Our approach to using AI as a practical tool - while keeping strategy, judgement, creativity and client relationships human-led.
Artificial intelligence is now part of everyday digital work. It can generate copy, suggest designs, write code, analyse data, organise research and automate tasks that used to take hours.
We think that is exciting. We also think it needs perspective.
At Enigmatio, we use AI because it can make good people faster, more informed and more capable. What we do not do is treat it as a substitute for experience, judgement or a real conversation with a client.
That is what we mean by People First, AI Second.
It is not an anti-AI position. Quite the opposite. It is a practical way of getting the value from powerful new technology without forgetting what clients are actually paying an agency for: people who understand the problem, make decisions, take responsibility and care about the outcome.
Use AI where it adds leverage. Keep humans where judgement matters.
AI is a tool, not a strategy
There is a temptation to talk about AI as if using it automatically makes a business innovative. It does not.
A tool is only useful when somebody knows what problem they are trying to solve. The same is true of a new website, a brand identity, a marketing campaign or an advertising account. The difficult part is rarely producing more output. The difficult part is deciding what the right output should be.
Before we think about tools, we want to understand the business: who the customer is, what is not working, what needs to change and what success actually looks like. AI can help us investigate options. It cannot take ownership of those decisions.
Where AI genuinely helps us
1. Research and discovery
Good projects begin with good questions. AI can help us organise large amounts of information, identify themes, compare options and explore a subject quickly before we dig deeper ourselves.
That might mean structuring competitor research, summarising technical documentation, exploring terminology used in a client’s sector or helping us turn a messy set of notes into a clearer brief.
The important distinction is that AI can help us investigate. It does not decide what is true, relevant or strategically important. We still verify, challenge and interpret the information before it affects a project.
2. Ideas and creative exploration
A blank page is not always the best place to begin. AI can be useful for rapidly exploring directions: alternative headlines, content structures, campaign angles, visual concepts or ways of explaining something complicated.
We treat those outputs as raw material, not finished creative work.
The value comes from selecting, rejecting, combining and refining ideas until they fit the brand, audience and purpose. Ten instant ideas are only useful if someone can recognise which one is worth pursuing – and when none of them are good enough.
3. Development and technical problem-solving
For developers, AI can be an excellent assistant. It can help explain unfamiliar code, suggest approaches, generate repetitive boilerplate, spot possible causes of errors and accelerate debugging.
But code that looks convincing can still be wrong, insecure, inefficient or inappropriate for the system around it.
We therefore treat AI-generated code in the same way we would treat code from any other source: it needs to be understood, reviewed, tested and owned by the developer implementing it.
4. Repetitive and administrative work
Some agency work is important but repetitive. Organising notes, cleaning data, creating first-pass summaries, formatting information or preparing routine variations can consume time without needing creative judgement at every step.
Those are often good places for automation and AI assistance. The benefit is simple: less time spent on mechanical work and more time available for the parts of a project that genuinely need a person.
5. Content support
AI is useful for brainstorming topics, testing alternative phrasing, restructuring drafts and checking whether content is clear. It can help turn specialist knowledge into a more accessible starting point.
What it cannot do reliably is understand a business in the same way the people inside it do. It can produce fluent generic language very easily – which is exactly why unchecked AI copy can make different companies sound strangely alike.
We want content to sound like the business it belongs to, not like the tool that helped produce a draft.
What we do not outsource to AI
- Strategy. AI can contribute information and possibilities. A person still needs to decide what the business should actually do.
- Creative direction. Generating options is easy. Building a coherent identity, experience or campaign requires taste, context and judgement.
- Client relationships. We do not want communication with an agency to feel like talking to a machine. Questions, feedback and difficult decisions deserve a real person.
- Final quality control. Nothing should be accepted simply because an AI system produced it confidently. Work needs to be checked by someone who understands the brief and the consequences.
- Accountability. Software cannot take responsibility for a missed requirement, a poor recommendation or a technical mistake. We can.
Our rule of thumb
If a task is mainly about speed, repetition or exploration, AI may be able to help. If it requires judgement, responsibility, context or trust, a person needs to own it.
AI can be confidently wrong
One of the most useful things to understand about modern AI is that confident output is not the same as correct output.
A response can sound authoritative while misunderstanding the brief, inventing a detail, recommending an outdated approach or producing code that fails in a real environment. That is not a reason to avoid AI, but a reason to use it with appropriate supervision.
Experience matters because it gives you a way to evaluate the answer. If you cannot tell whether the output is good, fast generation does not solve the problem – it just lets you produce questionable work faster.
Faster should not mean lower standards
AI can reduce the time needed for certain tasks. We think clients should benefit from that efficiency. But the objective is not to squeeze as much output as possible into the shortest time. It is to spend less time on low-value repetition and more time on decisions that improve the finished work.
For example, if a developer can diagnose an issue faster, that creates more time to test the fix properly. If research can be organised more quickly, we can spend more time deciding what it means. If early copy ideas are generated rapidly, we can spend longer refining the message that actually makes it onto the website.
Efficiency is valuable when it creates room for better thinking.
What human-led looks like on a real project
The phrase human-led is easy to say, so it is worth being specific about what we mean.
- We listen before we generate. The project starts with the client’s goals, constraints and customers – not with an AI prompt.
- We choose the tools after we understand the problem. AI is used where it is useful, not added to a workflow because it is fashionable.
- AÂ named person owns the work. There is always someone responsible for the design, development, marketing decision or recommendation.
- We review before we deliver. AI-assisted work is still checked against the brief, brand, technical requirements and real-world context.
- Clients speak to people. Feedback, explanations and important decisions remain direct and personal.
AI and design: more options do not equal better design
Generative tools can produce visual ideas in seconds. That is useful for exploration, mood, references and quickly testing a direction.
But design is not simply the production of attractive images. A designer has to consider hierarchy, accessibility, brand consistency, usability, production constraints and what an audience should notice first. The most visually dramatic option can easily be the wrong one.
We therefore see AI as another instrument in the creative toolkit. It can widen the field of possibilities, but it does not replace the designer deciding which possibilities make sense.
AI and development: assistance is not ownership
The same principle applies to code. AI coding tools have become remarkably capable. They can accelerate routine work and can be particularly useful when exploring APIs, unfamiliar syntax or possible solutions to an error.
But a working-looking code snippet is not the same thing as a production-ready implementation. A developer still needs to consider security, performance, maintainability, browser behaviour, integrations, accessibility and the wider codebase. More importantly, someone needs to understand what the code is doing well enough to support it later.
We use AI to support development, but not as an excuse to stop understanding what we build.
Privacy and client information still matter
Convenience should not override common sense about data.
We do not assume that every document, dataset, login detail, private conversation or piece of client information belongs in an AI system. The right tool depends on the sensitivity of the task and the information involved.
That means thinking before uploading, limiting unnecessary data, using appropriate business tools and keeping normal security and confidentiality responsibilities in place even when AI is part of the workflow.
New technology does not remove old responsibilities.
Why this matters to clients
For most clients, the question is not whether their agency used AI to help with a task. The question is whether the work is good, accurate, appropriate and worth paying for.
Our approach is designed around that outcome. Used properly, AI can mean quicker research, faster problem-solving, broader exploration and less time wasted on repetitive work. Human oversight means those efficiencies do not come at the cost of context, originality, accountability or personal service.
We think that combination is more useful than either extreme: pretending AI has no value, or pretending it can replace the people responsible for the work.
Technology should amplify expertise – not replace the person accountable for the result.
Our approach will evolve. The principle will not.
AI will keep changing. The tools we use in a year’s time may be very different from the ones we use today, and we expect our workflows to keep evolving with them. We want that. A digital agency should be curious about better tools and better ways of working.
But the principle underneath them is unlikely to change.
Clients come to us because they need someone to understand a problem, apply experience, make recommendations and turn ideas into something that works. AI can make us better at doing those things, but it does not remove the responsibility to do them properly.
So our position remains simple:
People first. AI second.
We embrace AI as a powerful part of modern digital work. We use it where it makes us faster, smarter or more effective – but the thinking, judgement and relationship stay human.
Work with a digital agency that still talks to you
Enigmatio is a Scotland-based digital agency, remote by design and working with businesses across the UK and beyond.
We combine web design and development, branding, hosting and maintenance, and digital marketing – using modern technology where it adds value while keeping projects personal and human-led.
If you have a project in mind, tell us what you are trying to achieve. We will talk through the problem first and the tools second.
LET’S BUILD SOMETHING MEANINGFUL
Web design, development, branding, hosting and digital marketing – human-led, technology-enabled.