Embracing AI Operations

This is a tale of how my approach to design evolved over time from reactionary, to a deep appreciation for structured methodologies. An appreciation now critically important in the age of AI Operations - a role which aims to apply AI to enhance and automate how a business functions.

A long time ago…

When I first started in the world of work I was a web designer.

At the time it was seen as a role on the bottom of the pile - even though nobody else was capable of doing this particular kind of work. A strange dichotomy of opinions but, that’s common in business.

I had no process, no preferred approach. I was largely reactive to the brief at hand.

I hated the idea of doing something the same way every time. I felt it would be boring and kill my ability to learn new things.

I was young and impatient. Much to learn, I had.

When coding a website I’d be following the processes involved in CSS & JS development, but not really for design. Design is inherently messy, and my approach changed every time. When I designed anything I’d take whatever route I felt on the day.

I started noticing articles in publications like Smashing Magazine that talked about processes.

Once I said to my friend Westy, “processes are for computers, not people!”

“Hey, what’s your process?” - Well, I didn’t have one.

Perhaps that’s why I felt on the bottom of the pile - I was fooled by my own mind tricks.

The List Awakens

Once I was responsible for managing my own team of designers, I had to make a change.

Sometimes we must let go of our pride and do what is requested of us.

Everyone was working in different ways, components looked different and, despite similar abilities amongst designers, project work had wildly variable outcomes.

I had to update multiple Illustrator files when new components were designed, explain to everyone what had changed each time, and hope that they remembered to follow this approach.

It was manual, but it was a process. Part of a design process.

And it helped.

As my team grew and my staff changed, I needed to document these small processes for my own sanity as well as for everyone else to follow.

As a manager, it became a more relaxing way of working, to have it all externalised and ready to be referenced should I be on holiday.

I got really into this new methodology (new for me) and it branched into my own life.

I became the list man.

• I’d take interest in certain topics & capture details about them in my notes
• If a thing wasn’t on my list, then the thing wouldn’t get done
• If I made a list, everyone could follow it
• Always pass on what you have learned.

But, the ability to make a list is no match for the power of a framework…

The Double Diamond

At one point I hired a young French designer called Camille.

One thing that stood out in her portfolio was that each project had a little symbol in the corner showing what parts of the double diamond design process she had been involved in during that project.

I’d never seen this framework before & instantly had to learn more about it.

I learnt that The Design Council, a British organisation, created this process many moons ago.

It all made sense to me:

Discovery leads to Definition, Definition leads to Design, and Design leads to Delivery.

It’s less useful for small tasks, better for projects that can take a week or more.

There’s a lot more beneath that surface level description of its four main phases, which is where it gets interesting.

You can modify the Double Diamond to suit your needs - it’s only a starting point.

For example, in product development it’s clearly missing a phase after delivery where tracking user behaviour happens.

Using a framework like this brings an extra dimension to listing out steps of a process.

It’s an approach that’s brought new levels of consistency to many projects I’ve been involved in to this day.

A New Startup

In 2018, when Life Moments started, I had the opportunity to start a fresh way of working.

Since day one I began collating how I do things, partly through a healthy interest in enabling our team to thrive, and partly in an attempt to remember the myriad of skills needed to work in a small startup.

Design systems

It’s not just my design team that uses Figma. We’ve helped our entire business skill up & jump into the Figma workspace.

This makes a huge difference to our operational speed, and how processes can be designed:

• Figma is our source of truth for product development
• Every client product has its own project, named with a convention
• Each project includes standardised files for component libraries, page maps, and asset creation
• And it all stems from a core project that sets the precedents

We even use Figma for our slides (Figma Design not Figma Slides, to be clear) as so much of our body of design assets we’ve already created are reusable in sales decks.

And just yesterday I figured out a way exporting decks as PDFs but at 10% of the file size.

Immediately I documented that 5 step process, and made a screen recording to send on Slack letting everyone know that this was a great way of getting stuff done.

Where was this documented? In Notion, of course 😉

Notion - My Second brain

Ah, the second brain… It’s a commonly-used phrase, especially for Notion users.

I’ve built our design team handbook in Notion, and it’s a lovely, organised place to be.

We have connected databases for clients, projects, tasks, time tracking, hiring & onboarding, tools, a glossary, team roles & structure…

Without this I’d have to remember it all. And remembering things is not my strong suit so I find value in documenting even the smallest of processes.

In a small team such as ours, we each cover a handful of roles.

I’m responsible for design as a whole, but I’m also product managing, researching, coding, and managing people.

Roles lead to skills, skills lead to actions, and actions need processes.

Processes & documentation are going through the roof now, with productivity following suit.

Until AI became truly useful, it was all about documenting and improving manual work processes.

The AI Era

While not the earliest of adopters, I can lay claim to having used ChatGPT to help me build my website way before code editors like Cursor & Windsurf existed, and definitely before ‘vibe coding’ was a recognised term.

Around that time I played around with some image generation tools, & using Notion AI to help me build a skills matrix to assess my team in an objective way. Nothing too outrageous.

Then we started building RAG systems at Life Moments.

At first it all seemed like a backend developer’s world: Python code & mysterious concepts that we’d not worked with before.

I decided to get closer to the action to understand it better. I began visually mapping the information flow that was being coded.

It reminded me of an excellent article from Anthropic about prompt chaining, and all the different ways multiple LLM calls could be used to transform a few simple words into gloriously complex, yet (somewhat) controlled outputs.

I dug into the AI tech space, learning more about its history, emerging terminology, the frameworks that hold it all together, the tools that are built upon those frameworks…

I stumbled over a tool called Rivet that looked like my process maps, only it would actually function, outputting processed responses in a chain.

Rivet led to Langchain, Langchain led to Langflow, and Langflow led to to n8n

It set my mind alight! Now low-code automation builders could make use of AI.

This was a revolution.

The Dark Side

Ugh, YouTube… for all the incredible content it offers you will never find a more wretched hive of scum and villainy.

It all starts by searching for how to build a specific type of system like an n8n workflow that uses Slack as the front end.

There’s a lot to learn in this area, and so tutorials are welcome. But you need to ask yourself why this person has made the tutorial in the first place…

So many of these people are more ‘creator’ than AI Ops ‘specialist’.

They love to provide their credentials at some point in the video in the form of how they scaled their AI Automation Agency to over $100k per month - impressive, most impressive numbers.

They teach the obvious techniques, but not the golden nuggets of information, oh no. Those are behind the Skool community paywall.

And it’s that Skool community they’ve set up that forms both the reasonf or making the tutorial videos, and where they actually make all that money (assuming the numbers were actually truthful).

Nobody is making that much money selling automations. But they may well have 1,000 people paying them $100 each month via Skool…

But they use this revenue to sell the dream of running an agency. It’s all so scummy and needs calling out.

These people are fools trying to cash in on unsuspecting viewers.

But who’s the more foolish: the fool or the fool who follows him?

Back Down To Earth

Leaving all that stratospheric revenue behind, I returned to the real world & wanted to know more about this new industry from real people

Refusing to rely on the internet fakery, I spoke to founders who are making use of AI in their businesses, and discovered a job title that’s new to me: Go To Market Engineer (GTM Guy is the buzzword).

In fact it’s fairly new to everyone, Google Trends shows that people only started searching for this term at the beginning of 2025.

Typically these people are laser focussed on setting up cold outreach systems using a tool called Clay.com

These can also be created using n8n or Make, but Clay is a dedicated tool for data cleaning & personalising cold email, and so likely easier to sell the automation dream to clients than with a more generic tool like n8n.

This is what I find more interesting, the broader set of use cases that come from looking at any function in a business and applying AI to supercharge it.

Applying what I’d learnt to Life Moments operations revealed a completely different experience to how it’s sold up on YouTube.

• True use cases don’t just jump up & down waving a flag - you have to examine how people are going about their work to discover where AI can help.
• You can’t rely on pre-defined workflows that you have up your sleeve & throw them at people, it takes time & effort to understand their problems to solve before you can define something that will solve it
• The most interesting part of designing a workflow is the prompt engineering - the system instructions, the workflow structure, without this all you have is a bunch of empty nodes
• It’s much harder to build a robust workflow - catching errors, testing and refining the responses as they flow into each other, and ultimately checking to see if it does actually solve the problem or improve a team’s output

…and there it was again: the double diamond.

AI Operations can be approached using the same process as a design project. It’s just designing and deploying a system.

I think this kind of role suits those with a design background as there’s a significant skills overlap:

• Interviewing people about problems in their journey to a goal
• Mapping out steps of a flow
• Defining top level logic for how parts of a system interact

So far, for Life Moments, I’ve set up our automation environment, discovered a handful of use cases for various people, presented how AI automations can help grow our business, and am in the design stage for a few of them.

• Once delivered, on-brand images will be easy for any of us to generate using AI for sales decks and marketing campaigns
• Audits that once took a month of work will be 90% completed by AI in 5 minutes
• Plus research packs that follow our specific approach to personas, journeys, video scripts interactive tools, and how to engage customers will be generated by AI and saved in a central location

As a business we’ll be progressing so much faster, in a completely different gear. All thanks to AI Operations.

It's about designing, deploying, monitoring, and continuously improving these systems that power our work. Making them efficient, while maintaining our high quality standards.

Even at this early stage I can tell it’s been worth the effort to focus on evolving AI Operations in our business, despite the distractions of my other responsibilities.

Stay on target, stay on target…

Team Adoption

So far I’ve yet to discover a single person online talking about gaining buy-in to apply AI automations in a business.

This is where the tech meets humans, and it presents an ethics problem.

Automations are now so powerful, able to replace enormous amounts of manual work, and also, therefore, the need for humans to be employed.

When you’re the person advocating for the adoption of AI, you need to be aware that those doing the work manually may fear for their jobs, and see you as the aggressor.

If that’s what they think, they won’t want your newfangled operations anywhere near them.

“Hey, I can automate your job!”

“I’ve got a bad feeling about this…”

It could the beginning of a chain reaction that ruins culture and stifles AI innovation.

Is there a better approach to this very real and very human problem?

A New Hope

So, we have the Godfather of AI, the true experts, the early adopters, the overly vocal content creators, and their dogs.

They’re all saying that, unless you’re a plumber (or a similar role based in physical work), the robots are coming to take your job.

Well, maybe we should never underestimate a droid, but I don’t think it’s that simple.

For one thing nobody really knows. Predictions change weekly.

Hard to see - always in motion is the future.

I think AI will likely replace a good proportion of today’s roles, especially in the digital world.

But it will also create new roles, prompt engineers, agent managers etc.

Still, I think the replacement of roles is the wrong way to think about how businesses should adopt AI.

Cost cutting by replacing people with AI agents, while helpful, has finite impact. Once a team has been streamlined, there’s nobody left to replace.

My hope is that founders will look at the other side of their bottom line: what business growth lies in store when AI supercharges your existing human colleagues?

AI can make ten people feel like a hundred.

Here your opportunities are limitless, this is the way.

As AI continues to evolve at ludicrous speed, the real differentiator won't be how many jobs it replaces, but how strategically we leverage it to elevate human capability.

If you're building a business, consider not just where AI can cut costs, but where it can amplify your team's unique strengths and unlock new levels of growth.

How might AI empower your people to achieve what was once impossible?