Over the last week I’ve been working with a real estate servicing company that had a very specific, very annoying problem: They deal with hundreds of plot deeds every month – different formats, different structures, half of them scanned PDFs – and their team was manually: - Reading through each deed - Extracting key details (owner, plot number, boundaries, consideration, dates, etc.) - Re-entering all of that into their internal system / Excel It was slow, error-prone, and every “urgent” file jumped the queue and broke their process. So I built them a system that does three things: 1. Ingests plot deeds (PDFs, scans, etc.) 2. Extracts the key legal + property data in a structured way 3. Pushes everything into a clean, standardized format they can use for drafting, review, or downstream automations Some details that ended up mattering more than I expected: - Handling inconsistent deed formats (old templates, new templates, different registrars) - Catching missing / conflicting details (e.g. party names not matching across pages) - Making it easy for a human to quickly verify the output instead of re-reading the entire deed Impact so far for them: - Turnaround time per deed went from “whenever the team gets to it” to a few minutes - Their senior people now spend more time checking edge cases instead of doing data entry - They can actually take on more files without immediately hiring more staff Not sharing this as a flex or offer – more as a case study for anyone here who’s: - Working with real estate processes - Dealing with legal / property documents at volume - Or exploring AI for document-heavy workflows where accuracy matters If you’re doing something similar (real estate, legal ops, title, servicing companies, etc.) and want to swap notes, I’m happy to walk through how this is set up, what worked, and what broke along the way. Drop a comment or DM me and I can share more of the architecture, tools and lessons learned – could save you some time if you’re building in this space