AI is creating tremendous opportunities across industries. But without controlled and responsible use, it can also create serious challenges around cost, engineering capability, sustainability and market balance.
Everyone is talking about AI, and we are seeing an AI burst across almost every industry. I wanted to share a few thoughts on how organisations can gain its benefits while introducing the controls needed to use it responsibly.
One of the biggest AI use cases in 2026 is the AI coding agent. Many organisations are restructuring engineering teams around these tools. At the same time, they are seeing substantial AI infrastructure costs caused by unreasonable or poorly directed usage.
Many developers now prompt AI for almost every task. Sometimes people create unnecessary scripts or repeatedly use AI simply to demonstrate how extensively they are using it. AI is intended to improve productivity, but uncontrolled usage can increase cost without producing proportional value.
Looking back at developer tools
There was a time when developers wrote code in Notepad and compiled it using command-line tools. Then integrated development environments arrived. Looking back, we could describe them as an early form of intelligence for developers.
IDEs gradually became smarter, introducing:
- Auto-completion
- Syntax highlighting
- Auto-formatting
- Debugging support
Plugins later generated code snippets and templates while automating repetitive work. After 2025, AI coding agents took this progression much further: instead of helping with syntax or snippets, they could interpret prompts and generate complete code.
Every stage of this evolution shared one goal: reduce developer effort and accelerate software development.
Earlier advances did not eliminate developer jobs. They helped developers become more productive. With AI coding agents, however, usage is not yet consistently streamlined. Some teams use them effectively, while others overuse them, producing a mixture of positive and negative outcomes.
AI needs governance and logical controls
AI coding agents should be viewed as the next major productivity improvement in software engineering. The important difference is that organisations now need appropriate governance and logical controls, allowing them to gain value without creating unnecessary challenges around cost, employment or sustainability.
Engineering organisations already have established roles and levels—from junior engineers through engineers, senior engineers, leads and architects. AI platforms should allow administrators to allocate tokens, usage quotas or other sensible limits based on these roles and responsibilities.
This approach could help organisations:
- Keep AI costs predictable.
- Prevent unnecessary AI usage.
- Encourage developers to use AI where it genuinely adds value.
- Help developers use an allocated AI budget effectively rather than generating unlimited prompts.
This is similar to the way organisations already manage cloud budgets, infrastructure and software licences.
Why responsible usage matters
1. Developers may stop thinking
If AI solves every problem, developers may slowly lose the habit of analysing and solving problems themselves. AI should improve thinking—not replace it.
2. We may fail to develop the next generation of strong engineers
Today’s senior engineers reached their level by solving problems over many years. If junior developers depend entirely on AI, will we develop engineers with the same depth of capability ten or fifteen years from now? It is a question worth considering carefully.
3. Environmental impact
As AI demand increases, data centres must expand significantly to support AI models. More infrastructure means greater energy consumption, making responsible usage important not only for cost but also for the environment.
4. Monopoly and imbalance
Overdependence on a small number of AI providers may create monopolies and imbalance in the technology ecosystem. Like breaking a link in a food chain, the impact may not be immediately visible but could become a serious concern over time.
Final thoughts
I am not against AI. In fact, I believe AI coding agents are among the greatest productivity improvements software engineering has ever seen.
But like every powerful technology before them, they should be used responsibly. With appropriate governance, controlled usage and thoughtful engineering practices, we can benefit from AI while avoiding unnecessary costs and long-term risks.
Should organisations introduce structured AI governance, or should AI remain completely unrestricted?
Let us collaborate and build strong practices for using AI going forward.