One hub, a spoke for every team — built on a database M2G owns, and around Claude, the LLM your own research already chose.
A $2B+ placemaking platform that acquires, develops, leases and operates distinctive mixed-use, hospitality, retail and industrial places — and runs the signature events and SOAR programs that make them matter.
Your AI Champion team surveyed the company. The finding was clean: across departments, the work that needs help is the same three jobs. You even named the framework.
Finding the right file fast — "parse every Stockyards deck from 2023–2025 to find when we asked Majestic to approve a 3rd-party expense."
Abstracting long documents — leases, JV agreements, loan docs, OMs — into the specific answers a decision needs.
First-pass writing that pulls from existing material — RFP responses, memos, brand decks, captions, bios.
Champions in place: Rogelio (Accounting) · Jacob (Development) · Lauren (Asset Mgmt) · Will (Finance) · JP (Ops) · Annie (Marketing).
When the Champion team tried to wire AI into real workflows, the same obstacle surfaced in every department's notes. It isn't the AI — it's where the data lives.
The daily Outlook → SharePoint → Yardi → SharePoint swivel-chair. Rogelio's words: frustrating, and entirely manual.
Your own LLM memo recommended Claude as M2G's default. The deciding factor was real-estate fit — and it points straight at the Yardi problem.
"We recommend adopting Claude as the organization's default large language model in the near term" — comparable price, broader M2G-relevant use cases.
Yardi's Virtuoso Connector launched with Claude as the first and only officially supported AI — query portfolio, financials, work orders in natural language.
Public Benefit Corp, doesn't train on your data by default, respects M365 permissions — the trust posture a fiduciary firm needs.
The model question is settled. M2G OS is built around Claude.
The Champion team assessed the field hard — Glean, Copilot, Egnyte, Monday, ChatGPT. The frontrunner was honest about its fit.
The best enterprise tool reviewed. Your own conclusion: it is "more horsepower than M2G currently needs" — ~$65K/year, a $20K+ crawl, a custom Yardi integration, and a big data migration to land it.
"Agentic AI is still a ways away… AI still makes mistakes that can have material impact." Your verdict: be disciplined about ROI, don't give in to hype. These tools cost a junior employee.
Read your own findings together and the answer is inevitable. Same three needs everywhere → so build one hub. Data trapped in a closed Yardi → so own the ledger. Claude already chosen → so build around it. Glean too generic → so build the one tool that fits M2G exactly.
A single system of record M2G owns — every division, one live view of the truth, queryable in plain English.
Eight focused workspaces on top of that hub — each shaped to a division's real work, not a vendor's template.
Own the ledger and the swivel-chair dissolves. The 1–2 agents you set as a goal become the whole operating system.
No open API, no clean schema you control. AI can read copies at best — it can't act on the source.
Outlook → SharePoint → Yardi → SharePoint, by hand, every invoice — the daily friction the survey flagged firm-wide.
Trapped data can't be joined with market signals or asked a plain-English question. Own the ledger and all three dissolve.
The all-inclusive system doesn't rip out how M2G works today. It replaces what was never meant to be your system of record — and absorbs the stack you were about to rent.
The same three jobs — retrieve, synthesize, generate — delivered inside each division's real workflow, all reading and writing to one owned hub.
Deal pipeline, OM intake, comps, the master comp database.
Project status, schedules, RFP drafting, doc abstraction.
Leases, rent roll, renewal & termination flags, performance.
Investor CRM, memos, reporting, Marketing-Rule-aware drafts.
Run fundraising at your own assets — RSVPs, check-in, SOAR.
Brand-consistent decks, captions, bios, monitoring.
AP coding, statements, the bridge tab, board-ready packets.
Scheduling, handbook & compliance Q&A, company comms.
Eight spokes, one ledger. The more of M2G that runs on it, the more it becomes how M2G operates.
Every example below is lifted from your own survey and testing rubric — mapped to the OS, with the time it gives back. Claude drafts and codes; a human approves.
Signature events at Mule Alley, ALCO and the Stockyards aren't a side project — they're how M2G builds the relationships that close deals and raise funds. $2M+ raised through them. No off-the-shelf platform serves this; the OS makes it first-class.
Guest lists, invitations, RSVPs and live check-in for a fundraiser at your own asset — one workflow from save-the-date to the night-of, instead of five spreadsheets and an inbox.
Every guest, donor, LP and prospect in one relationship graph — Claude drafts the timely follow-up, flags the warm intro, and surfaces who you met where. The SOAR mission and the deal pipeline, on the same ledger.
The hours-saved is real — but the durable prize is the asset you build underneath it. Proprietary operational data you own, plus the layer your own roadmap already wants but no vendor sells.
Every lease, deal, event and decision on a ledger M2G controls — not rented copies in a closed system you'll pay for forever.
"Blend external market signals with internal information about asset classes, submarkets, and expirations." Your roadmap's own words — and no vendor offers it. Owning the data is the only way to get it.
Glean and Copilot give every firm the same release. An OS shaped to M2G's data is an edge competitors can't buy off a shelf.
Your own estimate of a custom RE-Intelligence build: $350K+ on the low end. This is that capability — built into the operating system, not bolted on.
Impact isn't a promise — it's a before-and-after on a metric M2G's leadership already watches.
G&A as a % of AUM — or AUM (SF) managed per employee. The thesis in one number: run more portfolio with the same team.
Month-end close in days, or hours/week of manual admin returned — unambiguously OS-attributable. Your own framing: "Rogelio saves ~5 hrs/week."
Baseline confirmed with M2G — the pilot proves it in 60 days. These brackets are filled together, on a metric your leadership already owns.
The OS honors your AI Policy v3.1 as written. Claude does the first pass; a qualified person approves before anything is relied on or sent. No autonomy theater.
Straight from your policy: "Picture the AI as an eager, highly intelligent intern… AI is a writing partner — but you are the writer." The OS is disciplined on ROI and clear-eyed about what AI can't do. Every output is documented, reviewable, and human-owned.
M2G OS is live today — a clickable operating system seeded with an M2G-shaped portfolio, with Claude really in the loop. We won't put a percentage on a company we don't fully know yet. Here's what already works.
This is the exact question Kristie raised in your testing rubric — "buy a property with ten tenants, which one has a termination option?"
Because the OS owns the ledger, Claude answers it in plain English, cites the source leases, and lets you act — open the asset, draft the notice — without leaving the thread.
No big-bang cutover. We deliver your 2026 goal of 1–2 reliable agents first, prove the ROI in your own language, then widen — and only strangler-migrate off Yardi once each function is proven on owned data.
Rogelio's AP flow + Lauren's lease abstraction live on the OS — the 1–2 agents your roadmap already targets.
Checkpoint: ~5 + 2.5 hrs/wk back, measuredFinance, Investments, Marketing, Events & Giving — each spoke added on proof, on one owned hub.
Checkpoint: time saved per divisionFunction by function, the OS owns the ledger — reporting, then AP, then leases. Yardi runs until each piece is solid.
Checkpoint: owned vs. rented, function by functionStay on Yardi the whole way. Switch each function off only when M2G OS has earned it.