The official ChatGPT thread - the next big thing?

[Jeff Lebowski]

Wow.

[MartyFunkhouser]

One of my big annoyances in working with a small firm is document automation. We just don't have it. Something as simple as building a caption for a legal filing, or repetitive legal documents. I've tried mail merges in the past and they are annoying and I haven't got them to work right. It has been pretty much open a file under the client, save as, and then start working. Either that or just having an already filled out caption. It works, but I've always been annoyed by it.

Well, I am now getting what I need through tools I'm building out through Claude. I now have a base file for each client that has their information in it. It pulls from that to create a caption. It is a simple command line structure. I am also building out other automation tools. For example, I today build out having a text file that has hearing information in it (date, judicial officer, parties present, etc.) and it will generate a word doc with the correct caption, intro paragraph, notice of intent to submit, and a certificate of service. Just add the body text of the order and you are there. I can use AI tools also to draft the text of the body.

I'm finding myself doing something entirely new, I'm working from a command line more and more. Maybe it isn't more efficient, but it is more fun and more satisfying.

[Jeff Lebowski]

That is awesome funk.

[YOhio]

Google has come out with Agentic Vision and from what I've seen it's super impressive. Here's a demo sandbox with a bunch of different examples.

The one that kind of stunned me was where the prompt asked it to analyze an image that contained a flat image with 39 animal emojis. The prompt said, "Crop out all the animals, and use them as icons in a matplotlib plot showing the lifespan of those animals. Sort by lifespan." Within a minute each animal image was identified, extracted, and sorted onto a matlap plot by lifespan. Wild.

[MartyFunkhouser]

Just built a tool that creates 3 documents needed for a simple ex parte motion with the court. It takes less than about 5 minutes to create, clean it up, and file. It was a motion that I would normally give to a paralegal. However, just giving the assignment and reviewing the documents takes more time than running this tool.

I'm now thinking through more and more automation that I can add to my practice.

[Jeff Lebowski]

This morning using a combination of Google AI Studio and Claude Code, I built a web-based decision support and data analysis app that replicates >80% of the functionality of another app we have had been working on for years. Superior interface, better design, better responsiveness, more lightweight. I built it in 90 minutes.

[BigPiney]

This morning using a combination of Google AI Studio and Claude Code, I built a web-based decision support and data analysis app that replicates >80% of the functionality of another app we have had been working on for years. Superior interface, better design, better responsiveness, more lightweight. I built it in 90 minutes.

that is crazy.

[YOhio]

This morning using a combination of Google AI Studio and Claude Code, I built a web-based decision support and data analysis app that replicates >80% of the functionality of another app we have had been working on for years. Superior interface, better design, better responsiveness, more lightweight. I built it in 90 minutes.

Nice work!

[The_Tick]

I rarely used AI. I am doing a little more of it lately.

My wife has the ability to "retire" from CalStrs in March of this year when she turns 55. She gets an okay amount if she starts drawing then. If she waits until 62 she gets a really good amount monthly. I used ChatGPT and the one built into google. Asked it what the break even point was before she'd start "losing" money by drawing early rather than late. It told us that 71y10m would be the break even point, and then she'd start losing money after that. so almost 17 years of drawing for the break even point. Her mom passed away at 64 years old, her dad at 78. We split the middle and decided to do it now.

I am new at this stuff and it was fun.

[BigPiney]

I rarely used AI. I am doing a little more of it lately.

My wife has the ability to "retire" from CalStrs in March of this year when she turns 55. She gets an okay amount if she starts drawing then. If she waits until 62 she gets a really good amount monthly. I used ChatGPT and the one built into google. Asked it what the break even point was before she'd start "losing" money by drawing early rather than late. It told us that 71y10m would be the break even point, and then she'd start losing money after that. so almost 17 years of drawing for the break even point. Her mom passed away at 64 years old, her dad at 78. We split the middle and decided to do it now.

I am new at this stuff and it was fun.

I'd love to see the prompt you used (minus the numbers of course) for this. Because I would like to ask it the same question. I have run the calculator on CalPERS but this would be cool to work with.

[Bo Diddley]

I rarely used AI. I am doing a little more of it lately.

My wife has the ability to "retire" from CalStrs in March of this year when she turns 55. She gets an okay amount if she starts drawing then. If she waits until 62 she gets a really good amount monthly. I used ChatGPT and the one built into google. Asked it what the break even point was before she'd start "losing" money by drawing early rather than late. It told us that 71y10m would be the break even point, and then she'd start losing money after that. so almost 17 years of drawing for the break even point. Her mom passed away at 64 years old, her dad at 78. We split the middle and decided to do it now.

I am new at this stuff and it was fun.

Nice I've been considering running my numbers thru the AI to play with some different scenarios.

[falafel]

This is pretty incredible writing. Has a bit of a Christian resonance to it.

Embedded Tweet

I mean, its just reshaping human expression in a generic way.

Not that impressed.

[Jeff Lebowski]

I'd love to see the prompt you used (minus the numbers of course) for this. Because I would like to ask it the same question. I have run the calculator on CalPERS but this would be cool to work with.

Just fire it up and describe your situation. It will ask you whatever questions it needs to in order get enough data to do the analysis. I’ve used ChatGPT for strategizing a number of retirement financial scenarios, and it works great.

[BigPiney]

Just fire it up and describe your situation. It will ask you whatever questions it needs to in order get enough data to do the analysis. I’ve used ChatGPT for strategizing a number of retirement financial scenarios, and it works great.

I'm also really lazy, but I may try this.

[The_Tick]

I'd love to see the prompt you used (minus the numbers of course) for this. Because I would like to ask it the same question. I have run the calculator on CalPERS but this would be cool to work with.

If one has the opportunity to start drawing X amount of retirement funds at the age of 55, versus Y amount at the age of 62, at what point will I be losing money if I choose to retire at 55 instead of 62?

And then it asked a couple of clarifying things (is there a guaranteed level of inflation/COLA or things like that.

I have started to play around with the amount of money that I'd like to see when I retire out of my 401k/ROTH on a monthly basis. I put in my current balance, with 8-10% average growth along with 15% annual input off of current salary, and then how much money I would have starting at the age of 62. How much would it be if I wait until 63? And then I ask it to assume how much it would be if I waiting an additional year and project that same question to ages 64-70. And then if I up my percentage to max my contributions for the next 10 years. Kinda fun.

I enjoy watching it "spin" while it is doing the computations.