Sean Thomas

Enjoy lifting weights. Multi-state real estate broker (TX, CO, GA, NC, SC, VA, MD, DC, OR, WA, AZ, FL, and CA)

[Sean Thomas @SeanThePaul]

This is often my message in trying to lead AI adoption across Orchard: Anything is possible!

[Lenny Rachitsky @lennysan]

Recurring advice from my recent podcast guests: Be more ambitious.

[Sean Thomas @SeanThePaul]

Damn, I want a screen on a wall I can talk to! And Astra looks sick

[OpenAI @OpenAI]

This is GPT-6 Astra.

Anything you can do on a computer, Astra can do for you. Fast.

[Sean Thomas @SeanThePaul]

By 2027, the top 20% of agents will control nearly all the deal flow.

Tough inventory pushed the hobbyists out, and buyers verify track records before returning a call. The agents winning right now gave their low-judgment paperwork to AI so they can move twice as fast on relationships.

A legacy brokerage name and six software subscriptions won't save an office when three lean agents can handle an entire territory.

[Sean Thomas @SeanThePaul]

Always respect Dan’s take on the new models.

Most exciting part: Fable is now a model you want to converse back and forth with.

[Dan Shipper @danshipper]

BREAKING:

Anthropic just dropped Fable 5.1—and CLAUDE IS SO BACK.

We’ve spent the last week testing it at @every across coding, writing, and knowledge work. Our verdict:

It's finally Fable for everyone.

It’s the strongest coding model we’ve used, but now it's fast, token-efficient, and CRUCIALLY actually speaks like a normal person.

Here’s our vibe check:

- A monster at coding. @kieranklaassen rebuilt a working version of Proof, our document editor, from one prompt. It added useful details he hadn’t requested, and it handles enormous coding jobs that run for days at a time. It built a computer use Mac app for me called Hands in one-shot that other models failed at.

- A Claude our writers want to use again. It has clearer prose, fewer AI tells, and it takes an edit without arguing. It's a significant upgrade over Opus 5. And won @kplikethebird's heart back.

- About half the tokens as Opus 5, and much faster. In our Slack-agent tests, it delivered comparable results to Opus 5 using about half as many tokens, in about 60% of the time.

- Knowledge work you can delegate. It can produce great knowledge work—like slide decks—end to end without making slop. And flew threw @hammermt's tests with flying colors.

- It now supports zero-data-retention agreements. Now businesses can actually use it! A big barrier to Fable adoption is gone.

Net Result:

It's obviously an Opus 5 killer. If that was your daily driver you should switch today.

If you're using GPT-5.6 in ChatGPT for Work, it's spinning the wheels on for big delegated tasks.

I still use ChatGPT for Work more day to day, but I use way more tokens in Fable 5.1. I send it off at the beginning of the day to do big programming projects, like end to end MVP builds, and check in every once in a while.

State of Play:

The big knock on Anthropic was they built a supergenius in a datacenter that was almost unusable. It was too slow, argued back, and talked in technical gibberish. They've managed to solve those problems and more with Fable 5.1!

[Sean Thomas @SeanThePaul]

Am I the only one who has a hard time figuring out how to use up all of the potential in my @OpenAI @claudeai @bot @cursor_ai subscriptions? What automations are valuable enough to max my subscription token use?

[Sean Thomas @SeanThePaul]

AI assistants are getting their own computers.

The brokerage test isn't whether they write content. It's whether they can turn a client consultation into next steps without sending, changing records or making decisions before a person reviews it.

https://t.co/2KJzryfELx

[Sean Thomas @SeanThePaul]

I've been learning as I go, so my prompting has often been elementary at best.

Even then, I've rarely seen a frontier model actually hallucinate.

When it does make something up, it's usually because I threw a half-baked question at it with zero context and expected mind reading.

[Sean Thomas @SeanThePaul]

I am tightening The Real Estate Leverage Report around AI tools that matter inside a brokerage.

Each issue: a few tools worth knowing, what they could change for brokerage work, and one practical workflow, prompt, or template.

What would you want covered first?

[Sean Thomas @SeanThePaul]

Most of what eats your energy as an agent isn't even in your control.

We started an internal podcast at Orchard this week. One of the agents we interviewed doesn't sit around waiting for conditions to be perfect. She picks the next thing she can actually do and moves.

I kept coming back to that same idea in a broker education session. Figure out what's yours to control. Work on that. Ignore the rest.

[Sean Thomas @SeanThePaul]

@daniel_mac8 This is really cool. How do you customize the widget?

[Sean Thomas @SeanThePaul]

Run this in one market:

1. Ask ChatGPT, Gemini and Grok the same buyer or seller question.
2. Open the first source each cites.
3. Assign one factual correction.

Which source do you expect AI to use?

[Sean Thomas @SeanThePaul]

@ChatGPT This is awesome - Can it reuse logins for frequently used websites, or do we have to log in each time?

[Sean Thomas @SeanThePaul]

A homeowner asks AI which agent should list a home. Your top producer may not appear.

I built a 20-minute check: one market, three AI tools, three cited pages, one assigned correction.

[Sean Thomas @SeanThePaul]

I've been building with this stuff enough that I got less interested in another chatbot.

I love chatbots. I use them all day. I just think the real leverage is when the work disappears without me opening a tab.

Wrote about that here:
https://t.co/ssgTR5AHUD

[Sean Thomas @SeanThePaul]

Fair Housing people said something this week that stuck with me: the AI vendor is not the one with the license. You are.

I'm not going back to doing all of this by hand. I dump work into these tools constantly.

I just don't pretend the model is the one who has to answer for it.

[Sean Thomas @SeanThePaul]

I use this stuff on real work all the time. First drafts, follow-up, the boring certification and paperwork loops.

It is genuinely fun. It is also still my name on the file if the output is wrong.

The tool can take the work. It cannot take the judgment.

[Sean Thomas @SeanThePaul]

I've been trying new tools the way some people try restaurants.

Most of them are just a nicer way to ask a question. Every so often one actually takes a job off my list.

That's the only review that matters to me.

[Sean Thomas @SeanThePaul]

A lot of software is just bolting a chat box onto the same old product and calling it AI.

I keep going back to the tools that just do the task.

I don't want another place to prompt. I want the chore gone.

[Sean Thomas @SeanThePaul]

There's a new faster model every other week now.

I care less about the speed and more about whether I can throw a messy task at it and get something I can actually use.

Waiting 10 seconds vs 1 second is not my bottleneck. Me cleaning up a confident wrong answer is.

[Sean Thomas @SeanThePaul]

OpenAI said this week that businesses are now a bigger part of their revenue than regular ChatGPT users.

Doesn't surprise me. You fall in love with it in the chat window. You pay for it when it starts doing actual work.