Most growing companies are held together by spreadsheets that nobody fully understands — built by someone who left three jobs ago, maintained by someone who doesn't know why it exists, and quietly critical to daily operations. In this episode, Jeff Mains sits down with Garrett Fritz, co-founder of MetaCTO, a fractional CTO firm that helps mid-market companies transform outdated operational processes into custom, scalable software.

Garrett breaks down why so many organizations are trapped in the "if it ain't broke, don't fix it" mindset, how AI has lowered the barrier to custom software without eliminating the need for expertise, and when it actually makes sense to build your own tool versus buying off-the-shelf SaaS. He also shares how internal tools can evolve into white-labeled revenue generators — and the most common mistake founders make when they try to take that leap too fast.

Whether you're drowning in manual processes, questioning your SaaS spend, or wondering how to implement AI responsibly, this episode delivers a practical, no-hype roadmap.

Key Takeaways

4:37 — **The #1 operational inefficiency Garrett sees:** Hundreds or thousands of employees running mission-critical operations on a spreadsheet built a decade ago by someone who's since been promoted — and nobody knows why it has the formulas it has.

6:15 — **What "turning spreadsheets into apps" actually means:** MetaCTO embeds in the business, decodes the spreadsheets, understands the workflows, and builds working software that can replace the internal process — or be taken to market as a SaaS product.

7:54 — **Profitable from day one:** Because Garrett and his partner came with a thick Rolodex from 15–20 years in tech leadership, MetaCTO launched with clients already lined up — no burning cash to find product-market fit.

13:27 — **70% of AI POCs never see the light of day:** The excitement dies when teams realize how much effort is involved. MetaCTO's focus is getting those 90%-done prototypes all the way to the finish line. 18:34 — **Build custom vs. buy SaaS — the real decision framework:** After 2–4 weeks embedded in a business, MetaCTO looks at licensing costs, actual feature utilization (often just 2% of the SaaS product), man-hours wasted, and growth trajectory to determine the ROI break-even point.

28:25 — **Niches win:** SaaS isn't dead — it's narrowing. The companies gaining ground are building hyper-specific tools for specific industries (think: Procore, but only for commercial plumbers) where the UI, reports, and workflows are built around exactly how that niche operates.

31:33 — **The #1 mistake when productizing internal software:** Not talking to the second customer. Your problems aren't always everyone else's problems. Validate outside your organization before building for market, or you risk six months of rework when the deltas turn out to be core to the platform.

33:40 — **How to actually quantify the ROI of custom software:** Bake usage analytics into every product from day one. Track utilization, time on platform, transactions processed, and revenue generated — then compare to the man-hour cost baseline captured during discovery.

39:14 — **Responsible AI implementation starts with one rule: Resist "Accept All."** Don't grant admin tokens to AI agents for convenience. Suffer through permissions early so you don't face irreparable reputation or business damage when a bad actor exploits an over-permissioned agent.

41:22 — **The smartest first step for any leader feeling stuck:** Use AI tools like Replit to build a prototype with fake data. Don't try to connect it to real systems — just use it to force yourself through the problem-solving process. Come to the conversation with a working wireframe and you'll skip weeks of expensive discovery.

Tweetable Quotes

At the heart of it is some Excel spreadsheet that some employee made 10 years ago — and it is critical to the operation." — Garrett Fritz

"70% of AI proof of concept projects have never seen the light of day. It's pretty common to get excited about something and then realize, oh, this is a lot more effort than we thought." — Garrett Fritz

"You can't just give a layman a chainsaw and expect to be a carpenter. A little bit of finesse and experience goes a long way." — Garrett Fritz

"The niches win. The companies gaining ground are building hyper-specific tools for specific industries — where the UI, reports, and workflows are built around exactly how that niche operates." — Garrett Fritz

"We never build it and run away. And as you can imagine, anyone who's created a piece of software has never said 'I'm done' either." — Garrett Fritz

"Resist 'Accept All.' Give the AI admin access for convenience, and you're one bad actor away from irreparable damage to your business." — Garrett Fritz

"AI is most valuable when it's applied to real business friction — not just trendy experiments or chatbots. Nobody needs another one of those." — Jeff Mains

SaaS Leadership Lessons

1. Familiarity is the enemy of efficiency. The "if it ain't broke, don't fix it" mentality keeps organizations locked in spreadsheet-driven operations for years — sometimes decades. The pain point has to get big enough to justify change, but by then the cost of switching is enormous. Don't wait for a crisis to modernize.

2. The barrier to custom software has dropped — but expertise still matters. AI tools like Replit and Lovable have made it possible for non-developers to prototype software. But there's a massive gap between a 90%-done prototype and a production-ready, secure, maintainable application. Knowing what you're doing still matters.

3. Don't buy features you'll never use. Most enterprise SaaS customers use 2% of the product's functionality — but pay for 100% of the license. When your team is only using 2% of the product and only 50% of the people who should be using it actually are, you're compounding inefficiency at every layer.

4. Build for the second customer before you build for the market. If you think your internal tool has market potential, validate it with people outside your organization before investing further. Your problems are not automatically everyone else's problems. The cost of discovering core delta requirements after six months of development is enormous.

5. Measure everything from day one. Custom software that doesn't have baked-in usage analytics is a black box. You can't demonstrate ROI, you can't justify ongoing investment, and you can't make intelligent roadmap decisions. Instrument every product with utilization metrics, transaction data, and performance monitoring from the start.

6. AI governance isn't optional — it's the first conversation. The most dangerous thing you can do is grant your AI agents broad permissions during development and never revisit it. Treat AI like a junior employee: define its scope, limit its access, and require human approval for anything with downstream consequences. Someone always has to be the final buck.

Guest Resources

garrett@metacto.com

https://metacto.com/

https://www.linkedin.com/in/grfritz/

Episode Sponsor

The Futureproof Series - https://www.youtube.com/playlist?list=PLfkXKUPZ5xuOqMPR7_gzGybncTtavyR1N

The Captain's Keys

Small Fish, Big Pond – https://smallfishbigpond.com/ Use the promo code ‘SaaSFuel’

Champion Leadership Group – https://championleadership.com/

SaaS Fuel Resources

Website - https://championleadership.com/

Jeff Mains on LinkedIn - https://www.linkedin.com/in/jeffkmains/

Twitter - https://twitter.com/jeffkmains

Facebook - https://www.facebook.com/thesaasguy/

Instagram - https://instagram.com/jeffkmains