AI Design Process

Having delved into ”vibe coding” and observed agent behaviour and performance on many ai software creation tools, I can safely say that AI tools are now a mainstay in my entire design workflow, whether it be day1 discovery or detailed prototyping, I am using AI at almost every stage to;

1) enhance quality of ideation and the output
2) drastically reduce cost and time for organisations I work with
3) to upskill for the future and ensure my teams stay on the cutting edge of possibility

At Northern Trust, I was recently teaching teams to use Lovable.dev ( a full stack, ai software generation tool) to prompt their way to quality, accurate, coded prototypes in minutes. There of course are many learnings and skills with communicating with the agents inside these tools, that I wrote about, when I released my first vibe coded, deployed voice notes app in December 2024, view my observations here.

Enterprise teams that adopt an AI first design process, are likely to increase release velocity and therefore growth, as deploying real coded software is now easier than designing in figma, meaning rapid experimentation at a greater fidelity can now take place at scale across organisations, thus the feedback loop and real build costs are drastically reduced, with the only barrier, software adoption and light training, both of which I advocate for and run, effectively.

As a technology leader, it’s clear the same can be said on the engineering (cursor) and product management pillars, where there are overlaps in certain tools, (Lovable claims to be full stack), and so with this, new ai teams and an ai software strategy is now possible, something I am also exploring.

Curious? check out a version of “Radar” that I created in Lovable in one hour, (here)

(a greenfield risk intelligence UI that took approx 18 months to research, build and launch the old way)