This was a build week. Four new tasks live (plus all the sub-tasks and skills behind those tasks), fourteen existing ones upgraded to v2, and the finishing touches on something we've been waiting months to show you. Behind the Glass The four new tasks: Analyze Multifamily Rent Roll now has a full analysis layer on top of the existing parser, so you can run occupancy trends, loss-to-lease, lease expiration exposure, and delinquency risk on any multifamily rent roll you upload. Analyze T12 Operating History takes a trailing twelve and runs variance and anomaly detection, with severity-graded flags and specific follow-up questions for your diligence team. Property Level Overview for Land Development pulls parcel, ownership, zoning, flood, wetlands, demographics, schools, and a location analysis into a single diligence package, complete with a draft utility inquiry email. Deep Parcel Research runs a tailored parcel workflow depending on whether you're screening an acquisition, doing off-market outreach, underwriting for development, or prepping a pre-meeting brief. The fourteen v2 upgrades touched some of our most-used tasks: Build Stabilized Pro Forma, Deep Location Analysis, both RUBS and stabilized property tax, Simple Lease Abstract, Loan Sizing, Zoning Memo, OM to A.CRE Value-Add Model, and several others. The library is bigger, and the existing tasks are harder to break. We also shipped ten new subtasks that run inside primary tasks and make them more powerful without you ever needing to see them. The thing I've been waiting to show you is Visual Apps. We're putting the finishing touches on them now. These are highly polished, shareable outputs that go beyond Excel files, Word documents, and static dashboards. Think beautiful Offering Memorandums, dynamic financial models, data dashboards, investment committee memo forms. This has been a missing piece in building a platform that covers the full range of work we do in CRE. The first ones go live next week. The Unlock Most of my week was spent one-on-one with Founders, building the custom tasks each of you gets as part of this group. (Friendly reminder: if you haven't shared the custom task you want us to build for you, you're missing out on one of the best perks.) Those conversations keep surfacing how much the data layer matters. Here's a live example: we're working through a challenge right now where AI does not know the parcel boundaries of the properties it works on for you. It can view a map of the property, but until it also knows exactly where that property starts and ends, it can't do things like analyze topography, flag wetlands impacting your buildable area, or calculate street frontage. We're on it, and we hope to have it solved soon. This is the invisible work. It never shows up in a task list, and it's the reason the platform keeps getting sharper. The Horizon Visual Apps go live next week. Expect several tasks to start emitting them right away, and more to be upgraded over the weeks that follow. Also: the Airtable integration is ready. Which means CRE Agents can now build databases for you. If you've ever wanted a structured, queryable record of every deal you've looked at, every owner you're tracking, or every comp you've run, we can help you stand that up. Bigger Picture The data point that hit me hardest this week: Anthropic and OpenAI each recently crossed roughly $30 billion in run-rate revenue, with each company sitting at roughly 4,500 to 5,000 employees. To put that into perspective, when Google hit $30 billion in run-rate revenue, it had nearly 25,000 employees. Roughly five times the headcount! The implication for CRE is direct. If eighteen months from now you are not earning more revenue per head, or managing more AUM per head, than you are today, you are doing AI wrong. The leverage is real, and it has already been demonstrated at a scale nobody thought possible five years ago. The only question is who does it best in our industry. It's a genuine privilege to be building these AI tasks on behalf of this group. I mean that. Spencer