I have been thinking about AI and vibe programming in a slightly different way recently.

Think about IKEA furniture.

For years, millions of people assembled flat-pack furniture with an ordinary screwdriver. It worked. You could follow the instructions, turn every screw by hand and eventually end up with the finished product.

Then you picked up an electric screwdriver.

The furniture didn't change. The principles of assembling it didn't change. You still needed to know which pieces went where. You could still put something on backwards, use the wrong screw or ignore the instructions entirely.

What changed was the effort.

Something that required dozens of repetitive manual actions suddenly became considerably faster.

I think we are approaching the electric screwdriver moment for enterprise software.

We've been trying to move technology closer to the business for years

Elite have always advocated for this. The idea itself isn't new.

For years, the technology industry has talked about low-code and no-code platforms putting more capability into the hands of business users.

It is also something I have strongly believed in.

If the person who understands the commercial problem can participate directly in designing the solution, we remove layers of translation, shorten delivery cycles and make technology more responsive to the business.

But there has always been an important qualification.

Just because someone can configure something doesn't necessarily mean they should be given unrestricted ability to do so.

Large enterprise applications aren't simply collections of screens and workflows.

They sit inside an ecosystem of data, integrations, security, calculations, controls, business rules, regulatory requirements and other applications.

A seemingly simple decision today can create significant consequences tomorrow.

That is why good enterprise software has always required architecture, system design, technical governance and controlled change.

AI doesn't make any of those things disappear.

What AI changes is the effort between an idea and a working solution

This is where I think natural language, AI-assisted configuration and what has become known as "vibe programming" become genuinely interesting.

The traditional enterprise delivery process contains a significant amount of translation.

• A business user explains a requirement.
• Someone documents it.
• Someone designs it.
• Someone translates that design into technical requirements.
• Someone configures or develops it.
• Someone tests whether what was built actually represents what the business originally wanted.

AI has the potential to compress much of that process and is doing so in many industries.

Increasingly, a knowledgeable person can describe an outcome in natural language and have technology perform a substantial amount of the mechanical work required to produce it.

That doesn't eliminate expertise.

It amplifies expertise.

And that distinction matters.

The bottleneck moves from execution to judgement

If AI makes creating software significantly easier, then the scarce resource is no longer necessarily the person capable of physically configuring it.

The scarce resource becomes the person who knows what should be configured.

• Someone still needs to understand the business.
• Someone needs to understand the data.
• Someone needs to recognise that a change to one calculation affects three downstream processes.
• Someone needs to understand the commercial agreement behind the calculation.
• Someone needs to determine whether the solution being created is secure, supportable, auditable and consistent with the wider enterprise architecture.

AI can dramatically accelerate execution.

It doesn't automatically provide good judgement.

This has significant implications for organisations adopting AI, but it also has significant implications for companies like ours that implement enterprise technology.

Our role has to change too

Historically, a significant part of a specialist systems implementation company's value has been its ability to provide people capable of configuring complex applications.

If AI dramatically increases the productivity of those people, simply selling configuration capacity becomes a less compelling proposition.

I see that as an opportunity.

The future shouldn't be about putting ten consultants into a room to configure something that three highly experienced people working alongside the client and using AI-enabled tools could potentially accomplish.

It should be about using that increased capacity to achieve considerably more.

For Elite Energy , that means continuing to move towards working with our clients rather than simply building for them.

Our value increasingly comes from combining deep energy-industry knowledge, architecture, system design, governance and platforms such as EnergySys with the productivity that AI can provide.

The objective isn't to maximise the amount of configuration we perform.

It is to minimise the distance between a genuine business requirement and sustainable operational value.

And I deliberately say sustainable value, not MVP

The technology industry has become obsessed with the Minimum Viable Product.

There is absolutely a place for MVP thinking. Getting something into people's hands quickly can be enormously valuable.

But when AI makes building something increasingly easy, simply being able to create an application quickly becomes less remarkable.

The harder question is what happens afterwards.

• Can it operate reliably?
• Can it scale?
• Can it be changed safely?
• Does it fit the enterprise architecture?
• Can another person understand it three years from now?
• Are the calculations and decisions auditable?
• Is security designed into it?
• Can the business take greater ownership without creating uncontrolled risk?

And, most importantly, does it continue producing value as the organisation changes?

The objective should therefore move beyond minimum viable product towards long-term sustainable value.

Speed matters.

But speed without architecture and governance simply allows us to create tomorrow's technical debt faster.

There is a warning here for organisations embracing vibe programming

If an electric screwdriver allows one person to assemble furniture five times faster, it also allows them to put the wrong screw into five times as many pieces of furniture.

AI creates the same paradox.

The democratisation of software creation is incredibly exciting.

But organisations could easily move from a world where technology change was frustratingly slow to one where hundreds of applications, agents, workflows and automations are being created without an understanding of how they fit together.

That isn't digital transformation.

It's architectural chaos delivered at unprecedented speed.

The answer isn't to prevent people from using the tools.

It is to create the environment in which they can use them safely.

That means governed platforms, clear architectural principles, appropriate security, controlled data models, testing, auditability and sensible boundaries around what can be changed.

In other words, democratise creation without democratising chaos.

This could actually make enterprise platforms more important

There is an interesting contradiction in all of this.

If AI can increasingly create applications, why do we need enterprise platforms?

I think the opposite question may prove more useful.

If organisations are about to create software at a rate we have never experienced before, where is all of that software going to live?

Enterprises still need trusted data.

• They still need security.
• They need calculations that can be explained.
• They need audit trails.
• They need integrations.
• They need permissions.
• They need controlled production environments.
• They need resilience.

And they need confidence that a critical business process isn't dependent on something somebody vibe-programmed on a Friday afternoon and subsequently forgot about.

Platforms provide that foundation.

In our world, I believe there is a powerful combination emerging between configurable enterprise platforms such as EnergySys, our portals technology, deep domain expertise and AI.

The platform provides the governed foundation.

AI dramatically accelerates what can be accomplished on top of it.

And experienced people ensure that what gets built actually makes sense.

Doing more with less shouldn't simply mean cutting costs

There is another important perspective here.

Whenever technology dramatically improves productivity, the immediate conversation tends to become: how many people can we remove?

That is far too narrow.

If a team becomes twice as productive, there are two ways to look at it.

We can produce yesterday's output with half the resources.

Or we can ask what becomes possible if the same talented people can now accomplish twice as much.

In many organisations I work with, the backlog of worthwhile technology change is enormous.

• There are commercial processes that should be automated.
• Manual reconciliations that shouldn't exist.
• Spreadsheets performing critical functions.
• Data that should be available immediately but takes days to assemble.
• Business rules waiting months for technology teams to implement them.

AI gives us an opportunity to attack that backlog.

So "doing more with less" shouldn't simply mean fewer people.

It should mean more capability, more responsiveness and more business value for every dollar invested in technology.

The electric screwdriver didn't eliminate the craft

When electric screwdrivers became commonplace, we didn't conclude that anybody holding one had suddenly become a carpenter.

• The tool increased productivity.
• Knowledge still mattered.
• Experience still mattered.
• Design still mattered.
• And craftsmanship still mattered.

I think enterprise AI will follow a similar path.

The organisations that win won't necessarily be those that give everyone an AI tool and tell them to start building.

They will be the organisations that combine the extraordinary productivity of these tools with business expertise, architecture, governance and good judgement.

And the same applies to technology companies.

Our future isn't about protecting the amount of manual work involved in delivering software.

It is about embracing tools that allow us and our clients to accomplish dramatically more.

The electric screwdriver didn't make the person building the furniture irrelevant.

It just meant they could build a lot more.