There was a period when “bespoke software” felt like a quaint idea — a luxury for large enterprises, unnecessary for anyone who could assemble a stack of SaaS tools. Then came the AI wave that was supposed to replace custom development entirely. Neither of these things turned out to be true.
Here’s what I’m actually seeing in the market and why I think the pendulum is swinging back toward purpose-built systems.
The SaaS Stack Problem
The mid-sized business that assembled 12-15 SaaS tools a few years ago is now looking at:
- $8,000-$25,000 per month in SaaS subscription fees
- Significant staff time maintaining integrations between systems
- Data scattered across platforms with no unified view
- Workflows that work around tool limitations rather than fitting the business
- Increasing prices every renewal cycle
The promise of SaaS was simplicity. The reality for many businesses is complexity — but the complexity of managing multiple misaligned tools rather than the complexity of one well-built system.
What the AI Wave Got Wrong
The narrative was: AI code generation would make custom software so cheap to build that it would proliferate. In some sense this is happening, but not in the way the narrative predicted.
What’s actually happening: experienced developers with AI tooling are building custom software faster than before. The reduced cost has made custom software accessible to businesses that couldn’t previously justify it.
What didn’t happen: AI tools replacing experienced developers. Building software that works reliably in production for your specific business still requires expertise. AI generates code. Expertise builds systems.
The businesses that tried to build software by pointing AI tools at their requirements documents without experienced developer involvement have largely had poor experiences. The AI part was fine. The missing part was the judgment.
The Specific Problems SaaS Can’t Solve
I keep seeing the same patterns in businesses that come to me after trying to assemble SaaS solutions:
Process that doesn’t match any available template. Your business does something specific that no SaaS product models correctly. You’ve spent years configuring workarounds. The workarounds have become load-bearing and nobody fully understands them.
Data ownership and portability. When your operations data lives in five different SaaS products, you don’t really own it. Switching a single tool means migrating data. Switching all of them is nearly impossible. And when any one of them changes pricing or stops offering the features you depend on, you’re at their mercy.
Integration reliability. Zapier and Make can connect SaaS tools, but when an integration breaks — and they break — you don’t necessarily know until you find data that didn’t sync. Mission-critical workflows shouldn’t depend on third-party integration middleware.
Performance on your data volume. SaaS products are designed for a median customer. If your data volume, transaction rate, or reporting complexity exceeds the median, you pay premium prices for mediocre performance.
The New Economics of Bespoke Software
The cost of custom software has come down materially in the last two years. AI-assisted development means an experienced developer can deliver more in less time. Projects that would have been $80,000 three years ago might be $55,000 today.
Meanwhile, SaaS costs have gone up. Price increases at renewal are common. Tier limitations force expensive upgrades. Total cost of ownership for a SaaS stack has risen while the fit-to-business often hasn’t improved.
The break-even between “configure a stack of SaaS tools” and “build custom software” has shifted. For more businesses than before, custom is now the better economic choice over a 3-5 year horizon.
What Bespoke Actually Means
Bespoke software isn’t about writing every line of code from scratch. It’s about building systems that model your business accurately and specifically.
This often includes:
- Off-the-shelf components where they fit (authentication libraries, payment processing, email delivery)
- Custom business logic for the parts that are specific to you
- Integrations built to your requirements, not the tool’s capabilities
- A data model that matches your domain, not a generic template
The “bespoke” part is the fit, not the absence of leverage. Using Rails as the foundation isn’t the opposite of bespoke — it’s using the right tool as the foundation while building the specific thing your business needs on top of it.
The Pattern That’s Working
The businesses getting the most value from custom software in 2025:
- Started with clear requirements — they knew what problem they were solving
- Used an experienced developer who understood the domain
- Invested in discovery before development
- Built the right scope initially (not everything at once)
- Maintained and extended the system as the business evolved
This isn’t exotic. It’s just good software development practice. And the businesses doing it are getting competitive advantages that their SaaS-constrained competitors can’t easily replicate.
Let’s talk about whether bespoke software makes sense for your situation. I’ll give you an honest assessment, not a sales pitch.