The Internal Tool Renaissance: Why AI Makes Build-Your-Own Viable Again
From roughly 2015 to 2023, the answer to "should we build or buy?" was almost always buy. SaaS was cheap, plentiful, and getting better every quarter. Building internal tools felt like vanity — why pay engineers six figures to rebuild something Salesforce already does?
That logic made sense when custom development was expensive. It no longer is.
The Economics Have Flipped
AI-augmented engineering has collapsed the cost of building custom software by an order of magnitude. A system that would have taken a team of four engineers three months now takes one or two engineers a few weeks. The math has fundamentally changed.
Consider a mid-market company paying $80,000/year for a CRM that their sales team tolerates but never loves. They've customised it as far as the platform allows, built workarounds for the gaps, and trained new hires on the quirks. Now compare that to a purpose-built CRM that matches their exact sales process, integrates with their specific data sources, and costs $30,000 to build with ongoing maintenance that's a fraction of the SaaS bill.
The custom tool wins. Not on features — on fit.
Exact Fit Beats Close Enough
SaaS products are built for the average customer. They have to be. When you serve thousands of companies, you optimise for the common denominator. Your workflow gets bent to fit the tool, not the other way around.
Internal tools have no such constraint. They can be opinionated in exactly the ways your business is opinionated. Some examples we've seen work brilliantly:
A custom knowledge base with domain-specific search. A manufacturing company's engineers were drowning in a generic wiki. We built a search system that understood their part numbering scheme, cross-referenced technical specs, and surfaced related failure modes. Confluence could never do that.
A reporting dashboard that answers actual questions. Leadership at a logistics company was getting dashboards full of charts nobody asked for. The custom alternative had six views, each answering a specific question their CEO asked every Monday morning. No drilling. No filtering. Just answers.
An onboarding system that matched an exact process. An agency had a 23-step client onboarding flow that spanned five different SaaS tools and a spreadsheet. One custom application replaced all of it. New clients went from confused to productive in half the time.
The Hidden Cost of "Close Enough"
Every workaround your team builds around a SaaS limitation is a tax. Every manual step that exists because the tool doesn't quite fit is friction. Every training session for new hires on "how we use this tool differently" is wasted time.
These costs are invisible because they're distributed across every employee, every day. They never show up on an invoice. But they're real, and they compound.
When to Still Buy
Not everything should be custom-built. Commodity functions — email, file storage, source control — are better bought. The rule is simple: if the tool touches your competitive advantage or your unique process, build it. If it doesn't, buy it.
Accounting software? Buy it. A system that manages the specific way you deliver value to clients? Build it.
The Opportunity
Most companies are sitting on a backlog of internal tool ideas that never cleared the ROI bar under traditional development economics. AI has lowered that bar dramatically.
The companies that recognise this shift early will operate with tools built for exactly how they work. Everyone else will keep paying to be average.