Using ChatGPT Alone Will Cost You Deals. Here's the Fix.

TL;DR

• "Just ask ChatGPT" is the most common AI advice in real estate, and the most useless.
• ChatGPT is a writing tool. It is not a research tool, a deal analyzer, or an operating system.
• Investors who treat one chatbot as their entire AI strategy lose to operators running a multi-tool stack.
• The fix is not better prompts. The fix is using the right tool for the right job.
• Your competition is already doing this. Catch up or get out-executed.

What "just ask ChatGPT" actually gets you

Open a thread. Paste in an address. Ask for an ARV and a rehab estimate.

ChatGPT will give you a confident, well-written, professionally formatted answer that is wrong.

The comps will be fabricated. The cap rate will be plausible and made up. The rehab number will be a national average dressed up as a local quote. The output will look like analysis. It will read like analysis. It is not analysis.

This is not a ChatGPT problem. ChatGPT is doing exactly what it was built to do: predict the next likely token. The problem is that real estate investors are using a language model as a market research platform, and then making offers based on the output.

That is how ChatGPT alone costs investors deals. And it has been the default advice for two years.

Why one chatbot will never run your business

There is no best AI tool. There are best-in-class tools for specific jobs. Operators who understand this win. Everyone else is still typing "act as a real estate expert" into a single chat window and wondering why their offers keep getting rejected.

We have been over the multi-tool argument in detail in Why One AI Tool Isn't Enough. The short version: every AI has a job. At Clark St, Claude does our writing and reasoning. Perplexity handles research with citations. Gemini handles graphics and video. Claude Code builds our internal software. The stack is the strategy. A single tool is not.

A general contractor does not show up with one tool. A plumber does not. A lender does not run a deal on one spreadsheet. Why does the average investor try to run an entire AI operation out of one chatbot?

Because someone on a podcast told them to.

Fact block. ChatGPT, Claude, and Perplexity are different categories of software. ChatGPT and Claude are language models optimized for writing and reasoning. Perplexity is a research engine that returns cited sources. Treating them as interchangeable produces three different qualities of answer for the same question.

The four ways "just ask ChatGPT" fails investors

1. It hallucinates comps and cap rates

ChatGPT does not have real-time MLS access. It does not have the local CMA data your title company has. It will invent comps that sound real. We have seen ChatGPT confidently produce a $325,000 ARV on a property where the right number was $278,000. That is a $47,000 swing on a 70% MAO calculation. That kills the deal. Or worse, it doesn't kill the deal, and you close on a bad number.

2. It writes seller scripts that miss the seller

ChatGPT will write you a beautiful seller follow-up email. The seller does not read email. The seller wants a phone call inside 60 seconds and a human voice that sounds like it cares. We covered this in Motivated Seller Scripts: Timing Beats Wording Every Time. The wording is not the problem. The delivery channel and the response time are.

3. It rewards the prompt, not the operator

The "prompt engineering" industry has convinced investors that the problem is their prompts. The problem is not the prompts. The problem is the workflow. A bad operator with a great prompt is still a bad operator. A good operator with a mediocre prompt and the right tool stack will out-execute every time.

4. It creates a single point of failure

If your business runs on one chatbot, your business runs on one company's uptime, one company's pricing decisions, and one company's product roadmap. When that tool goes down or changes overnight, you stop. Operators with a multi-tool stack keep working.

What to do instead

Build a stack. Assign jobs. Stop trying to make one tool do everything.

Here is what we actually run at Clark St:

• Writing and reasoning. Claude. It writes better than ChatGPT, holds longer context, and produces fewer confident-sounding mistakes on real estate topics. Pick whatever you want, but pick one and stick with it.
• Research with citations. Perplexity. Anything that needs to be true and verifiable runs through Perplexity, not a chatbot. Market data, regulatory questions, comp methodology, financing structures.
• Deal analysis. A purpose-built tool with real data. Not a chatbot. Our free deal calculator is one option. The point is: use a tool that knows the numbers, not a tool that guesses the numbers.
• Speed-to-lead. A purpose-built voice agent that responds in under a minute, every time, 24/7. Not a person checking email. Not a chatbot in a browser tab.

We laid out the full operator stack in The 5 AI Tools I Use to Run a Real Estate Business. Read it. Or build your own version. But pick more than one tool.

Curious how we put the stack to work? Ed talks through specific use cases on the Real Estate Underground podcast, including the episodes on the multi-tool workflow we run at Clark St.

The "just ask ChatGPT" mindset is a tax on your business

Every hour an investor spends polishing a prompt to make ChatGPT do a job it cannot do is an hour not spent calling sellers, walking properties, or closing deals. The mindset is the cost. The lost deal is the bill.

We built our own software because the off-the-shelf flip tools were not solving the problem either. The full story is in We Stopped Using Crappy Flip Software. The pattern is the same: stop trying to force a tool to do a job it was not built for. Get the right tool. Get on with it.

What "right" looks like

The investors winning in 2026 are not the ones with the cleverest ChatGPT prompts. They are the ones who built a small, deliberate stack of AI tools and assigned each one a job they are actually good at.

That is not flashy. It does not make for great podcast advice. It works.

If you take one thing from this article: ChatGPT alone will cost you deals. The fix is not a better prompt. The fix is a stack.

---

Try the free deal calculator. If you are still running deals through a chatbot, run them through a tool built for the job. The Elevista deal calculator is free, no signup, and gives you MAO, ARV-based offer math, and rehab modeling in under two minutes.

Also free:

• Real Estate Underground podcast — operator-level conversations on running a real estate business in 2026.
• Elevista prompt library — 30 prompts we actually use at Clark St, with the "why this works" reasoning behind each one.

---

About Ed Mathews

Ed is the founder of Clark St Capital, Clark St Homes, and Elevista. Clark St has operated across single-family, multifamily, and land development, and Ed is also invested as a limited partner in 1,000+ unit multifamily projects. Before real estate, he spent years building and advising companies in Silicon Valley. He's now building the leading AI SaaS company for the real estate investing industry at Elevista and hosts the Real Estate Underground podcast.