How much do companies waste on unused SaaS licenses?

A lot — a large share of licenses (commonly cited around 30–46%) go unused, and big enterprises waste tens of millions of dollars a year on licenses nobody touches, reclaiming only a small fraction. Much of this is thin, per-seat tooling where you pay for every seat regardless of use, which is exactly the slice worth scrutinizing.

Is it cheaper to build internal apps than buy SaaS in 2026?

For simple, thin-feature tools, increasingly yes — AI-assisted development has lowered the build cost and you escape per-seat fees. But only if the platform absorbs the backend and maintenance; otherwise custom software's lifetime maintenance cost can erase the savings. For complex or compliance-heavy systems, buying is still usually cheaper and safer.

Which SaaS tools are worth replacing with a custom app?

Thin internal tools: CRUD apps, admin panels, dashboards, and simple automation glue. These are standard shapes you rent per seat and can own instead. Avoid replacing deep systems of record (ERP, payroll, GRC), products with network effects, or anything needing vendor liability and certifications — there, buying wins.

Does AI actually make building internal software faster?

For simple apps, it genuinely lowers the bar — including for non-developers. But the gains are contested for complex work: a controlled study found experienced developers were slower with AI on hard tasks even while feeling faster. The honest read is that AI collapses the cost of simple, well-scoped apps, not of complex specialized software.

What are the risks of replacing SaaS with internal apps?

Mainly maintenance and governance. Most in-house builds miss budget or timeline or get abandoned, and custom software's lifetime cost is dominated by maintenance — so you can trade license waste for a perpetual patching burden. Building outside IT oversight also adds security and compliance exposure. The risk shrinks only if a platform handles the backend, hosting, and upkeep.

Why is per-seat SaaS pricing breaking down?

Because AI decouples value from headcount — when an agent does the work, per-seat pricing no longer tracks what's delivered, and idle seats become pure waste. Analysts expect most vendors to move off pure per-seat pricing toward consumption and outcome models within a few years, which is part of why owning a usage-priced app can be cheaper.

Who owns the data when you build instead of buy?

You do. That's a core reason teams replace thin SaaS with internal apps — instead of your data living in a vendor's multi-tenant system under their terms, it stays in your own database and environment. That ownership matters for privacy, compliance, portability, and avoiding lock-in, on top of the cost savings.

← Resources

By Marcus Okafor — Platform Engineer, Apps · 2026-06-12 · INSIGHT

Replacing SaaS With Internal Apps: Build vs Buy in the AI Era

AI has lowered the cost of building simple internal apps, making it viable to replace some per-seat SaaS subscriptions. This guide covers the SaaS sprawl and waste numbers, where AI genuinely helps (and where it doesn't), which tools are worth replacing, and the maintenance and governance risks.

The SaaS sprawl and waste problem

The average company runs around 100 SaaS apps, large enterprises more, and the spend keeps rising even as app counts plateau — vendors extract more through AI add-ons, tier upgrades, and consumption billing. The waste is the headline: a large share of licenses (commonly cited around 30–46%) go unused, and big enterprises waste tens of millions of dollars a year on licenses nobody touches, reclaiming only a fraction of it.

Much of this is thin, per-seat tooling — admin panels, simple workflow glue, single-purpose trackers — where you pay for every seat whether or not it's used. That's the slice where the build-versus-buy math is shifting.

Where AI lowers build cost — honestly

AI-assisted development has genuinely lowered the bar for building simple apps; the majority of organizations expect to build more internal software, and natural-language "describe it and AI writes it" tooling has put basic app-building within reach of non-developers.

But be honest about the limits, because credibility depends on it. A controlled study found experienced developers were actually slower with AI on complex tasks even while feeling faster, and team-level gains in task throughput came with much higher review time. The accurate framing: AI collapses the cost of simple, well-scoped internal apps — it does not make complex, specialized software cheap. Replace the thin tools; don't assume you can vibe-code your ERP.

What's worth replacing — and what isn't

Worth replacing: internal CRUD apps, admin panels, dashboards, and simple automation glue — the standard shapes covered in our [build-internal-tools guide](/articles/build-internal-tools-with-ai). These are thin-feature tools you're renting per seat, and an app you own can do the same job with your data in-house.

Not worth replacing: deep, specialized, compliance-heavy systems of record — ERP, payroll, GRC — products with strong network effects, and anything where you need a vendor's liability, certifications, or constant regulatory updates. The 2026 framing has shifted from "build vs buy" to component-level decisions: buy the compliance-heavy core, build the differentiating and glue layers, and use AI for the parts that are genuinely simple.

The consolidation upside

Beyond cost, replacing thin SaaS with owned apps consolidates. You collapse a stack of subscriptions onto one platform, keep your data in-house instead of spread across vendor tenants, and escape per-seat pricing — which is breaking down anyway as AI agents act like "users" and decouple value from headcount. Analysts expect the majority of vendors to move off pure per-seat pricing toward consumption and outcomes within a few years.

The pattern in practice: replace a recurring per-seat bill for a thin-feature tool with a custom app you own, where data stays in-house and cost scales with usage rather than headcount.

The risks: maintenance and governance

The honest counterweight is that in-house builds fail often. Most in-house IT projects miss their timeline or budget or get abandoned, and custom software carries the majority of its lifetime cost in maintenance. Replace a SaaS subscription with an app you then have to patch, secure, and keep running forever, and you may have just traded license waste for a maintenance tax.

The other risk is governance lag — building outside IT oversight trades license waste for security and compliance exposure, and shadow-AI-related breaches carry a real cost premium. The replacement only pays off durably if the platform absorbs the backend, hosting, auth, and maintenance so you're not re-creating that 60–80% maintenance burden yourself.

How osFoundry fits the durable slice

osFoundry targets exactly the durable part of this shift: the simple, high-waste internal apps that eat per-seat SaaS subscriptions while a large share of licenses go unused. Because apps ship as code with an included backend — database, auth, secrets, cron, functions, hosting — it removes the maintenance tax that sinks most in-house builds, and your data stays in your environment rather than a vendor's tenant. AI-native authoring lowers the bar to build these apps without re-hiring an infrastructure team, and [usage-based pricing](/articles/per-seat-vs-usage-based-ai-pricing) means cost scales with what the app does, not headcount. The honest positioning: replace the thin per-seat tools you can own end-to-end, and keep buying the deep systems of record where a vendor's liability and certifications are the point.

Frequently asked questions

How much do companies waste on unused SaaS licenses?: A lot — a large share of licenses (commonly cited around 30–46%) go unused, and big enterprises waste tens of millions of dollars a year on licenses nobody touches, reclaiming only a small fraction. Much of this is thin, per-seat tooling where you pay for every seat regardless of use, which is exactly the slice worth scrutinizing.
Is it cheaper to build internal apps than buy SaaS in 2026?: For simple, thin-feature tools, increasingly yes — AI-assisted development has lowered the build cost and you escape per-seat fees. But only if the platform absorbs the backend and maintenance; otherwise custom software's lifetime maintenance cost can erase the savings. For complex or compliance-heavy systems, buying is still usually cheaper and safer.
Which SaaS tools are worth replacing with a custom app?: Thin internal tools: CRUD apps, admin panels, dashboards, and simple automation glue. These are standard shapes you rent per seat and can own instead. Avoid replacing deep systems of record (ERP, payroll, GRC), products with network effects, or anything needing vendor liability and certifications — there, buying wins.
Does AI actually make building internal software faster?: For simple apps, it genuinely lowers the bar — including for non-developers. But the gains are contested for complex work: a controlled study found experienced developers were slower with AI on hard tasks even while feeling faster. The honest read is that AI collapses the cost of simple, well-scoped apps, not of complex specialized software.
What are the risks of replacing SaaS with internal apps?: Mainly maintenance and governance. Most in-house builds miss budget or timeline or get abandoned, and custom software's lifetime cost is dominated by maintenance — so you can trade license waste for a perpetual patching burden. Building outside IT oversight also adds security and compliance exposure. The risk shrinks only if a platform handles the backend, hosting, and upkeep.
Why is per-seat SaaS pricing breaking down?: Because AI decouples value from headcount — when an agent does the work, per-seat pricing no longer tracks what's delivered, and idle seats become pure waste. Analysts expect most vendors to move off pure per-seat pricing toward consumption and outcome models within a few years, which is part of why owning a usage-priced app can be cheaper.
Who owns the data when you build instead of buy?: You do. That's a core reason teams replace thin SaaS with internal apps — instead of your data living in a vendor's multi-tenant system under their terms, it stays in your own database and environment. That ownership matters for privacy, compliance, portability, and avoiding lock-in, on top of the cost savings.