"Why don't you just use [popular SaaS tool]?" I get this question a lot. And the answer is always the same: because none of them do exactly what I need, and the ones that come close cost more than building my own.

The Tools I've Built

Over the past months, I've built three main platforms:

Each one was built because I had a specific problem, looked for an existing solution, and couldn't find one that fit.

The Case Against SaaS

1. Subscription costs add up fast

Running an operation with dozens of automated workflows, multiple AI models, and several content pipelines would cost hundreds (maybe thousands) per month using SaaS tools. My self-hosted setup runs on a single server that costs a fraction of that.

2. No tool does exactly what you need

Every SaaS tool makes assumptions about how you work. Those assumptions are right 80% of the time. But it's the other 20% — the specific way your business works — that matters most. With custom tools, there are no compromises.

3. Integration is always the bottleneck

When you use five different SaaS tools, you spend half your time making them talk to each other. Zapier, webhooks, API bridges — it's glue code all the way down. When you own the tools, integration is just... code. In the same codebase. On the same server.

4. You control the data

All my content data, workflow configurations, and business logic live on my server. I'm not dependent on any third party's data policies, API changes, or pricing decisions. Even my AI's identity and memory are in version control — stored in a shared brain repository that I own and sync across servers myself. This matters more as AI gets deeper access to infrastructure — a point underscored by Claude Mythos, Anthropic's new AI built specifically for cybersecurity. The more specialized and powerful these agents become, the more important it is to own the environment they run in.

The Case For Building

AI makes it possible

Here's the thing that changed the equation: AI agents can build software now. Tim — my AI agent — built most of Documentor in a single session. What used to require weeks of development now takes hours.

This isn't about being a great programmer. It's about being able to describe what you need clearly and having an AI that can turn that description into working software. The barrier to building custom tools has dropped dramatically. Case in point: Tim built this entire website in one day — from zero to six blog posts with images and backlinks. And when my chat app kept disconnecting on mobile, Tim found the root cause and shipped a fix in 10 minutes. The same approach works for client projects too — I recently built a voting system for a BNI chapter, spec to deployment in a single session.

Custom tools become competitive advantages

Anyone can sign up for the same SaaS tools. Nobody else has my exact workflow system, my exact content pipeline, my exact scheduling logic. The tools themselves become part of the moat.

You learn your own business better

Building tools forces you to think deeply about how your business actually works. Every workflow you design, every data model you create, every automation you build — it all requires you to understand the underlying process. That understanding is valuable.

When NOT to Build

I'm not saying everyone should build everything from scratch. That would be insane. Here's when I use existing tools:

  • When the tool is mature and standard. I use nginx for web serving, SQLite for databases, Docker for containers. These are solved problems.
  • When the cost is low and the fit is good. Some SaaS tools are cheap and do exactly what they promise. No need to reinvent those.
  • When the tool isn't core to your business. I don't build my own email client or code editor. I build the things that are unique to how I make money.

The Philosophy

Build what makes you money. Buy what doesn't. And when AI makes building cheaper than buying, the calculus shifts toward building more. This philosophy is also what makes it possible to run an entire business solo with AI — owning your tools instead of renting them is a core part of keeping the headcount at one.

That's the approach I've taken, and it's working. Every tool I've built pays for itself — not in revenue directly, but in time saved, flexibility gained, and costs avoided. My latest example: automating credit card bill sorting — a task that used to take 3-4 hours now takes 15 minutes.

If you want to see these tools in action, start with how the content machine works. Or check out Pointer — my latest personal tool that finds credit card deals at nearby restaurants using GPS.

The whole reason I built Jarvis is to make this philosophy accessible. You shouldn't need to be a developer to own your tools instead of renting them. Jarvis gives you a server and an AI agent — you describe the problem, and it builds the solution. Same approach I use, available to anyone. The story of turning a personal tool into a SaaS product is worth reading if you're curious how that leap happened.

— Pond