Vertical AI Outgrew Intake: Three Companies That Proved It
HomeLight's AI escrow agent, Enzo Health's agentic EHR, and Legora's Cadastral acquisition show vertical AI moving from intake automation to full lifecycle workflows.
By Springvanta
Three things happened in the first week of June that, taken together, say something specific about where vertical AI is headed. None of them are about intake forms or chatbots anymore.
HomeLight launched EVA, an AI escrow agent that automates the 120-plus discrete tasks required to close a residential real estate transaction. Enzo Health shipped the first agentic EHR built for home health agencies, connecting referral to reimbursement in a single system. And Legora, the Stockholm-based legal AI startup valued at $5.55 billion, acquired Cadastral, a commercial real estate AI platform — its fourth acquisition in five months.
What connects them: each company stopped trying to improve one step of a vertical workflow and started trying to own the whole thing. Intake to closing. Referral to billing. First contact to final resolution.
HomeLight EVA: from reading emails to wiring funds
HomeLight's EVA (Escrow Virtual Assistant) is a collection of AI agents with access to 80-plus tools. It monitors the escrow officer's inbox, opens incoming PDFs, extracts key data, enters it into the system of record, contacts agents when information is missing, kicks off title searches, pulls HOA documents from state databases, purchases them with HomeLight's credit card, and sends intake forms to buyers and sellers.
CEO Drew Uher told Inman that accuracy started at 25-30% when the first workflows went live in early 2025, climbed to 80-90%, then stalled for six months. "As recently as December 2025, I was honestly frustrated with the project," he said. "I wasn't sure if this was even a fully solvable problem."
By Q1 2026, all four major workflows reached what Uher calls 100% accuracy — defined as zero false confidence. EVA is allowed to escalate corner cases it can't handle. What it can't do is guess wrong and act on it. Given that escrow involves life savings, that distinction matters.
EVA has automated roughly 25% of the total escrow surface area so far. HomeLight secured $40 million in debt financing from BlackRock funds to scale nationally. The company isn't licensing EVA to other title agencies — it uses the technology internally as an AI-first title and escrow agency.
Victor Lund of WAV Group flagged two things worth watching. First, the data exhaust from automated escrows could become a licensing asset on par with companies like Cotality and Black Knight. Second, AI-handled wire transfers eliminate the human-screen vector that makes wire fraud so common in real estate. Whether EVA is actually more secure remains unproven.
Enzo Health: agentic EHR for home health
Enzo Health launched what it calls the first AI-native electronic health record for home health agencies. The difference from legacy EHR systems: Enzo EHR was built as an agentic system from the ground up, not an old platform with AI features bolted on.
The numbers are aggressive. Intake time drops from 70 minutes to roughly 5 minutes — the system reads incoming referral packages, verifies eligibility, and handles PDGM coding. Scheduling drops from 15 minutes to about 30 seconds. Visit charting time falls by 75% through real-time documentation from visit conversations. The system also monitors quality compliance, tracks CMS star ratings, flags documentation gaps, and auto-generates billing claims while catching potential denials before submission.
CEO Zach Newman's framing: "Every other vendor in this space tried to solve that by layering AI features onto old architecture. We built something different."
Enzo Health, founded in 2024, says it supports organizations covering over 500,000 patients annually and has seen revenue growth of more than 40x in the past twelve months. The company raised a $20M Series A led by N47. The demand driver is demographic: roughly 10,000 Americans turn 65 every day, and home health is becoming the preferred care model.
Legora acquires Cadastral: legal AI meets commercial real estate
Legora's acquisition of Cadastral is the fourth deal the Stockholm startup has closed in 2026, following Graceview (Australian regulatory intelligence), Qura (Swedish legal research), and Walter AI (Canadian legal AI). Legora raised a $600 million Series D at a $5.55 billion valuation in April.
Cadastral, founded in 2024 in New York, built an agentic AI platform for commercial real estate legal workflows — document drafting, investment memos, data room analysis, Excel models, PowerPoint decks. It landed over 50 firms in just over a year, including JLL, AvalonBay, Equity Residential, and Empire State Realty Trust, with revenue growing 40% month over month.
The acquisition extends Legora's "agentic operating system" (Legora aOS) into commercial real estate, one of the most document-intensive industries on the planet. It also anchors Legora's first major US engineering hub in New York, with plans for 200-plus people there and 300 across North America by year-end.
CFO David Eckstein told Law.com the company is actively pursuing more deals in IP, patent, trademark, and litigation — practice areas where its customer advisory board is asking for deeper capabilities.
What the convergence means

These three stories share an architecture decision, not just a timeline. Each company concluded that bolting AI onto existing vertical software produces incremental improvements at best. So they're building systems where agentic AI runs the full workflow — not a smarter intake form, but an agent that handles intake, processing, documentation, compliance, and output.
For SpringVanta buyers evaluating vertical AI in their own operations, the practical takeaway is straightforward. The vendors worth watching are the ones making a full-lifecycle claim — referral to reimbursement, first call to closing, intake to resolution — and backing it up with production metrics, not roadmap promises.
HomeLight's accuracy progression is instructive: 25% at launch, an 80-90% plateau that lasted half a year, then 100% on the four core workflows between February and May 2026. The plateau is where most vertical AI projects stall. HomeLight broke through it. Enzo and Legora are earlier in that curve, but their architecture assumes the same trajectory.
Sources:
- HomeLight Launches AI Agent to Automate Real Estate Closings (AI Investor Picks / Inman, June 5, 2026)
- Enzo Health Launches First Agentic EHR for Home Health (PR Newswire, June 1, 2026)
- Legora Acquires Commercial Real Estate AI Platform Cadastral (Law.com, June 2, 2026)
- Legora Acquires Cadastral (Legal Practice Intelligence, June 5, 2026)