

Jay Long
Software Engineer & Founder
Published March 8, 2026
Gus has an authenticated Upwork API client. It can perform all kinds of tasks on the platform. I did an initial run that gave it basic functionality, then went back and added more capabilities because I had to add scopes to the token.
The daily Upwork task works like this: it has filters based on conversations I've had with Gus about where I've been, what I've done, what my experience is, what I'm good at. Just honest, open discussions. What I love doing, where I'm interested in going. He's also got the deep strategic company plans in memory from all our sessions. Factoring all of that in, he helps me come up with filters and searches job postings on the platform.
I didn't realize the power of some of the things we did while building it. During setup and testing, Gus proactively suggested changes to my profile. I told him to look at my profile, but he came back with specific recommendations: swap some tags around, update some phrasing, take things that were buried in the back and front them. Based on everything I'd told him about what I'm good at, what I love doing, where I want to go, and what I've done, he could see that my profile wasn't aligned with any of that.
The result? I used to get garbage spam from Upwork telling me to go look at job postings. They've gotten notably better. I wouldn't say great. That platform is not what it used to be. But better.
Every day around 3 PM, I get a report from Gus with real recommendations. Good, solid recommendations with thought behind them. He explains why he's recommending things and highlights stuff I just don't have time to look up: their funding, their history, their reviews. These are all important things as a freelancer. You don't want a bad client. You're interviewing them as much as they're interviewing you.
One thing the daily report has really highlighted is the gaps in my skills relative to where I want to go. So the next thing I'm adding to the Upwork functionality is tech trend tracking. When you're looking at these job postings, look at the technologies they mention and trend them. There's so much changing so fast that you were always going to miss some of it as a developer. If you were engaged in the community, you'd pick up on trends, but things are moving at a different speed now. You really need an AI to go out there, analyze all the trends, and put in front of you what it suspects you're going to care about.
This is a simple feature that's going to take seconds to add. And that fact is worth sitting with for a minute.
Now that I'm ramping up my ability to orchestrate many agents at once, configure them to run subagents, and set some of them up to be proactive and self-improving, it's an epiphany. You suddenly realize why things are changing so fast: they're changing fast because they can. It doesn't take a billion-dollar company or VC funding. It doesn't take a huge team of human developers. It doesn't take any of that to be involved in fast changes like this.
The market I'm targeting naturally overlaps with people who want help with AI automations. But what I'm really doing is targeting digital marketing gigs so that CyberWorld Builders can position itself as a firm that takes ownership of tech for small companies that are underserviced. Basically, we're providing an IT service.
The engagements vary. Some companies are real mom-and-pop shops with nobody handling tech at all. They need someone to take full ownership and become a tight partnership. Other companies are a little larger, they have a staff, and somebody on that staff is a techie person. That's more of a consulting thing. Part-time work, maybe some sprints to help them make leaps and bounds, with the intention of handing off maintenance to an internal resource.
The upsell is automation services. We'd naturally transition into automating things for them once we understand their operations.
A lot of what I'm doing runs on a dedicated machine. But the trend I'm seeing on Upwork is that people don't want to manage physical machines. They want it on a VPS in the cloud. That's totally reasonable, and it actually makes it a lot easier for a remote resource like me to manage.
I don't know why this didn't occur to me sooner: I should offer that hosting service on AWS. I can take advantage of economies of scale, put this stuff in an ECS cluster, manage many clients on elastic, low-maintenance infrastructure where the maintenance responsibility is outsourced to Amazon. If I have enough clients, ECS becomes worth the cost, and the pricing is actually pretty reasonable. You obviously save money just provisioning an EC2 instance, but ECS gives you the scalability.
A lot of these companies already have complex business operations that depend on pre-existing n8n workflows. What I need to do is either have a specialist agent that can take our automations and put them into n8n so I have a familiar graphical interface to manage everything from a dashboard, or reverse engineer n8n just like I did with OpenClaw so I own the codebase and have full insight into the architecture.
My theory was correct: I really do need to demonstrate competency with n8n. A lot of these companies are going to say "here are my n8n workflows, I need you to automate these," and they're not going to be confident in me if I can't speak to deep experience with the platform. The Upwork research validated the strategy of learning n8n. Not because n8n is the future, but because it's a strategic thing to know to help transition people who are already established on it into what comes next.