AI is changing the way we work, communicate, search, build websites and interact with technology. At Pooka & Co, we’re curious about AI, but we’re not blindly enthusiastic. We want to make the most of what it can do while being honest about its risks – and making sure technology remains a tool for people, not a replacement for them.
We actively explore new AI tools and technologies, we test them, and we talk to suppliers and technology partners we work with about what they’re developing. We look for places where AI can make our work more efficient, helping us identify areas we might otherwise have missed, or take some of the heavy lifting out of a repetitive process.
But there’s a principle at the heart of how we use it: AI can be in the process. A human is always in the loop.
We see AI as a tool to support our work, not something that does our work for us. We aren’t interested in replacing people with AI and we don’t believe because something can be automated, it necessarily should be.
Instead, we ask a fairly simple question: Is this better done by AI, better done by a human, or – as is so often the case – better done by both?
What we mean by Human in the Loop
For us, Human in the Loop means that meaningful human judgement, oversight and accountability remain key in the work we do.
AI might help us investigate something, but it doesn’t get the final say. What Pooka puts its name to has been reviewed, considered and approved by a human. That matters because we understand both what AI is capable of and, importantly, what it isn’t. AI can process enormous quantities of information incredibly quickly. It can recognise patterns, generate ideas and accelerate certain tasks.
But it doesn’t understand your organisation, your community or your users in the way people do. It doesn’t have lived experience, it doesn’t possess judgement, empathy or accountability. And it can confidently get things wrong.
Those aren’t small caveats, they’re fundamental to deciding when and how AI should be used.
AI in our development work
AI is increasingly becoming part of the toolset available to developers, and we see genuine value in that. Our development team uses AI-assisted tools to help write code, identify potential issues, speed up repetitive development tasks or support troubleshooting.
But what we don’t do, is hand a project to AI and accept whatever comes back.
AI-assisted code writing is part of our development and quality assurance process. It is reviewed and tested by our Developers and, where appropriate, scrutinised by our Directors, Marcus and Paul.
Nothing bypasses human oversight, and when work is presented to a client or deployed to a live website, responsibility for that work sits with Pooka’s people, not with an AI system.
AI in Communications
We see similar opportunities in communications. Anyone who has spent half an hour staring at a blinking cursor knows that sometimes the hardest part of writing is simply getting started.
That’s one of the reasons we developed Pooka AI for WordPress.
Our Generate First Draft tool is deliberately called First Draft. It can use information about an organisation, its mission, values and tone of voice to help turn an idea or collection of notes into something tangible and reflective of the organisation. And it’s often shockingly good at that.
But then the really important work begins. The person using it reviews it, challenges it, changes it and adds their own knowledge, experience, personality and perspective. The First Draft tool will get you well underway. But it won’t remove the human who signs their name to the article. It isn’t the Author.
The same principle applies when we use AI ourselves. It can help with research, structure, ideas, summaries and first drafts, but the finished communication is something a person takes responsibility for.
Curiosity without chasing shiny things
AI is developing extraordinarily quickly, we don’t think the answer is to ignore it. We also don’t think the answer is to bolt AI onto everything because everyone else appears to be doing it. Our approach is to stay curious.
We actively test new capabilities and explore what the platforms, suppliers and technologies we work with are developing. We want to understand where AI genuinely improves experiences for our clients and their users and where it simply introduces another layer of complexity.
That means asking why before asking how.
- What problem are we trying to solve?
- Will AI genuinely make this better?
- What are the risks?
- What information does the system need access to?
- What happens if it gets something wrong?
- And does a person still have the knowledge and opportunity to challenge the result?
Sometimes an approach is AI based. Sometimes an approach is human based. The better approaches increasingly touch on both worlds, but accountability always lies with the person responsible for the task.
AI, security and the changing threat landscape
Our interest in AI isn’t limited to the opportunities it creates. We’re also very conscious that the same technologies can be used by people looking for vulnerabilities, automating attacks or attempting to undermine online systems.
The tools available to bad actors are becoming increasingly sophisticated. That makes active maintenance, monitoring and professional oversight of websites more important, not less.
It is also why we don’t see launching a website as the end of our responsibility. At Pooka, we have mechanisms in place that allow us to roll out important security updates across multiple websites concurrently. When security fixes become available, we can act quickly to protect the websites we manage.
This is the type of work that we perform quickly and quietly for all of our clients, leaving them to think about the content of the site, with peace of mind about the underlying technicalities.
AI may change the nature and speed of some digital threats. Our response is the same principle we apply elsewhere: knowledgeable humans, supported by good technology, actively looking after the systems entrusted to us.
Responsibility stays with us
Perhaps the simplest way to describe our approach is this: we don’t outsource responsibility to AI.
If we’ve designed it, developed it, recommended it, written it or delivered it, we take responsibility for it. AI may have helped somewhere along the way. It may have saved time, uncovered something useful, suggested an approach or taken care of some of the repetitive work.
But there’s a Pooka person in the loop. Because ultimately our clients don’t employ an algorithm. They employ us.
