Should we stop buying SaaS and build our own tools with AI?

It depends on the tool and the workflow. Keep buying for common, heavily regulated or business-critical systems such as accounting, payroll, CRM and document storage, where vendors carry security, compliance and upkeep. Consider building where your workflow is specific to how you make money, where people copy data between tools by hand, or where you pay for a large platform and use a small part of it. Build only if someone will own the tool after launch. For most firms the answer is a mix, decided one workflow at a time.

Who this is for

CFOs, COOs and technology leads who pay for a growing list of software subscriptions and are wondering whether AI changes the build-or-buy decision.

The short version

AI has lowered the cost of building software. It has not lowered the cost of owning it.
Build where your process is a real advantage or where tools force awkward workarounds.
Buy where the problem is common and the vendor carries compliance and maintenance.
Before building anything, check that your team is using the tools you already have.

Why this is a live question

In October 2026, a widely shared discussion on X described a finance firm that stopped buying some off-the-shelf software and hired engineers to build its own tools with AI coding agents. Replies split between two views: custom tools as a competitive edge, and worries about maintaining and securing code that was produced quickly. We are not repeating the story here. It is a useful hook for a decision many firms now face, whatever their industry.

What actually changed

A few years ago, a custom internal tool for a mid-sized firm usually meant a long and expensive project. As of October 2026, AI coding assistants let an experienced engineer produce a working internal tool much faster than before. That shifts the question from “can we afford to build this?” to “should we own this?”

The second question is harder. Owning software means reviewing it, securing it, testing it, documenting it and keeping it running when AI models, connected systems and the people who built it change.

When building makes sense

The workflow is part of your edge. Your way of screening opportunities, pricing work or reporting to clients is what sets you apart, and generic software flattens it.
People stitch tools together by hand. Staff copy data between systems every week. A small integration or agent can remove that work. See how AI agents connect to your CRM and email.
You use a small slice of a big platform. You pay for broad software and rely on one or two features.
The data is too sensitive to send out. Keeping processing in your own environment is a firm requirement. See local vs cloud AI.

When buying is still better

The problem is common. Accounting, payroll, CRM and document storage vendors spread the cost of compliance and upkeep across many customers.
Regulation is heavy. A vendor’s audits and certifications are hard and costly to copy.
No one will own it. Software nobody maintains becomes a risk, however quickly it was built.
The vendor is adding the AI features you want. Many established software products now include AI features, as of October 2026. Check your current tools before replacing them.

The middle options people skip

Building versus buying is rarely all or nothing. Between the two sit cheaper options:

Use your current tools better. Training often unlocks features your team already pays for. See why your team isn’t using the AI tools you paid for.
Configure what you have. Saved instructions, team assistants and built-in automations cover many needs.
Connect, do not replace. A small integration between two tools can remove the manual step without rebuilding either.
Build a thin layer on top. Keep the vendor system as the record and build a small tool or agent for the step that is specific to you.

Our guides to build vs buy for AI tools and custom AI vs off-the-shelf tools cover these options in more depth.

Questions to answer before you build

Who owns this tool in 12 months, by name?
How are access and data protected, and who reviews the code?
What happens if the person who built it leaves?
How will you know if it breaks quietly?
What will running it cost each month, including AI usage? See controlling AI costs across a team.

If you cannot answer these, you are not ready to replace a vendor product.

A practical way to decide

List your software spend and the workflows each tool supports.
Mark the pain. Where do people work around a tool or re-enter data?
Score each workflow on three things: how specific it is to your firm, how often it runs and how sensitive the data is.
Try the light option first: training, configuration or a connector.
Build the smallest useful version for one high-scoring workflow, with a named owner and a plan for maintaining it after launch.
Cancel the old subscription only after the new tool has run reliably for a full cycle of real work.

Common mistakes

Replacing a mature product to save license fees, then paying more in maintenance.
Building before the team understands what AI can already do in current tools.
Letting one enthusiast build critical tools with no review or documentation.
Cancelling the vendor too early, before the replacement has proved itself.

When you do not need us

If you have an experienced engineer, a clear owner and one well-defined workflow, you can run the steps above yourself. Outside help is most useful when the decision spans many tools or teams, or when no one inside the firm can review and maintain what gets built.

How Ruby helps

AI strategy: a clear build-or-buy view across your workflows, including when to keep what you have.
AI engineering: secure, documented agents, integrations and internal tools that you own.
AI training: so your team gets full value from the software you already pay for.

If you are weighing whether to replace a tool, book a call with Ruby.