Retool’s 2026 Build vs. Buy Report surveyed 817 builders across enterprise organizations.
The headline finding is blunt: 35% of enterprises have already replaced at least one SaaS tool with custom-built software.
Not “plan to.” They have already done it.
The rest of the numbers validate the trend. 78% plan to build more internal tools in 2026. Roughly half are saving 6+ hours per week through AI-assisted building. And 60% of builders have created software outside IT oversight entirely.
Shadow IT is thriving because employees are frustrated.
These aren’t speculative numbers from a vendor with an agenda. They describe a shift that’s happening right now.
The easiest tools to replace go first.
Not all SaaS categories are equally vulnerable. Retool’s data shows clear patterns.
| Category | Replacement rate |
|---|---|
| Workflow automation tools | 35% |
| Internal admin tools | 33% |
| Business intelligence dashboards | 29% |
| Customer support tools | Common |
| Project management tools | Common |
| CRMs | Common |
| Form builders | Common |
Companies are targeting low-hanging fruit. They want out of tools where the alternative is clear and the switching cost is manageable. If a tool is overpriced or underperforming relative to what AI-assisted development can now produce, it’s on the chopping block.
Deeply embedded tools with complex integration webs—enterprise ERP systems, core accounting software—are safe for now. The data migration alone would take months.
What is the SaaStr 90/10 rule?
SaaStr’s practical framework for navigating the replacement wave offers a clear mental model.
Buy 90% of your software off-the-shelf. Do this when adequate solutions exist and the total cost of ownership favors buying.
Build 10% where no adequate solution exists. Or, crucially, where the existing tool has zero AI functionality.
That last condition is the new threshold. SaaStr’s team replaced a paid portal tool in one day using Claude and vibe-coding. They’ve built 12+ vibe-coded apps used over 800,000 times.
The threshold for “build” didn’t change because companies suddenly love building software. It changed because AI reduced the cost of building from months to days.
The threshold for “build” didn’t change because companies suddenly love building software. It changed because AI reduced the cost of building from months to days.
The bar for what counts as an acceptable SaaS product just got infinitely higher.
How do you decide what to replace and what to keep?
The Retool data gives us the macro picture. The practical question is which tools in your specific stack need to go.
Look closely at your vendor’s AI features. Are they meaningful? A chatbot on a help page or generic content suggestions don’t count. If the AI is purely cosmetic, the gap between what modern alternatives can do and what your tool does is only going to widen.
Watch your costs. Per-seat pricing that increases 15–20% annually while your team size stays the same is a glaring signal. You are paying more every year for the exact same value.
Beware of sunk cost in a vendor’s ecosystem. Proprietary data formats that don’t export cleanly are a trap. Templates using a vendor-specific language that no other platform recognizes create artificial lock-in.
Lock-in is a balance-sheet liability. It compounds with every renewal.
But you shouldn’t rip everything out.
If a tool is deeply embedded in your daily operations, stability matters. When processes are built around specific features and the switching cost would disrupt operations for weeks, the replacement calculus changes.
Do the math. If it costs $50,000 to migrate and you save $20,000 a year, your payback is 2.5 years. If the payback stretches to 4+ years, the disruption probably isn’t worth it.
Some SaaS vendors are genuinely innovating. If your vendor’s AI actually works with your data, operates across workflows, and delivers measurable value, they might be adapting fast enough to justify staying.
Maintenance costs are the silent killer of DIY apps.
The “build everything with AI” narrative skips an uncomfortable truth. Maintenance costs are real.
SaaStr reports spending 30–60 minutes daily maintaining each of their production vibe-coded apps. For a team that’s replaced five SaaS tools, that is up to 25 hours of maintenance every single week.
The build-vs-buy decision is about the total cost of ownership over several years.
DIY builds are fast to create but demand ongoing attention. They work well for simple internal workflows. They fail spectacularly for customer-facing applications or systems where downtime costs actual money.
The sweet spot is clear.
Use purpose-built platforms for your core infrastructure. A modern website on production-grade hosting doesn’t need daily maintenance. Rely on AI-assisted building strictly for the unique internal workflows where off-the-shelf solutions fall short.
Start where the ROI is obvious.
If you’re convinced you need to replace some tools, you have to start somewhere.
Most companies don’t even know their total SaaS spend. List every tool, its annual cost, its renewal date, and its actual utility. The final number is always a shock.
Find your worst value-to-cost ratio. It’s usually a platform where you use 20% of the features but pay for 100%. Or it’s a legacy tool that hikes prices annually while modern alternatives cost a fraction of the price.
For most mid-market companies, the website is the easiest, highest-ROI replacement.
Proven open-source frameworks like Astro and Next.js, paired with a headless CMS, offer a well-understood migration path. The cost savings are immediate. You often save $30,000–$40,000 a year in platform fees alone. The new site is faster, more flexible, and ready for AI from day one.
It is also highly visible. When your team sees a faster, more capable website running at a fraction of the old cost, it builds the necessary confidence to tackle the next tool.
Build only what you can’t buy. For unique internal workflows, use AI-assisted development. Just make sure you build on proper infrastructure, not on duct tape.
35% is the floor, not the ceiling.
Retool’s 35% replacement rate is just a snapshot of early 2026.
Every indicator suggests this pace is accelerating. AI-assisted development is faster. Open-source alternatives are more robust. SaaS pricing pressure is more aggressive.
Companies evaluating their spend right now will be miles ahead of those who wait for their next painful renewal cycle to force a panicked decision.
You don’t want to be in the last wave.
Want to know where your website stands? Assess your site — our free AI assessment evaluates your website’s tech stack, performance, and AI readiness, and delivers a score with a personalized roadmap. Takes 60 seconds.