Posts related to artificial-intelligence category

Consequence Is The Moat

Vinay Patankar · 13 Sep, 2026 · Artificial-intelligence

I have been reading the AI agent pitches aimed at property management, which is a category I find useful precisely because nobody writes think pieces about it. Small teams, tight margins, real consequences. A good place to see what is actually true. Here is the field as it stands, and then the thing the list made obvious to me. 1. Dash. Ours, so read the bias in. It sits in Slack or Microsoft Teams, connects to the property software and the mail and the accounting the office already runs on, does the fetching and compiling and drafting, then stops. Anything that goes out or spends money comes back into the channel with an approval on it. 2. EliseAI. AI for housing and healthcare, built around automating how an organisation talks to prospects and residents at scale. Big portfolios, big volume, platform decision. 3. Zuma. Its assistant is called Kelsey. Handles enquiries and scheduling through your CRM, chases rent with promise-to-pay tracking, runs move-outs through to deposit returns, and answers the phone. A team of human experts sits behind it, which is a genuinely interesting design choice. 4. Funnel. Calls itself multifamily's first AI front-office-only platform. Agentic follow-up across email, chat, SMS and voice, a renter CRM, a contact centre, a resident portal. The most complete answer on the list and the largest commitment. 5. LeadSimple. Leads, calls, processes and work orders on one timeline with AI underneath. Less glamorous, closer to where the week actually goes. 6. Property Meld. Maintenance and only maintenance, which for a lot of companies is the entire problem. 7 and 8. The coding agents everyone keeps suggesting. Codex and Claude Code. Both excellent. Both built for engineers working in a codebase. They come up in every one of these conversations and they are the wrong shape, which is worth saying carefully because it is not a criticism of them. Here is what the list told me. Every serious entry is organised around a single function, or around replacing the front office entirely. Almost nothing is organised around the gap between functions, which is exactly where the week disappears. A work order, a rent roll, an owner email and a ledger entry all describe the same unit and sit in four systems, and a person spends their days carrying information between them. That job has no product category. And the coding agents show you why the gap is hard to fill. They are the most capable agents in existence right now and they are built on an assumption that does not transfer. In a codebase, a wrong move is a revert. You branch, you try, you throw it away. That assumption is what lets them be fast and autonomous. It is also why nobody sane points one at a rent ledger. A notice that reaches the wrong resident is not a revert. So the frontier in this category is not capability. It is consequence. Reach into the real systems, restraint before anything leaves the building, traceability so you can ask where a number came from, and living where the team already talks so adoption does not die in a new tab. None of those four are model problems. They are product problems, and they are specific to a domain. Which is why I think this market splits rather than consolidates. Horizontal agents keep winning on breadth and will improve faster than anyone expects. Domain agents win where being wrong is expensive, because that is where the last ten percent of the work lives and no benchmark measures it. An operator said a thing to me that I keep coming back to. He did not want an agent that was clever. He wanted one that would stop and ask, the way a good new hire does in week one. That is a much harder product than it sounds, and it is worth more than a few more points on a reasoning score.