Saturday, February 24, 2024

How I'm Using AI In My Real Estate Business

Like many people, I'm playing around with different generative AI tools - investing time and money - to see how they can help my business. At the moment, it's helping around the edges, saving pockets of time, but it hasn't structurally changed how I do things yet. Here are some of the tools I'm using: promises that you don't have to take meeting notes. It generates transcripts, meeting summaries, and action items. I wanted to try it for two reasons. First, I wanted it for myself so I wouldn't have to take as many notes. Second, I wanted to offer it as an added value to the people I coach - they would have a record of our conversations for later reference.

It's OK... I find that I don't have to keep as copious of notes - just the highlights - which is excellent. I like having it recorded, and searching for specific parts of the meeting is easy. The summaries are OK, but the action items are useless.

I also found it helpful for group meetings when someone can't make it. I can send them the meeting notes, which include the summary and transcript. One feature that seems cool, but I have yet to really try, is using the built-in chatbot to ask questions about the meeting. It's like a fancy search engine, which seems like it could be helpful.


The idea behind Descript is that you can edit a video or podcast (*cough* Furlo Capital Real Estate Podcast *cough*) as if it were a Word document. So, you add your raw video, it transcribes it, and then, as you edit the transcript, it edits the video.

The editing is cool, but I still prefer doing it in Final Cut Pro. I have a hunch that getting better at multi-track editing in Descript would be faster than FCP, but the learning curve is a little steep.

Where Descript has been SUPER helpful is on the text side. I'll edit a video and then import it to Descript. Once it's transcribed, I have it help me with a few items:

  1. Output subtitles and a transcript. I used to use Rev for this.
  2. For shorts, subtitles can be added to the video. I used to do this by hand, and it was miserable.
  3. Write podcast and video descriptions. They're always written in 3rd person, and I haven't found a prompt that works to make it first-person (without ruining the overall description).
  4. Identify chapter markers. This is borderline perfect, though it usually gives more than I want, but it's easy to remove them.
  5. Suggest titles. I'll use these to make my own before analyzing it with a headline analyzer.


There are some big chat-style generative AIs to play with: Anthrop\c's Claude, Google's Gemini, and Microsoft's Copilot, to name a few. But something about the Wild West feel of OpenAI's ChatGPT attracts me. Sure, I want my AI to be careful and respectful, but I don't want it to be too cautious (I'm a "safety third" type of person).

The free version is OK, but if you're going to integrate it into your life/business, the Plus version is noticeably better. I'm finding it useful for three general tasks:

  1. Summaries: I can feed it a document, like a transcript, and ask for the key points. I haven't had to do this yet, but I could probably upload a rental agreement and ask questions about it (though, in general, I prefer to read the actual text).
  2. Idea generation: I've done the classic "Give me 10 fears that [target market, like data scientists] have about [doing something, like investing in real estate]," and it's excellent for that. I also like to give it a podcast transcript and a prompt like "What questions might someone have after listening/reading this?
  3. Transposing: I'm still new at this, but I've been playing around with creating podcast images from text prompts.

Arc Search

This iPhone app is a new tool for me. Instead of searching and getting links, I can get multi-media summaries and answers. If I don't like the result, I can easily do an old-school search (Just think, someday, Google may call it Google Classic Search). I'm unsure if it's "better" or faster yet - because it takes time to learn a new workflow - but it's cool.

Where's the line of too much AI?

Part of this play is helping me discover the best uses. Sure, I can ask it to write this article. It would be bland (plus, I like writing because it helps me think), so it's not a real option. But someday, it'll be able to write in my voice after I give it an unorganized blob of ideas. Then what? Do I get credit for writing it?

I use Grammarly for editing, and that seems to be OK. I take credit for doing data analysis, even though I use tools like Excel and R and not slide rulers. If I were a web developer, I'd use GitHub Copilot to help write code. Is that OK? It's just a smarter version of autocomplete, right? Does anyone care as long as the output is high quality?

To me, the line seems fuzzy. As a business owner, I'm not super concerned about how it's done, as long as it's done well. Would I care if I had a theoretical employee who used these tools to do their work in half the time but still charged the full time? I'd care about having a lazy (dishonest?) employee, but I'm paying for results, not time spent. The issue is the person's character, not the tool's use. I pay MORE for contractors to go faster, so it's not always about saving money.

Again, does anyone care if I use a WYSIWYG editor like Squarespace on a website instead of hand-coding it? No. Does anyone care if I use a pivot table in Excel to make a chart? Nope, it's considered an advanced skill. Does it matter if I wrote a helpful article using chatGPT for the first draft with some light editing? Well... No?

In each of these examples, I devised the idea and validated the final result. What if I hired someone to do all the work for me? Doesn't anyone really care as long as it's helpful? This is what businesses do, right? CEOs of mid-to-large companies can't do everything.

Perhaps the rub is that it's an implied expectation that humans do certain things (like write and edit videos), and somehow, we feel lied to when we find out a human didn't do it. Or, maybe it's because we're in the uncanny valley of AI quality. If that's the case, this will be less and less of an issue as society changes its expectations and the output improves. Does anyone else miss DVD special features? I do, but streaming is so convenient!

What's My Core Value?

A more helpful framework has been thinking about my core value. What do I bring to my business that's hard to replicate?

Let's use my podcast as an example. My core value is my experience, knowledge, and how I explain them. So, AI can help me think of topics, but I should be the one to choose the topic and talk about it. Then afterward, creating subtitles, descriptions, and chapter markers are all low-value tasks. So, I'm OK delegating those to AI (with my final approval).

I also think video editing is lower-value (for me), but AI isn't great at it yet, and my budget doesn't allow me to delegate it to someone every time. Basically, I'm OK with delegating everything (to humans or AI) around making the podcast, except for the on-screen part, because that's the highest value I bring.

Future Uses?

I do other lower-value tasks, and it's fun imagining a time when tools, AI or otherwise, can help with them.

  1. Bookkeeping takes a while. I've started delegating that to a property manager. Hopefully, someday, she can use AI to help her go faster. For example, Quickbooks will suggest categories, and that'll get smarter over time.
  2. I don't write a lot of commercial rental contracts, but I could see it helping with that. Or reading a contact and advising me on things to look out for.
  3. So... email... I want an AI to analyze my previous emails to match my writing style for a one-click auto-responder. And not just a one-liner auto-responder, but a complete message. I've read about some pretty extreme solutions, but they're more hostile to senders than "solutions." (The best -and most expensive - option is hiring a virtual assistant to reply on your behalf.)
  4. I have an odd plumbing problem. I can't find a suitable solution on Google or ChatGPT. I'd rather not pay a plumber, but it's increasingly looking like I don't have a choice.
  5. Imagine this: I see a property listing. I call the seller to learn more and I take a tour. I want to feed the listing, the call transcript, and the tour photos into a system that analyzes the property. It'll output a P&L, repair estimates, and project KPIs into a spreadsheet. I can then take over from there. I think? We're flirting with my biggest value-add now.
  6. Or, I give it a current property in my portfolio and its P&L over time. The tool looks at the market and can suggest sell/hold and rent rates. I haven't tried this yet, but I bet it can already help make rental listings (though, I have a template that's easy to use). Or, auto-general some good-looking reports?

All of these are still helping me to do what I'm already doing more efficiently. None of them fundamentally change how I'm already doing business. What would an AI-first real estate investor do?

I suppose I'd want a tool that analyzes every property in the US for me and generates offers I can submit. It would auto-generate rental contracts and do the follow-up on rent payments and notices. It would listen to every conversation and suggest the next steps.

Or, maybe I never directly interact with people! Instead of writing emails, I write/speak my thoughts, and it writes and sends them. I never directly analyze a property, but I write/speak/send the stats, and it analyzes it for me. If I'm working with a potential seller, I might tell the AI to keep following up until I get a yes or a no. The AI knows the analysis and my criteria and can automatically do a back-and-forth with the seller, searching for a mutually acceptable deal by tweaking the price and terms based on the seller's response.

Maybe? That seems far off, and at least for now, it's a part of my core value in my business. Perhaps a reasonable next step is to get better at telling the AI my raw, unstructured thoughts and letting it draft a response.

No comments:

Post a Comment