I was worried I reached the end of the internet this week.

My stockpile of podcasts (more accurately my YouTube history) was looking a little barren.

But worry not, I still ended up with more things to talk about then I could fit into one newsletter.

Here’s what made the list:

• The Perfect Caption

• Strategizing with o1-preview

• Cloud Costs

• Twitch Streamer Tools Hackathon

• The Intelligence Age

The Perfect Caption

Last week, I was mulling over the idea of using someone else’s tools to get social media clips rolling versus taking time to use my own tools, and after a chat with GPT, I was convinced to stick with my own tools.

It would be really nice to say that all the clips I posted were done by me and not someone else’s product.

In the long run, it probably doesn’t matter that much, but starting out it may be a nice selling point.

With that, I stayed focus on my own progress, but unfortunately still have some work to do.

I wasted a little too much time being a perfectionist with how the captioning looked, switching between different pixel-looking fonts on Google Fonts, then finding that the other font options had weird line-spacing that I couldn’t fix using the options in the caption file, so I finally took the advice I normally give others.

Let the market decide.

I could spend weeks perfecting the font, or I could get videos out and start testing what works.

Here’s where I landed for now:

Not bad!

I’m going to start with something like that and see what kind of feedback I get, so let me know what you think!

Clips should start uploading to YouTube next week if I can keep the distractions to a minimum.

With that, on to some distractions!

Strategizing with o1-preview

I was waiting for Whisper AI to transcribe just the first 2 hour block of one stream, which was taking about as long as listening to the audio file itself, so I’m going to need to get that process working overnight or something, but since it was taking a long time, I decided to do some strategizing with o1-preview.

I laid out basically everything in the process so far, telling it how I’m recording, stripping audio, cutting silences, transcribing the remaining audio, then searching the transcript to find social worthy clips.

I also told it the tech I wanted to use, and my goals for where I wanted to end up and what I wanted to prioritize.

And in maybe 30 seconds of thinking in total, I had a comprehensive roadmap to follow so I can better stay on task.

I was surprisingly excited to see a clear path forward outlined in bullets and weeks.

• Phase 1: MVP Development and Initial Deployment (Weeks 1-2)

• ⁠​Phase 2: Iterative Refinement and Basic UI Development (Weeks 3-4)

• Phase 3: Enhanced Automation and Feedback Integration (Weeks 5-6)

• Phase 4: Optimization and Scaling Preparations (Weeks 7-8)

• Phase 5: Advanced Features and Automation (Weeks 9-10)

• Phase 6: Continuous Improvement and Scaling (Ongoing)

Awesome, now I have a timeline to try to beat.

And it’s feature complete, helping me remember aspects of the project I would inevitably forget.

It’s still insane to think you can have access to something like this at any time, all the time.

Hard not to imagine what you can create with a team of AIs, but more on that later.

For now I’m excited to knock out that roadmap faster than a 2 week sprint cadence.

Thanks, o1-preview!

Cloud Costs

Since I’ve de-prioritized some of the divs initiatives for the moment, I have also been thinking about how to reduce costs.

Basically I have a few sites operating on Webflow subscriptions that don’t need to be active in Webflow.

Luckily I’ve already devised a way to fix this issue when I considered the SEO route.

I can export all the content for the sites that aren’t active and host them somewhere else for much cheaper - probably Google Cloud.

Webflow charges $29 a month to host a website with a custom domain and a minimal CMS (content management system) for storing content.

I have 6 websites hosted on Google Cloud right now and it usually costs about $.11 per month to host all of them.

GCP (Google Cloud Platform) is more manual work, but also more control and less cost, so it seems like the better play for now.

And one tool that could help out with this, which I just learned about thanks to Fireship, is Pulumi.

Terraform had been on my radar for cloud management but I hadn’t considered alternatives yet, and Pulumi seems like a promising alternative.

One command to handle GCP uploads and and one to handle shutdowns sounds very convenient so that will be on the list to look into soon, especially as the monthly Webflow payment notifications roll in and make me want to take action sooner.

More from Fireship on how to make the cloud better:

Twitch Streamer Tools Hackathon

Twitch is running the Twitch Streamer Tools Hackathon right now through November 4th.

Since hackathon’s have been a theme in this newsletter and I’ve done a podcast about them, it felt right to include the idea here.

My friend’s hackathon group plans to meet up and submit a project later next month, and they also already have the idea.

I was flattered when he told me they wanted to use one of my old project ideas from a long time ago.

When I was starting to stream on Twitch, I was thinking of things that could be fun to develop to add some interactivity to live streams.

I came up with the idea of creating a fake stock market where each stream channel would have a stock valuation and you could use fake money to buy, sell and trade streamer stocks.

We’d somehow come up with a price for the stock based on a few stats and try to update channel stock prices in real time so viewers and streamers could track prices and have something to interact with.

It sounded fun in general, but this also happens to fall into one of the categories for the hackathon:

The perfect opportunity to give this idea another chance.

A lot of work to do on that, but I’m looking forward to hacking something together next month.

I have a feeling we’ll produce something entertaining.

The Intelligence Age

And let’s wrap up with a wild right turn from video game streaming to the coming Intelligence Age Sam Altman wrote about on his blog this week.

Honestly I think the two concepts are surprisingly linked.

This could go on for a long time, but I’ll try to keep it short.

Altman sees a future where superintelligence exists “in a few thousand days.”

I can’t help but agree.

But if that’s the case, what do humans do?

If AI can learn, understand and execute faster and more effectively than a human being could ever hope to, what’s left for humans.

A combination of 2 ideas are my big bet for the future if this happens soon.

1. People will always want entertainment.

2. People are obsessed with people.

I’ve mostly thought about the entertainment idea on my own, but it’s also pretty obvious so I can’t take credit for it.

There’s literally too much entertainment to consume and yet more is created everyday.

People will always want something new and different to be entertained by, and everyone has different tastes, so the demand for entertainment will never end and will always change.

As for people being obsessed, I remember this clip from Ben Horowitz:

“Humans are obsessed with humans.”

Cheetahs have been able to run faster than humans for forever, but we still watch humans race other humans.

Computers are already better chess players than humans, but there’s 12,000 people, as of writing this sentence, watching live streams of humans playing chess (although some portion of those viewers are bots…).

Putting the 2 ideas together, there’s a good chance people will always want entertainment from other people.

In a world where there’s no need for people to work anymore, and whatever you can think of can be generated by AI in seconds, there will undoubtedly still be a demand for entertainment and goods from other people.

How that will work in the post-capitalism world Altman has also speculated, I’m not as sure.

I could see business becoming a sport, competing to build the most successful company, with a friendly, all knowing AI there to facilitate the rules of the game for the players.

And maybe this would all play out on live streams like the ones today, although hopefully in a more impressive way, like ultra-realistic VR - no bulky headset required.

But at the end of that future day, people will still be watching people do people things.

Clearly I’m an optimist for the future of people and AI.

There’s always a chance it could go the complete opposite direction, but I think the odds are very low.

Maybe not ignore-them-completely low, but certainly much less than the-world-is-ending-tomorrow low.

At the same time, I’m sure the road ahead will not be all sunshine and rainbows either.

It won’t happen overnight and mistakes will be made.

But I think we’re headed in the right direction.

I look forward to the Intelligence Age as a net positive.

And at the same time…

I for one welcome our new superintelligent overlords.

And that’s it for this week! The intelligence age is coming, but the entertainment doesn’t have to end!

Those are the links that stuck with me throughout the week and a glimpse into what I personally worked on.

If you want to start a newsletter like this on beehiiv and support me in the process, here’s my referral link: https://www.beehiiv.com/?via=jay-peters. Otherwise, let me know what you think at @jaypetersdotdev or email [email protected], I’d love to hear your feedback. Thanks for reading!