# Using more AI for real work

Source: https://paragraph.xyz/@ccarella/using-more-ai-for-real-work
Kiwi News: https://news.kiwistand.com/stories/Using-more-AI-for-real-work?index=0x66d631f72949734dd4b4c0f57af2c0a62b6648a7a8fd61c1b4db95382e2285f3d8f76f94
Submitted by misha on 2024-09-02 UTC, 1 upvote, 0 comments

## Summary

The author describes a 30-day writing challenge with Saadiq, a product leader they have known since 2010, with each person staking $100 on a handshake agreement. Saadiq showed the author how he uses AI to build apps and prototypes, which prompted the author to spend a few days setting up AI workflows. The author says Claude was the first real breakthrough, particularly Claude Artifacts for building small apps and testing concepts. Using Claude, the author made sense of a massive CSV dataset and created a prototype Artifact in 15 minutes. The author now calls the Claude, Projects and Artifacts combination their highest-leverage toolkit and says the goal is to apply AI-first thinking to every problem.

## Article

Today marks the 16th post in my 30-day writing challenge with Saadiq. We've each staked $100 on the line, with a simple handshake agreement: miss a day, and the other claims the pot. It's not a crypto thing, just good old-fashioned accountability.

Saadiq is one of my favorite product leaders in my network. I got to know him in 2010 when he was the co-founder of Hot Potato (acquired by FB) and I was working with them on adding game mechanics to it for Hot Potato 2.0. Fast forward all these years and we are both still interested in emerging technologies like cryptocurrency and AI. He made the leap into crypto full time first with Candy Digital and then at PROOF.

A couple of weeks ago, we caught up, and Saadiq showed me how he's using AI to create apps and prototypes. This is the dream for any creative technologist. I have a computer science degree and I'm technically capable, but I'm decidedly more on the product design and strategy side. I know if I spend the time to really learn these tools, I can do a lot of rapid prototyping of products. But like all good things, such as going to the gym, it's hard to make time to master new tools even when I know they're transformational.

Reading Saadiq's essays during this challenge has been eye-opening. At first, I realized I was missing out on ways to use AI in my work. Then came the FOMO, followed by a bit of self-reproach for not embracing AI more fully. Finally, I buckled down and spent a few days flexing these tools and setting up my workflows.

It's not that I was completely AI-illiterate. I've been using AI for image generation in my art since 2021 and regularly tap ChatGPT for various tasks. But Claude was my first real breakthrough. Stumbling upon Claude Artifacts (which I frustratingly can't find now) was a lightbulb moment – I suddenly saw how I could use Artifacts to build small apps and test concepts.

The most crucial shift, though, has been in mindset. AI used to be near the bottom of my problem-solving toolkit. Now, my first thought is, "How can AI help with this?"

Just last Friday, I was procrastinating on a tedious task involving a massive CSV dataset. Every time I opened that file, I'd find an excuse to do something else. In desperation, I threw it at Claude. Despite my experience with CSVs and ChatGPT, it hadn't occurred to me to use AI here. With Claude's help, it took just 15 minutes to make sense of the data, reverse-engineer the original form, and create a prototype Artifact.

I've tested many tools recently, and I'm particularly excited about Cursor. But so far, the Claude+Projects+Artifacts combo has been my highest-leverage toolkit. My challenge now is to consistently apply this AI-first thinking to every problem I encounter.