Image generated by Midjourney. Obviously.

I recently went hiking with a friend of mine. At some point during the hike we started talking about how I’d stopped writing retrospectives. His main feedback was: “You feel much further away.”

This got me thinking. And the idea that I’m losing touch with a group of people I like made me a bit sad. So I decided to try and provide a short update on my work life—a retrospective, if you will.¹

What I’ve been doing

These are the main things I’ve been up to, and how it’s been going.

Pegasus

SaaS Pegasus continues to be my “main hustle” and takes up about half my time. A lot of my time goes into support—both supporting customers and keeping Pegasus up to date, fixing bugs, cleaning up code, etc.

Outside of support my main project for the last few months was rolling out an overhaul to Pegasus pricing, which I finally launched a couple weeks ago.

Here’s the old pricing page:

Pegasus’s previous pricing.

And here’s the new one:

Pegasus’s current/new pricing. The sale period will end this week.

Mostly, this is me raising prices, though I added a lower tier with many features disabled so that I still have a more affordable option.

I launched the changes with a “sale” that kept the original prices in place for a couple weeks. This sale will end early next week, at which point we’ll see how the new prices do. The two weeks since I’ve launched have been the most profitable for the project outside of Black Friday, so I’m happy with the results so far.

The ability to tie different licenses to different features under the hood was a big technical project and where the lion’s share of the effort went. Hopefully it paves the way for more experiments in pricing moving forwards—though I’ll let the dust settle on the new pricing for at least a few months before making more changes.

The new feature tiers that I introduced to Pegasus. Less blurry version available on the pricing page.

Community Keeper

As I mentioned in Quit Monkey, Grit Monkey, I launched a new product—called Community Keeper. At first I was positioning it as a permanent archive for your Slack communities, but it’s since morphed into a hodge-podge of tools that help me run the Pegasus Slack.

The project is simultaneously going great—in that I’m getting tons of value and having fun with it—and going terribly—in that nobody besides the Pegasus community is using it. The most interesting pieces are now related to AI (see below).

Adventures in AI

Turning 40 in the year of the AI revolution has really accelerated my progress towards feeling like an old man.

— Cory Zue (@czue) March 20, 2023

Like everyone else in the world, I’ve gotten quite interested in the latest advances in AI and specifically large language models (LLMs) like ChatGPT and GPT4. I’ve been learning and experimenting with the tech in various ways.

The first thing I did was add OpenAI (ChatGPT and DALL-E) integrations to Pegasus. That was like five lines of code. Then I started going deeper.

I built a custom chat bot on top of the Pegasus docs. It knows everything in the docs and will provide Pegasus-specific answers to questions without needing any specific prompting. The core tech for that was also silly easy (implemented in a couple hours).

Talking with my Pegasus-specific chatbot in Slack.

This functionality has been integrated into Community Keeper, which is now an AI knowledge-base/Slack archive/chatbot Frankenstein project that is working quite well for the Pegasus community. I’m still unsure whether there’s a more general product in here, but am exploring that question while building and learning. It seems like the market will saturate quickly, and these things will be a commodity soon—so I’ll either need to be very fast (maybe already too late?) or very good (still possible?).

Finally, these advances in LLMs have put Pegasus in a unique situation.

In the short term, Pegasus being a tool that lets anyone take their 5-lines of AI magic and quickly turn it into a business has been a great “shovel in a gold rush” moment—especially since most of the leading-edge AI stuff is happening in Python. Pegasus has had a very good start to the year, and I suspect it is heavily driven by new AI projects.

In the long term, the extent to which AI can replace the need for code boilerplates as an entire category remains unclear. I think it’s certainly possible that Pegasus could get destroyed by some new codebase-generating-thing built on top of LLMs. It’s also possible that it won’t. And finally it’s possible that Pegasus will become that thing—which is another reason to invest in learning as much about LLMs as I can.

Writing

Writing was a big goal for the year, though getting into LLMs has derailed it a bit.²

I published Quit Monkey, Grit Monkey at the end of February and—though I thought it was pretty good—it didn’t get any real attention apart from a handful of dedicated readers. Ideally, the fact that I was proud of it should have been enough, but it’s hard not to be disappointed when you put effort into something and the world shrugs. Creating is hard.

I’ve drafted one other writeup that I’ll hopefully publish in the next month—a technical post on how Pegasus is put together. But it’s possible my adventures in AI might detract from my writing ambitions moving forwards.

YouTube

I’ve been documenting a lot of my AI exploration on my YouTube Channel, which continues to grow, albeit at a relatively slow clip. I had a goal of getting to 1,000 Youtube subscribers by the end of the year, and I’m on pace to achieve that.

My YouTube subscriber growth this year. It’s gone up a bit faster since I started doing more with AI.

YouTube screencasts have become a low-stakes way to quickly document and publish my technical work, and also serve as an indirect marketing channel for me/Pegasus. YouTube streams are also a bit more AI-resilient than writing—at least for a few more months/years till the AIs can do that too!

How I’m feeling

I’m currently more motivated to work than I can remember being in years. Pegasus continues to grow. And between learning about LLMs and trying to get better at video streams I have plenty of learning and growth.

At the same time, I do have this incredible feeling of uncertainty around the future—around the impact of AI on Pegasus, software developers, and…well…the entire world, really. I can quickly jump from existential dread to incredible optimism depending on what I happen to read. It’s quite a time to be alive and in tech!

Recommendations

A couple people mentioned missing my content recommendations as part of my retros, so here’s a few of my favorite things from the past few months.

• What’s Our Problem (book). Tim Urban has produced a masterpiece about societies and how we’ve gotten all messed up in the last decade. Should be required reading for everyone to participate in online discourse. You can read a preview chapter here.
• Gurwinder (substack). Writing about interesting modern psychological phenomena. My favorites include The Perils of Audience Capture, and Why Smart People Believe Stupid Things.
• The New Gurus (podcast). Documenting the rise of online gurus in various random spaces (crypto, productivity, health, etc.).
• Sold a Story (podcast). How a combination of well-intentioned wrong people and corporate interests managed to totally screw up how kids were taught to read for decades.
• Exhalation (book of short stories). I’m a bit late on this one, but Ted Chiang’s short stories are incredible. Kind of like Black Mirror for writing—each one exploring a creative, novel idea, with great storytelling and characters. On the topic of AI, I also recommend Chiang’s ChatGPT Is a Blurry JPEG of the Web in the New Yorker.
• Simon Willison’s blog. Simon has been producing some of my favorite content on AI from a developer’s perspective—and his opinions have largely informed/aligned with my own. He also happens to be one of the creators of Django, which gives him plenty of extra cred.

That’s it for now! I haven’t decided whether this was a one-time thing, or something I’ll do on a more regular basis.

If you appreciated reading this and want me to do more of them, send me a quick email, a like on Twitter, note by carrier pigeon, etc. so I know that it’s worth doing. It’s the only way I know that I’m not shouting into the void.

Notes