AI Is Transforming Software, Are Agencies Still Relevant?
Is it time for agencies to enter an existential crisis? Pat might have the answer.
Do digital agencies that build software have a future, with AI progressing the way it is?
It's a question that keeps me awake at night. I'm sure a lot of other agency owners will resonate with this statement.
The mad reality we are facing is that our entire workflow, and arguably our business model, is being turned upside down by AI.
We need to get comfortable with a world where large parts of the build process, both front-end and back-end, are completed by AI agents like Claude Code, and where clients can use tools such as Loveable to do a lot of the work that an agency would traditionally do.
However, an agency wielding AI in an experienced and robust manner will still be able to produce a much better result than a client wielding AI. This is where the value is.
As an agency owner, if you want to survive in this age of AI, you need to combine your traditional experience and skills with AI in order to elevate your output and maintain your value proposition.
There are opportunities throughout the project process, but let's focus on the design and build stage, as this is where the biggest changes are happening right now.
We now live in a reality where designers can write really solid code using AI, and developers can produce really high-quality interfaces, creating a growing overlap between teams.
That opens the door to much tighter collaboration. Imagine a developer spinning up a first pass of an interface, a designer refining it in Claude Code, then the developer validating and merging it. All within the same tool, iterating in real time, rather than designs living in Figma and being thrown over the wall to dev.
In some cases, dare I say it, you could even bring the client into that loop in a controlled way, allowing them to iterate as well. I think a lot of clients would really appreciate and enjoy that.
The process is going to become much more fluid, with small teams iterating in almost real time, working together to rapidly realise their vision.
And by keeping professionals in the loop throughout this process, and controlling the direction, the end result will always be far better than a client plugging away in Loveable.
We are actively reviewing this approach to ensure that we stay ahead of the curve as this change unfolds. Rolling this out will allow us to deliver much better products for our clients, much more quickly.
Also, you have to remember that a project is more than just design and build. You have:
Meeting the client, interpreting their requirements
UX research, including speaking to users, exploring user stories, designing user journeys, prioritising features, and scoping out the MVP
Validating everything the AI produces, from UX and code quality through to security
Getting that product in front end users on a regular basis, getting their feedback, and iterating and evolving in a focused way
Testing, testing, and more testing. Plus some extra testing
Deploying, supporting, and fixing things when they break
That is still a lot of work, difficult work, and it is all stuff that we can do better than our clients.
Right now, I'm optimistic about the agency model, as long as it can adapt.
Who knows what the landscape will be like in a month or two, though?