Finding people who are self directed and curious and passionate

Paths for problem discovery, using GPT-4 to build for me, and organizing an AI event

I’m endlessly fascinated by people who are self directed and curious and passionate and exploring different things. I like hearing the process discovery stories along the way. I’m drawn to artists or side projects or inquiries or founders and it was nice being in SF this past month feeling that energy as I wonder to myself about discovering valuable problems to solve and how I can find ways to pay rent that still give me agency and learning and cultivate my curiosity.

Recap of projects and thoughts and things over the last month. TLDR;

• GPT-4 allows you to build apps by describing what you want, rather than writing code. I used it to code a command line tool that summarizes any web page.

• I organized an AI Show & Tell in SF at NFX, with 20+ projects and ~75 attendees. Hosting informal conversations about what folks are working on inspires next steps and makes the AI space feel more welcoming for new devs.

• There are a variety of paths to finding a compelling problem to work on. I found a strong builder energy in SF, with lots of makers up for jamming on this.

Conversations on problem discovery

Over the past few weeks I asked founders how they chose what they’re working on now, especially curious of b2b or niche ideas. From a previous post: How did you find the problem you’re working on now? How did you know it was valuable? Paraphrasing a few convos:

• Consult and talk to customers. “I was consulting with several businesses and started working more specifically in content marketing for them I became an expert in it, and then created an AI product around it.”

• Join a company with someone who knows the space.“My company is in real estate, I found it because the job I was working in before was also in real estate. A friend pulled me into his company. I had no experience in it, but learned enough building the product. Then I noticed an opportunity around pricing for real estate, and that’s my company now.”

• Just build and launch. “What worked for us was to build the idea and launch. We put it out there. Funders saw the tweet and DM’d us. That started conversations. People see you launch and know that you’re serious.”

Building things 🚀🚀🚀

Using GPT4 to program for me. After seeing so many threads about coding with GPT-4, I had to try it out myself. I used GPT-4 to code a command line tool that summarizes any web page. It felt wonderful to collaborate with AI like this.

• Read more about what I made here: “How I used GPT-4 to code an idea into to a working prototype” There’s a link to the repo in the post too if you’re curious what the code looked like.

• Building apps by describing what you need, rather than learning how to code, could allow more people from more fields to build tools they need. This is inspiring! I’d love to explore this tool building process as a workshop with participants across industries to see what ideas come up.

• A buddy and I tried out this workshop idea. She shared her idea with me and we had a session using chatGPT to flesh out a product brief, technical spec, user research plan, and MVP for an app idea she had. Then we used GPT-4 to get started on the code for the web app. It was incredible jamming on the idea and making it more concrete.

Updating my AI chat apps with the chatGPT API. OpenAI released a chat API with gpt-3.5-turbo, which is much cheaper than using the completion API, the previous endpoint for build AI chats.

• I updated the chat tutorial to use the new chat api, you can find the updated repo here. I also updated my streaming example to use the new API too here.

• What this also means is that it’s cheaper to share apps using these endpoints too! If you’re interested in using any of these AI chats I’ve built, reply here or send me a note and I’ll share the link.

Throwing an AI Show & Tell in SF

I put together a generative AI show & tell at NFX. Here’s the epic selfie we took at the end, and more details of the event. Major props to NFX for hosting, and Manny for hopping in to help organize.

• The show & tells are fast and lightweight. The format is like a science fair + happy hour. People come with laptops, set them up somewhere around the room, and share what they’re working on. Halfway through I get everyone in a circle and we go around sharing 20 seconds or so on who we are, and either what we’re making or what we’re curious about — all to spark conversations after. It’s a fun energy! We get feedback on projects, see what people are working on, and just meet each other.

• There was a good mix of AI products and also creative AI applications. Often events I would go to would focus on the product side. An attendee remarked how it was nice to see so many creative apps too.

• My favorite outcome of the event was how it made building in the AI space feel more welcoming. One attendee shared that he was inspired to build projects afterwards — things felt more approachable because people shared the process in making things. He could understand how things worked.

• If you’re interested in helping host one, or attending one, let me know! Up next is New York. Would love to throw one where I’m based now in DC. (Other cities that come to mind — Boston, London, LA…)

• In a future post I’ll share more about my process organizing events — if you have any questions on this please reply and let me know! In the end I’m happy it worked out so well. Organizing it was a grind — I just repeated in my mind: set the intention, take action, let go of the outcome.

The builder energy in SF is everything

• SF is a Disney World for AI right now. The energy and hype and eagerness for connections is real, and it’s absolutely motivating. There’s several meetups and hackathons a week. Most nights you can find something. You go out and quickly get to know the community.

• I could just share whatever I was mulling on and get some ideas. The GPT-4 demo comes out, and later that day I meet someone new and just ask: you see the demo? And we get hyped on ideas from it. Or I spend an afternoon working with GPT-4, overwhelmed with what to do. A few hours later I meet an engineer who riffs with me on experiences with what worked well and what didn’t. I walk away with more ideas on what to try out.

• I meet a dev at a hackathon who’s on pivoting his AI startup and he shares what he’s learned from experience and other companies and we jam on moats, getting initial customers, and positioning his product. I run through his homepage to refine the story of his product. It’s energizing.

• My favorite event by far was Demos & Chill, inspired by WaffleJS, in turn inspired by BrooklynJS (both of which I love). Demos & Chill invites folks to share projects in an informal environment — no work stuff, encouraging things that are creative. You find yourself in a room full of self-directed curious people all excited to meet. When I moved to SF in 2016, I had a hard time finding this — people building their own things. Most were focusing on climbing the tech ladder. During the pandemic, people who lived in SF for work but hated it moved out, and those who wanted to be in SF stayed or moved in. The result is an eager and enthusiastic community.

More fun

• I demo’d doodle & diffuse at an art party, it was kind of ridiculous and fun.

• I joined a few drink & draw & hangs while in Oakland. I started these summer 2021 and love that they’re still running weekly.

What’s next

• Talk to more folks and find interesting problems to solve. Currently reading The Mom Test, and working on a plan to start sourcing more convos. Specifically looking for b2b problems to work on leveraging generative AI. Any tips or advice here, I’d love to know :)

• I’m fascinated by this pattern of AI in a loop. You create a loop with an AI of thoughts, actions, and observations. From the linked post: “Running ReAct for yourself for the first time is such a moment, like just the ohhhhhhhhhh of possibility space opening up.” 🤩 Going to get something running with langchain.

• Curious what it’ll feel like to write a plugin for ChatGPT, and what ideas will come from making one.

Thanks for reading! You can follow what I’m up to by subscribing here. If you know anyone that would find this post interesting, I’d really appreciate it if you forward it to them! And if you’d like to jam more on any of this, I’d love to chat here or on twitter.