The Historical Cost Barrier
Enterprise resource planning (ERP) systems—the software that manages a business's core operations—have traditionally been expensive. Off-the-shelf solutions like SAP, Oracle, or Microsoft Dynamics carry hefty licensing fees and require extensive customisation to fit specific business processes.
Custom ERP development has been even more expensive. Building a bespoke system from scratch could easily run into hundreds of thousands or even millions of pounds. This wasn't just about complexity—it was about the sheer volume of code required. Every form, every report, every workflow, every integration needed to be designed, built, tested, and documented.
For a small business with unique operational requirements, this created an impossible situation. Generic SaaS tools didn't fit their needs, but custom development was completely out of reach. They made do with spreadsheets, workarounds, and frustration.
What AI Changes
AI-assisted development doesn't replace developers or eliminate the need for expertise. What it does is dramatically accelerate certain categories of work.
Boilerplate generation. A significant portion of any business application is repetitive scaffolding: CRUD operations, form validation, data access layers, API endpoints. AI can generate this code quickly and accurately, freeing developers to focus on the genuinely complex parts.
Faster iteration. When a client describes a workflow, AI can rapidly produce a working prototype. Changes that might have taken days can now be explored in hours. This means more time can be spent getting things right, rather than locked into initial decisions.
Broader capability. A small development team can now tackle projects that would previously have required a larger team. Not because AI does their job, but because it handles the tedious parts, letting experienced developers apply their expertise more broadly.
Better documentation. AI is genuinely good at generating documentation, tests, and other supporting materials. These often get skimped on in traditional projects due to time pressure. With AI assistance, they can be produced alongside the code itself.
The Multiplier Effect
To be clear: AI doesn't write good software by itself. Left unsupervised, it produces plausible-looking code that often has subtle bugs, security issues, or architectural problems. The value comes from pairing AI capabilities with experienced developers who know what good software looks like.
An experienced developer using AI tools effectively can be significantly more productive than either working alone. The developer provides judgement, architecture, and quality control. The AI provides speed and breadth. Together, they can tackle projects that were previously impractical.
This multiplier effect is what changes the economics. A custom ERP that might have required a team of five for a year might now be achievable with a team of two in a similar timeframe. The expertise requirement hasn't changed—if anything, it's increased—but the raw labour component has been compressed.
New Possibilities for Small Businesses
This shift opens up genuine new possibilities. A small manufacturing company with unique production workflows can now consider custom software. A growing agency with specific client management needs doesn't have to force their processes into generic tools. A specialist retailer with unusual inventory requirements can have software built around how they actually work.
Combined with engagement models like CoSaaS that spread costs over time, the accessibility gap is closing from both directions: development costs are lower, and payment structures are more flexible.
This doesn't mean every small business should rush to build custom software. Generic tools work well for generic needs. But for businesses whose operations are genuinely distinctive—where their processes are part of their competitive advantage—custom software is now a realistic option rather than a fantasy.
What This Means for Quality
There's a reasonable concern that AI-assisted development might lead to lower quality software. If code can be generated faster, won't corners be cut?
In our experience, the opposite is true. The time saved on routine tasks can be invested in areas that often get shortchanged: architecture, testing, security review, performance optimisation. The boring parts go faster; the important parts get more attention.
The key is using AI as a productivity multiplier for good practices, not as a shortcut around them. More code review, not less. More testing, not less. More attention to security, not less. AI makes all of these faster too.
The Human Element
Perhaps counterintuitively, AI-assisted development has made human expertise more valuable, not less. Anyone can generate code now. What separates good software from bad software is the judgement applied: which code to generate, how to structure it, what to refactor, when to rewrite, how to handle edge cases.
The developers who thrive in this environment are those who can work at a higher level of abstraction. They're directing the AI, reviewing its output, correcting its mistakes, and integrating everything into a coherent whole. This requires deeper understanding, not less.
For businesses commissioning custom software, this means the same thing it always did: the quality of the people matters more than the tools they use. AI makes good developers more productive, but it doesn't make inexperienced developers good.
Looking Forward
We're still in the early stages of understanding how AI will reshape software development. The tools are improving rapidly, and practices are evolving to take advantage of them. What's clear is that the economics have shifted permanently.
Custom software that fits your business exactly is no longer a luxury reserved for enterprises with seven-figure IT budgets. For growing businesses with distinctive operational needs, it's becoming a practical option. The question is no longer "Can we afford custom software?" but "Would custom software give us a competitive advantage?"
If you're curious whether custom development might make sense for your business, we're happy to explore that question. Sometimes the answer is no—generic tools really are the right choice. But increasingly, for businesses willing to invest in software that truly fits, the answer is yes.