Wave nears it’s next phase.

The secret to success on Skool.

And more powerful AIs may be around the corner.

Topics for this week:

• Wave MVP

• n8n Video Starter Kit

• Skool Success

• AI Rankings

• Secret Models

Wave MVP

Wave crossed a huge milestone a few weeks ago now.

It made it to the App Store.

But there’s a few things left to take care of before officially calling the MVP complete.

A couple minor tweaks to existing functionality to maximum user impressions, and a few outstanding visual bugs need to be squashed for Wave to reach the next major goal.

MVP Complete.

Honestly at this point, it is well past a Minimum Viable Product.

I heard on the My First Million podcast the other day that DoorDash started as a website with 8 pdfs.

For contrast, Wave has:

• The “Create a Plan” system

• Push notifications

• Direct realtime messaging

• TikTok style video reel

• automatic video compression to optimize video

• social posts with image options

• a “like” system for posts and video

• A Stories feature

• a reporting system to flag inappropriate content

• a block user system

• and AI content moderation

Just to name a few.

But next week should hopefully mark the end of the MVP scope for iOS.

Then Google Play comes next.

Apologies to my fellow Android users.

Lots of work this weekend, and with it, the next big chapter will be closed!

Also this was a really great interview with the creator of NextJS and Vercel:

Video Starter Kit

Since AI Agents are all the rage, n8n offers a “self-hosted AI starter kit” right on their website, which is a really valuable tool.

Downloading their repo, installing Docker, and running a couple commands gets you a powerful setup for an AI agent on your machine, completely for free.

I’m not completely sold on custom AI agents through automation platforms, I’ll trust the pros to handle that functionality, and so far it seems fairly promising with OpenAI’s Agent Mode.

But what I thought would be really helpful is a video starter kit for n8n instead.

Basically all the things I’ve been making videos about lately, wrapped into one starter project that can be downloaded and executed in a single package.

The video is really focused on running n8n in Docker because I’ve been meaning to cover that, but seemed like a good excuse to introduce the video kit.

Thanks to a post from fellow automator Stephen G. Pope, I was inspired to “build in public” again.

So the starter kit is very basic at this point.

It has ffmpeg and Whisper installed, but the only thing it can do is transcribe video to text.

The next few steps will be adding options to return timestamps for segments and for words, and adding a basic method to access ffmpeg with a custom command.

But for now the name is appropriate.

It is certainly a video starter kit.

Skool Success

The current top earner in Skool, Nick Saraev, posted a video and breakdown about how he operates his community, outlining some of the strategies that have led to his success.

The single biggest piece of advice he gave was, unsurprisingly, consistency.

He pointed out his daily Skool activities, which are conveniently gamified in your profile, average well over 200 per day, while most community managers average 15 activities per day.

He reasoned that was a factor in his success just because he’s had 10 to 15x more connections than the average.

On top of that, he is very efficient with creating high quality videos and explains things well and offering actionable, direct advice, so there’s many other factors contributing to his success but the Skool activities seems like an obvious baseline.

So I’m making that part of my daily routine to track Skool activities from the day before and try to work my way up to a high average.

20 activities per day is a reasonable goal for now, and I’ll increase week by week as the community grows.

Nearly 50 members in my community after 2 weeks, and activity count should also get much easier as the membership grows but those activities will be in other communities until mine is a little busier.

So far so good on the Skool Activities.

Hundreds more to go!

AI Rankings

It’s been a few weeks since Grok 4 released and it finally made it’s way to the leaderboard.

Grok ended up settling at rank 3 for Text, tied with gpt-4.5.

I’m always on the lookout for the next best performer, and technically that rank is for the “Text” category, not specifically the WebDev category but it seems like a good overall indicator of performance.

Google remains at the top and remains my primary coding tool, but to be fair I haven’t really tested Grok with code lately.

I’ve been in a rhythm with Gemini, and with Gemini having a 1 million token context, I’ve gotten used to carelessly dumping 150k chunks of code in to give it all the files that might be involved in a change, rather than having to work with smaller context windows or one file at a time.

But I’m also starting to wonder at this point, if the models will all basically be at about the same level as far as humans can perceive.

Three different senior engineers with 20 years of experience will work differently, but each will most likely deliver their own perfectly viable solution.

Until an AI starts generating feature complete, functioning apps in a few prompts, any of the frontier models code could be equally usable.

So I’m sticking with Gemini for now mostly out of habit, but we’ll see.

As we approach fall and end of quarters, I have a feeling new model announcements will be right around the corner to push the threshold up further.

Secret Models

Speaking of new models, there are potentially 3 new hotly anticipated models from OpenAI that may be available soon.

GPT-5 has been mentioned a few times by Sam on various podcasts.

There’s also been mention of an open source model, which hopefully will be something on par with DeepSeek, Gwen and a few of the other open source leaders.

And just this week, “o3 alpha” was spotted in Web Arena testing options along with another model code named “starfish”.

Both claimed to be OpenAI related.

But from those that had access, o3 alpha was producing some impressive results.

Really complicated outputs from a single prompt.

The fishing rpg example is particularly striking, a full dashboard interface versus a small box with a single button.

I’ll be anxious to see how soon any of these models release and to see which one is impressively outperforming the current top spot on the Leaderboard, if it actually is o3 alpha or some other mystery model.

Looking forward to hopefully talking about one of these releases next week 🤞

And that’s it for this week! Wrapping up an MVP, building a video starter kit, and flexing that Skool activity muscle.

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!