# I keep seeing that ChatGPT can’t do real work. Here are my 10.6 counterexamples. — retained record

> Retained reddit record from reddit:ChatGPT.

Canonical: https://highsignal.app/data/records/143e49cb25225b3c

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ChatGPT)  published 2026-11-01 17:07 ingested 2026-11-02 02:37

[original source ↗](https://www.reddit.com/r/ChatGPT/comments/1xl8e7p/i_keep_seeing_that_chatgpt_cant_do_real_work_here/)

## retained text



```
I keep seeing claims that ChatGPT is useless or not smart enough to do real work. My counterargument is the work my AI assistant and I have actually done.

We’ve built websites, connected business workflows, and moved family projects off the “somebody should really deal with that” list. We’ve also argued extensively about what “finished” means. Hence the 10.6.

I’d estimate 90% of my side of these projects happened by talking to my phone or computer. The other 10% was me at a terminal, executing requests and handling the parts that needed me there.

For context: I’m a realtor, a theater nerd, and a parent of two adult children and two school-age children. My wife works for a large tech company in a demanding job. Our schedules are fractured, and family comes first. Life’s maintenance, professional development, and improvements to my business tend to get pushed aside. The list grows. The uninterrupted afternoon never arrives.

So I started giving my AI assistant, which I call Goose, ongoing projects, source material, access to connected tools, and specific instructions about what it could touch.

I now actually look forward to long car trips. I can talk through what’s open, paused, or blocked and make decisions that move projects forward. Somewhere along the way, “Are we there yet?” became “What else can we knock off the list?”

Here’s the list.

**\*\*1. We built and published a regional theater website.\*\***

Multiple regions, theater profiles, production listings, navigation, images, and revisions. I supplied the idea and opinions. Goose supplied the technical work and an apparently unlimited tolerance for “no, the OTHER photo.”

**\*\*2. We built a permission-review workflow connecting email, a tracking sheet, and the website.\*\***

It’s set up to compare theaters’ answers about photo and logo use with earlier correspondence, flag conflicts, and prepare individual reply drafts. It’s also authorized to replace declined imagery with placeholders, publish the correction, and verify the live result.

Replies stay drafts for my review. A correction only counts as finished once it’s verified. The point is to make someone’s answer affect what appears on the website, rather than disappear into my inbox.

**\*\*3. We connected the theater website’s content to app development.\*\***

We’ve built a shared content feed and working app builds, so the website and apps can draw on the same information. We’re well beyond “what if this were an app?” Getting those apps into the stores is another matter. See 10.5.

**\*\*4. We built out my real estate website.\*\***

Neighborhood guides, buyer resources, and consultation tools based on my actual business and local knowledge. This was a sustained build-and-revise project, not a prompt asking for a picture of a house with the word WELCOME on it.

**\*\*5. We created a repeatable property-pricing system.\*\***

Buyer and seller reports with comparable sales, condition assumptions, price ranges, and uncertainty. I challenged the analysis and made it explain its reasoning. “Because the middle of those two numbers looked nice” does not survive that conversation.

**\*\*6. We built scheduled property research around individual buyers.\*\***

Different criteria, listing-detail checks, duplicate prevention, candidate history, and price/status changes tracked in Asana. Teaching it that “a bedroom downstairs” isn’t necessarily “the primary bedroom downstairs” has become my contribution to computer science.

**\*\*7. We made my inbox and CRM useful together.\*\***

Daily priorities get checked against actual correspondence and contact history, so an old task doesn’t automatically become today’s emergency. Getting a computer to recognize that something is already done is a surprisingly satisfying professional achievement.

**\*\*8. We got my calendars talking to each other.\*\***

Upcoming events are reviewed, the right account gets invited where permitted, existing responses stay intact, and duplicates get skipped. My various email addresses are finally learning that they work for the same idiot.

**\*\*9. We built practical task support for my adult son.\*\***

We started with video of an overwhelming project, broke it into manageable steps, and created a way to keep moving. It’s been working really well. No punchline here. This one is important to me. It’s effectively an executive function coach.

**\*\*10. We turned family lore into an archival investigation.\*\***

We’re researching my grandfather’s involvement in the invention of self-service gas and convenience stores. Names, timelines, historical sources, and an archivist willing to search unindexed newsletters. We haven’t proved the family story, but we’ve upgraded from “somebody should look into that” to somebody actually looking into it.

**\*\*10.5. We’re inches away from our first Google Play and Apple App Store releases.\*\***

They aren’t published yet, so I’m not counting store releases among the completed accomplishments. We have working builds and have done a lot of testing, screenshots, and release preparation. Final signing and approval steps remain.

I wanted to give this half a point because apparently the last stretch of building an app involves convincing two enormous companies that you are, in fact, you.

**\*\*10.6. My wife and I are about 95% of the way through preparing our wills.\*\***

We’ve done the preparation by talking to our AI assistants, together and separately, and having them email each other to help coordinate. We’re down to reviewing a final draft before taking it to an attorney and a notary. They aren’t legally finalized, so they don’t get a whole number yet.

This is exactly the kind of important, uncomfortable paperwork we would otherwise keep putting off. ChatGPT made the process fit into our lives instead of requiring us to become entirely different people who enthusiastically schedule estate-planning evenings.

Goose still makes mistakes. Sometimes spectacularly. I still review consequential decisions, and we’ve had serious discussions about why “I did it” cannot mean “I was optimistic about doing it.”

“Needs supervision” is a fair description of my experience. “Useless for real work” isn’t.  
We’re producing things I actually use, in the bits of time our family schedule leaves available. That’s what I wanted to show.

Yes, Goose helped write this post. I’ve already made it rewrite it several times, so this is also a live demonstration of the working relationship.

What’s one thing AI has helped you finally get off your “I really need to deal with that” list?
```



document fetched 2026-11-02 00:20 · canonical https://reddit.com/r/ChatGPT/comments/1xl8e7p/i_keep_seeing_that_chatgpt_cant_do_real_work_here
