• In Depth

AI at Brightec: where we're going and why it matters

What we are doing is taking a serious, sustained look at how AI might reshape our work over the next 3-5 years — and what that means for a company like Brightec.

We’ve started to experiment. We’ve got early wins. We’ve hit a few nerves. And we’re not pretending to have all the answers.

Here’s what we’re learning — and what we’re protecting as we adapt.

The assumption we started with

Earlier this year, we asked a blunt question: What happens to a digital product agency that ignores AI?

There’s a growing sense across the industry that delivery speed, cost efficiency, and content generation will all become commoditised. AI is accelerating the standard. Clients will expect more, faster, for less. And we can’t ignore that pressure.

But here’s the first challenge: Are we sure that “faster and cheaper” is the right problem to solve?

Many of our clients come to us not because they lack delivery speed, but because they lack clarity. They need thought partnership. A team who’ll think with them, challenge assumptions, and own delivery end-to-end.

AI can help — but only if we apply it carefully.

So instead of assuming “AI will save us time”, we’re testing a better hypothesis: AI should amplify our judgment, not just automate our work.

Why we’re taking this seriously

We’re not chasing headlines. But we’re not naive either.

AI is already shifting how people:

• Think about value for money

• Engage with user experiences

• Expect transparency and speed in communication

It’s also exposing bottlenecks in traditional workflows: research that takes too long, feedback that’s hard to interpret, and documentation that gets stale quickly.

For a team that prides itself on craft and clarity, that’s an opportunity, not a threat.

The risk: shallow adoption

Let’s be honest about the danger here. It’s easy to use AI as a productivity badge. Fire up a ChatGPT prompt, generate some copy, check a box.

But that kind of shallow adoption leads to:

• Boring work that nobody’s proud of

• Misalignment with clients who value nuance

• Teams losing their sense of craft and ownership

We’re not going there. If AI is going to be part of how we work, it has to serve what makes Brightec great. That means making space for human judgment. Respecting context. And designing for outcomes, not just outputs.

Where we’re actually using it

So what are we doing now?

"I’ve been handing off a lot of the programming 'grunt work' — like refactors and writing boilerplate code — to ChatGPT. This is valuable and important work, but I find that if I get the ball rolling first, I can leave AI to it and switch my focus to more engaging engineering work like architectural decisions, implementing tricky logic, and designing new features."

Jonny, Senior Engineer at Brightec

We’ve chosen three areas to explore. These aren’t flashy — but they’re real.

a. Accessibility

AI helps us bake accessibility into the design process from day one. That can mean:

• Faster testing of contrast, structure, and content across screen sizes and conditions

• Smarter suggestions when it comes to copy or layouts that could confuse or exclude users

• Early flags on patterns that break WCAG guidelines

We’re not replacing our designers or researchers. We’re augmenting them — giving them more time to focus on the why, not just the what.

b. Prototyping

In the early stage of a project, we often need to show ideas quickly, long before any code gets written. AI lets us:

• Create simple, interactive prototypes based on client briefs or initial sketches

• Iterate faster without the overhead of full design sprint prep

• Explore divergent ideas in hours, not weeks

This isn’t about cutting corners. It’s about widening the option space, early and cheaply. That pays off later.

c. Communication

One of the biggest hidden costs in digital projects? Miscommunication.

We’re testing how AI can help:

• Summarise sprint updates or client feedback without losing nuance

• Draft clearer progress reports

• Spot patterns or blockers we might miss in raw notes

It’s early days. But the signal is promising. Clients feel more informed. The team feels more aligned. And decisions are happening with less back-and-forth.

What we’re not doing (and why that matters)

We’re not offering “AI transformation” services. We’re not pretending to be machine learning experts. And we’re not replacing developers with prompts.

That’s deliberate.

We believe in tools that enhance, not undermine, our culture:

• We care about learning, context, and judgment

• We want high-trust relationships, not high-volume transactions

• We build long-term partnerships — that doesn’t happen if your team feels disposable

If AI ever starts to compromise those values, we stop.

Resistance is a feature, not a bug

There’s been healthy scepticism internally, especially around legal risk. The team raised a fair question: What happens if we paste client code into a public AI tool?

That’s a real risk. So we’re working on real answers:

• Clear guidelines on what can and can’t be shared

• Engaging legal counsel to help shape safe practice

• Using closed or enterprise-grade AI tools where needed

The point here isn’t to move fast and break things. It’s to move thoughtfully, with the team’s trust and the client’s safety in mind.

This is not a side project

We’re not “letting a few people play with AI on the side.” This is strategic. (We’ve got enough side projects, thank you.)

But here’s the balance we’re aiming for:

• Top-down direction — to shape priorities, define the risk appetite, and stay focused on what matters.

• Bottom-up momentum — so that the team can explore, experiment, and share findings in the open.

We’re not waiting for a perfect framework. We’re creating space for progress.

A quick note on structure

Great things happen when direction and discovery pull together. Our leadership defines the goals. Our teams explore, learn, and build from the ground up. We believe in strategy shaped from above, and innovation unlocked from below.

We’ve rolled out team-wide access to ChatGPT with light guardrails:

• We use our judgment. AI can help us move faster, but it doesn’t replace thoughtful decisions.

• We review before sharing. Always check outputs before using them externally.

• We share what works. If something’s helped you, pop it in #ai-exploration (a Slack channel).

We’ve made this part of our actual workflow, not a separate lab project. That means accountability. And it means anyone can contribute ideas, shape tools, and push the thinking forward.

What this means for our clients

Let’s be clear: our clients aren’t asking us to use AI in our work. They’re asking for:

• Better outcomes

• Less waste

• More value

Some of them are asking us to build AI-powered features into their products — and we’re equipped for that. But they’re not asking us to generate user stories with ChatGPT or prototype interfaces using Midjourney. That’s our call to make, based on whether it improves the result.

When we use AI well, here’s what changes for them:

• They get prototypes sooner, not promises

• They feel clearer about what’s happening, why, and when

• They benefit from design systems and development practices that scale better and cost less

In other words: they get more of what they already value about Brightec.

That’s the goal.

What we’re still figuring out

This is a live conversation, not a solved problem. Some of the open questions on our radar:

• Ownership: Who owns the output when AI is involved?

• Trust: How do we make sure decisions made with AI are explainable and defensible?

• Design integrity: Will AI start flattening ideas into generic sameness?

• Value alignment: What happens when AI tools optimise for speed, but we care about clarity?

These aren’t abstract. They hit the core of how we work and what our clients pay us for.

What we’re protecting

Here’s what won’t change:

• We still believe in excellent software.

• We still want to be a company where people thrive.

• We still think trust, care, and good thinking matter more than trend-chasing.

AI can help - if we stay clear on what “great” looks like. (Great is still handcrafted. Just maybe with a co-pilot who doesn’t sleep.)

That’s the work.