Where I'm Going Next

The Epoch Moment

A lot of things happened at the same time, but I really think Open Claw was the catalyst for most of it. What Open Claw did was kind of like a four-minute mile. It took a bunch of stuff that already existed and threw it together in a creative way. The result was mind-blowing. Having the heartbeat, having the advanced memory system, having real skills that actually work — it completely changes your perspective on how to use AI.

And then some unfortunate timing forced my hand. Anthropic shut down the ability to use Open Claw with your Claude subscription and forced everyone onto API keys, which are terrible and extremely expensive by comparison. I'd already purchased my Claude Max plan for the month. Didn't have it in the budget to go buy Codex. I'd already been playing around with Claude Code anyway. I got my Max subscription like the day before they shut it down. Maybe the same day. So it forced me to reverse engineer Open Claw.

Now I feel like I have a degree of freedom I wouldn't have had otherwise. If you roll your own, you get the best of both worlds. You can just tell your memory system, "Hey, while you're indexing this file system we structured based on Open Claw, go ahead and add the Claude Code folder too." And it just does. Open Claw's memory system was based on Claude Code's file system to begin with — all he did was reverse engineer it and build on top. So now I'm sitting in between both ecosystems with the freedom to leverage either one.

The Pipeline That Changed Everything

I've got a really nice pipeline now where I can copy a voice transcript, paste it with one tap into a simple interface, and then the pull request opens up minutes later with a hero image and everything. Totally formatted, cleaned up. I read over them, but it's pretty evident that I don't have to. I never insist on changes. It's always in my voice. There's never anything stupid.

When you look at the editing process, the drafting, image generation, formatting, opening a pull request, creating TypeScript files — how much of that actually needs someone who's novel and visionary? That's all routine work. So that was one of the first things I automated. Now all I have to do is talk into a voice memo app for 20 or 30 minutes and the rest just happens behind the scenes. I spit out my thoughts, my perspective, my predictions, my recommendations. I give my human perspective while I drive, while I'm cleaning the office, during time that was previously a missed opportunity. I spit my humanity into a microphone for 20 or 30 minutes, and what comes out the other side is a pull request with a blog draft that I can review, approve, and deploy.

Instead of being the grunt at the company who's too busy with routine work to tackle the hard problems, I get to be the advanced problem solver. I come in and say, "Hey, Gus, what's your hardest problem right now?" And I just tackle that while SEO audits run in the background, blog publishing runs in the background.

What's Next: SEO End-to-End and GEO

Just like I did with the blog articles, I need to follow the same pattern with the SEO audit pipeline and take it all the way to opening pull requests, not just creating issues with reports. Right now I get an email with links to issue tickets where I've done SEO scans on all the sites I'm tracking. What I need is for that to go all the way and actually open a pull request with the recommended changes. That's really powerful because at that point, my SEO game is on autopilot.

I also need to do research into GEO — generative engine optimization. Making sure I'm optimizing for generative search, not just traditional search. I'd imagine it probably repeats a lot of the same patterns, and we just plug it in with the SEO audit. But that's going to be a real differentiator. If I can offer really cheap, really good SEO with maintenance, and throw GEO on top of that with an automation pipeline? That moves the needle forward in a big way.

The Unified Dashboard

When the blog pipeline opens a pull request, I added a feature where it sends me Telegram messages with links and pre-written text appropriate for each social channel — X, Facebook, LinkedIn. The header images render perfectly, the title renders perfectly, all the social meta is there. What I want to do next is have it actually draft posts on my social media accounts so all I have to do is approve them, just like approving a pull request deploys a blog post.

That plugs into the bigger vision: a CyberWorld admin dashboard that's a unified place for everything related to the website. Different dashboards and widgets so I don't have to bounce between Google Analytics, Facebook, and a dozen ad campaign dashboards. One place that lets me drill down into each individual element, but also has a smart home dashboard that's dynamic. It determines my highest-priority items and puts those widgets at the top. It shows me the data, gives me a description of what it means, and makes recommendations on what needs to be done. It knows where I need to go to perform the action. If someone commented on a post, it recommends that I respond, gives me a link straight to it, maybe suggests some things to say. But it leaves the actual decision to me.

The beauty in all of this is that I don't have to program these things like a machine every time I want to do something. I can have a conversation, like I would with a human, and it goes into the memory system, and it finds ways to improve itself. That's the power of the heartbeat, the proactive strategy, the memory system, the skill system. They all have a synergy with each other. The gateway, the heartbeat, the memory system, and the skill system all work together in a way that brings your AI assistant to life and makes it feel human in terms of productivity.

CyberWorld's Three-Tier Model

CyberWorld is emerging as a three-tiered service provider. The first tier is digital marketing — SEO audits, content creation pipelines, that kind of thing. The second tier is automations. The third is custom SaaS. But the separation isn't rigid. Part of your SEO is already automated. So the first test case for automation within your company is that we automate your SEO, and we've already got those automations in place.

It's really a funnel of value. While I provide those initial low-risk, simple digital marketing services, I'm already doing discovery. I'm developing relationships with your more technical team members. I'm gathering information about your company. There doesn't have to be a paid discovery period because I'm already learning while I provide the basics.

From there, I can start making proposals for automations. And then I work in simple things like adding a dashboard to track the digital marketing offerings we've implemented. That dashboard? That's the seed of a custom SaaS product. All we do is keep iterating on it. One thing leads to another. You're immediately getting value, and we're getting to know you. The more value we provide for you, the more value you can provide for your customers, and everyone compounds into a better situation.

Before you know it, you're going to have automated everything that makes sense to automate. You're going to be canceling subscriptions left and right. Because we're going to hear your griping about that bloated CMS everyone hates, the one with all these features you never use, the one you have to spend all this time training new people on. And I'm going to say, look, we need to get rid of that thing. I'll show you how I automated my blog publishing pipeline. No CMS, no database, just a human spitting raw knowledge into a microphone for 20 minutes, then a review and approve. That's not just increasing efficiency. It's eliminating a component that should not exist. It's deleting a part and simplifying everything. The CMS is the first obvious thing, and then we go after the CRM the same way.

SaaS Is Dying. This Is What Replaces It.

For almost 10 years, there was this gold rush in the SaaS startup market where VCs were throwing insane amounts of money at founders to build the next app that was going to blow up. Everybody was chasing that. People like me got the opportunity to learn really advanced coding skills and the freedom to play around with all the latest technology. I spent half my career building SaaS and the other half learning AWS, cloud architecture, deployments, security and compliance — the more enterprisy high-availability stuff.

I kind of left digital marketing behind for a long time and went and rode that SaaS gold rush wave. I loved working for all those people. It was just always so temporary. They'd make so many promises — "I want to be your only client, I'm always going to find work for you" — and then drop you on your ass as soon as something fell apart.

Now the SaaS industry is plummeting because it's so easy and cheap to have AI build your MVP. I still get clients who need me to take their vibe-coded MVP and deploy it into high-availability infrastructure, or handle compliance for banking, death care, medical. But this time next year, no one is going to pay you to code SaaS the way they used to. It's going to be companies like mine that already have relationships with your company. We're already handling your landing site, your lead capture, your sales funnels, your search engine optimization, your generative optimization. That gives us the angle to get to know your team and your company, and we're at a particularly advantageous position to offer solutions to your problems.

The Pace Right Now Is Unreal

I sat down in one day yesterday and built out that whole blog publishing pipeline. I almost built out the entire SEO pipeline to full automation — it's one step away. I built the Upwork scraper that analyzes jobs. I've already sent in a proposal for a great job I never would have found without a bot that learned my history and where I want to go with my company. The chatbot on the CyberWorld site is actually plugged into Gus through a secure Cloudflare tunnel. I ran into budget issues with the API keys, so I just asked, can we hook it up to Gus and let Gus handle this? We had a real serious talk about security, put guardrails around prompt injection attacks, hardened the thing. In less than an hour I had my chatbot plugged into my actual AI assistant in a secure way. It all works.

I did months of work yesterday. To completion. Work that would have taken me months to do poorly, I did great, in a day, and it's done and working.

If things keep going the way they're going, every week I'm going to be launching multiple products I've been wanting to get out for years. I'm already thinking about revisiting music generation AI collaboration projects. It won't be long before I've blown through all the pure software things I've always wanted to build, and I'm going to have so much time and budget that I'll start doing hands-on things with IoT and robotics.

I don't think shipping software is going to look the same in three to six months. Any small company willing to take their smartest, most techie employee and say, "Take a day or two of your week and just go do cool things with AI" — those companies are going to be the winners. And that person won't have to stay if you don't treat them right. But they will if you pay them well, show appreciation, and give them a piece of the pie.

This is an epoch moment. Everything changes from here.