-execute both inside your own niche.
Is it safe to put confidential deal documents into ChatGPT or Claude?
It is safe under specific conditions: a business or enterprise plan, data-training settings reviewed and disabled, and a firm-level rule about what never goes in (NDA-restricted material, anything you would not email an outside vendor). It is not safe in consumer accounts with default settings. The practical standard for a small firm is a one-page policy plus business-grade accounts, which together cost less than most firms spend on coffee and remove the largest genuine risk in the entire adoption process.
Why do most commercial real estate AI pilots fail?
Sequencing and shallowness, not technology. Deloitte’s 2026 outlook found organizational readiness and change management matter as much as the tools, and the JLL data shows institutions running many pilots with little depth in any. The typical failure: buy software before the team is fluent, run five simultaneous experiments, measure nothing, conclude AI is overhyped. The fix is the reverse: train first, pick one workflow, adopt deeply, keep score.
What should a small CRE firm automate first?
Whichever high-volume workflow your firm feels most: lease abstraction and document work for management and acquisitions shops, deal screening for investment teams, inbox and listing marketing for brokerages, rent-roll and reporting work for back-office-heavy firms. The selection matters less than the discipline: one workflow, adopted deeply, measured honestly, before the next one starts. Fluency training comes before all of them.
Should we buy proptech subscriptions or build custom AI?
Buy for commodity workflows, build only for workflows that constitute your firm’s actual competitive edge. Market data, accounting, and standard document generation are solved categories: CoStar, Crexi, Yardi, AppFolio, and Buildout exist for a reason. Custom automation (market range $25,000 to $150,000) makes sense for the two or three workflows where your firm genuinely differs, such as a proprietary screening model for your specific buy-box. Most small firms should buy more than they build, and build later than they think.
How much does AI adoption cost a small CRE firm?
The entry tier is modest: business-grade AI assistant plans run roughly $20 to $60 per seat per month, so a 10-person firm operates for a few hundred dollars monthly. Team training runs $2,000 to $15,000 at market rates. Custom automation, when a workflow justifies it, runs $25,000 to $150,000. The sequencing in this manifesto is deliberately cheapest-first: policy (free), fluency (thousands), one deep workflow (subscription cost plus attention), custom builds (only after the hurdle rate clears).
Do we need an AI policy before the team starts using these tools?
You need one, and a single page is enough. The policy names approved tools and account tiers, lists what may never be entered into any AI system, and designates one person to ask when unsure. Waiting for a comprehensive governance framework is how institutions ended up in the 88%-piloting-5%-succeeding trap; skipping policy entirely is how confidential deal terms end up in consumer chatbot accounts. The one-page version captures nearly all the protection at almost none of the cost.
Can we trust AI output for underwriting and lease abstraction?
Treat it as first-year analyst work: strong on format and speed, requiring review on substance. AI abstraction of a lease is a draft to verify against source clauses, not a final record; an AI-assisted screen is a reason to look closer, not a reason to close. Firms that build the review habit get the capacity gains without the liability. Firms that treat AI output as calculator output eventually publish an error they did not catch, and in a small firm your name is on it.
Do we need to clean up our data before AI is useful?
Cleanup is not a prerequisite, and waiting for clean data is a stalling pattern. Modern AI tools are unusually good at messy inputs: scanned rent rolls, inconsistent lease formats, decades of PDF filing chaos. Institutional adoption advice emphasizes data readiness because institutions integrate across dozens of systems; a 10-person firm can start with the documents it has today. Cleanup happens as a byproduct of the document-intelligence work, not as a prerequisite for it.
Closing: Signed, With a Date on It
Every claim in this manifesto has a shelf life, so we are explicit about the date. As of mid-2026: institutional AI success rates in CRE remain in the single digits by the institutions’ own surveys, business-grade AI assistants cost less per seat than a parking space, purpose-built CRE tools exist for every major workflow, and no ranking, survey, or vendor roadmap shows the giants converting their scale into adoption speed. The window in which a small firm can build an operating edge is open now. It will not stay open at this width.
The commitments, one last time. Fluency before software. The lease stack is an asset. Screen ten deals in the time one used to take. Nobody waits on your inbox. The back office closes itself. Buy boring, build advantage. Underneath all six: the one-page confidentiality discipline, and a principal who goes first.
The giants have the committees. You have the mornings. Use them.
— Arthur Wandzel, CEO, SFAI Labs
Arthur Wandzel
