The 200-Hour SaaS Build: A Real Workflow Breakdown

Steve Powell wanted to run an experiment.

Could he build something useful, in a language he didn’t know, without opening a development environment?

On a recent episode of SaaS That App with hosts Aaron Marchbanks and Justin Edwards, Steve Powell, one of the owners of Delta Systems, walked through how that experiment turned into myeaglecoach.com: a functional SaaS application built in under 200 hours with Claude, Claude Design, React, Laravel, and a lot of curiosity.

It wasn’t a toy project. It wasn’t a throwaway demo. It was a real product built to solve a real problem for Scouts, parents, and Eagle Scout coaches drowning in PDFs, lost files, broken edits, and endless review loops.

 

The Founder Experiment

Steve’s goal was simple: build like a founder.

Not like a developer with deep stack knowledge. Not like someone spinning up the perfect local environment. Not like someone who already knew the frameworks inside and out.

“I wanna build something that I don’t know how to build in a language that I don’t even know,” he said. “And I don’t even wanna open a development environment.”

That became the test.

Steve had ideas he had been carrying around for years. Once he started using AI tools more seriously, those ideas finally had somewhere to go. Claude became a thinking partner, a design assistant, a workflow builder, and eventually a way to turn one very specific problem into a working application.

 

The Real Problem Wasn’t Code. It Was PDFs.

The idea for myeaglecoach.com came from Steve’s work in Scouting America.

He was helping 11 Scouts work toward Eagle Scout, a process that can take months and requires a lot of planning, communication, approvals, signatures, and documentation. The problem is that much of the process still depends on PDFs.

Scouts download forms. They fill them out in a browser. They forget to save them. They overwrite completed versions with blank ones. They send PDFs back and forth. Coaches add comments. Someone needs to add photos. Someone else needs a signature. The file gets flattened, locked, lost, duplicated, or broken.

For Steve, the pain was obvious because he was living inside it.

Every proposal review took 15 to 25 minutes. He was checking the same kinds of things every time: missing information, inconsistent formatting, typos, incomplete tool lists, unclear project descriptions, missing safety details, and whether the Scout had explained the work in the right way.

That repetition became the first unlock.

 

From 25 Minutes to 12 Seconds

Before building the full app, Steve trained Claude to review Eagle Scout proposal PDFs.

He gave it the rules he used as a coach. Be specific. Use numbered feedback. Stay positive. Remember these are teenagers. Look for missing details. Flag formatting issues. Check whether the proposal includes things like a first aid kit, a map to the nearest hospital, and clear language around leading volunteers.

Eventually, the workflow became simple: upload the PDF, ask Claude to review it, copy the generated feedback, and send it back to the Scout.

Something that used to take 20 to 25 minutes took about 12 seconds.

That changed the entire coaching process. Instead of reviewing one PDF after a work session and following up later, Steve could review documents while standing in the kitchen talking to parents. Scouts could get feedback immediately, make edits, and keep moving.

That was the moment a repeatable AI workflow started looking like software.

 

A Skill Is Useful. A Product Is Different.

Steve could have stopped there.

He had a Claude workflow that worked. He could review proposals quickly. He could reuse the same instructions. He could even share the skill with other power users.

But that still wasn’t a real product.

As Steve pointed out, a Claude skill only helps people who already know how to use Claude. It requires the user to understand the environment, upload the right files, run the right review, and manage the workflow manually.

That’s not how most people want to work.

So Steve took the next step: turn the workflow into an actual application.

That meant thinking through screens, not just prompts. It meant mapping data back to multi-page PDFs. It meant handling imports and exports. It meant supporting digital signatures. It meant understanding how Scouts, parents, and coaches would move through the process in real life.

The AI could help build the product, but Steve still had to shape the experience.

 

Claude Design Became the UI Partner

Steve used Claude Design to build out the interface and workflow for myeaglecoach.com.

He already knew what the process should feel like because he understood the PDF workflow and the people using it. But Claude Design helped him turn that understanding into mockups, screens, and user flows.

Sometimes it suggested things he hadn’t considered. For example, Steve realized he didn’t necessarily need traditional passwords. Claude could help add Google login and Apple login. It could generate flows, revise screens, and help move the design closer to something usable.

Then came the wild part: Claude Design could hand work over to Claude Code.

As Aaron pointed out during the episode, that handoff button had not even existed a few weeks earlier. The tools were changing while Steve was building with them.

That was part of the experience: the stack was evolving in real time.

 

Under 200 Hours to a Functional Product

Steve built MyEagleCoach in under 200 hours.

At around 100 hours, it was already usable. By under 200 hours, it had become a feature-rich system with real Scouts using it.

The app can import completed Eagle applications. It can bring in Scoutbook data. It can import the 32-page Eagle Scout workbook. It can repopulate official PDFs and export them back out. It supports digital signatures. It helps Scouts move through the process without getting trapped inside broken local files.

One moment stood out.

A 14-year-old Scout needed to sign part of the document. She tried signing with a mouse, didn’t like the result, saw a QR code, scanned it, and signed from her phone. Steve barely had to explain anything.

That mattered.

“If we’re doing our job correctly, usability just comes with good design,” he said.

That may be the clearest lesson from the entire build. AI can help you ship faster, but speed only matters if the product makes sense to the people using it.

 

AI Still Needs Adult Supervision

The episode wasn’t just a celebration of AI tools.

Steve was clear about the parts that still require judgment.

He chose PostgreSQL, Aurora Serverless, and Laravel because he understood the trade-offs. He was cautious about all-in-one AI app builders and wrappers where you don’t know what’s happening behind the scenes. His concern was not whether they could build something, but whether you understood the expense, security, and architecture behind the green curtain.

That’s especially important when user data is involved.

As Justin put it, getting from a chat window to a deployed app with user data is sometimes fine, and sometimes very much not fine. Founders need to understand where their data lives and how it is secured.

Steve also ran into the familiar problem of AI forgetting what had changed. Claude Design generated architecture diagrams that included tools the project no longer used, like DocuSeal and Redis, because it found old specs and treated them as if they still mattered.

Claude could move fast, but it could also carry old decisions forward unless Steve corrected it.

His description was hard to beat: sometimes the AI feels like “a junior developer that’s had too much Baja Blast.”

 

The New SaaS Advantage

The bigger theme of the episode was not that anyone can magically build perfect software now.

It was that the distance between having an idea and testing a working product has collapsed.

Steve had a specific problem. He understood the users. He understood the workflow. He had enough technical judgment to make good decisions. AI gave him leverage.

That combination matters.

The future of SaaS may not belong only to the teams with the biggest budgets or the most engineers. It may also belong to people close enough to a problem to understand it deeply and curious enough to build the first version themselves.

As Justin put it, if you have a vision, a little expertise, initiative, and curiosity, you can do “crazy stuff” right now.

 

Steve’s Background

Steve Powell is the co-owner of Delta Systems. A longtime programmer, entrepreneur, angel investor, and startup advisor, Steve has spent his career helping companies design and build intuitive, powerful, and scalable applications. His expertise spans system architecture, database design, SQL optimization, UI, UX, QA, team management, and breaking systems before users do. Through Delta Systems, Steve and his team support heavily trafficked websites, eCommerce platforms, and business-critical systems that process hundreds of millions of dollars in online sales every year.

Listen to the new episode on: