UPDATED: 01/26
Three workflows I swear by for getting work done
AI TOOLS
PRODUCTIVITY
AI hasn't replaced my job. It has made me better at it. The hours in the day are the same, but I can now focus on the interesting problems instead of getting stuck in the tedious ones.
Here's the thing though: AI isn't automatic. It needs judgment, oversight, and the ability to know when to trust it or question its results. My work has settled into three workflows: content and research, rapid prototyping, and image generation.
Content, Research, & Design Thinking
At IBM, a lot of my job involves turning "engineer-speak" into something designers can actually use. PM specs range from being very technical to just plain vague. I used to spend hours deciphering them. Now I have help.
For anything proprietary, I use watsonx—IBM's AI platform—to summarize GitHub tickets and reshape specs into design briefs. For research, I've tested Gemini, GPT, Claude, and Grok. Grok consistently surfaces better sources, which I then feed into NotebookLM to create mind maps or podcast-style discussions. When I became the lead for Data Replication (a domain I knew nothing about), this method saved me.
Claude has become my primary workhorse—great for microcopy in IBM's voice, but also as a critique partner. I've gotten good at prompting for honest, non-sycophantic feedback by sharing rough sketches instead of polished work. It won't replace a veteran designer's eye, but it catches things I miss. For larger initiatives, I create dedicated projects in Claude with persistent context—more upfront work, but far more coherent responses.
Rapid prototyping & building
This is where AI has changed what's possible the most. When I needed to present a version of DataStage as a concept, I didn't have weeks. I had hours. My workflow connects different tools. I begin in Claude, feeding it all my requirements. It turns these into a detailed PRD, structured and ready for the next step. That PRD transforms into a Figma design spec, which I then input into Figma Make to create an initial prototype.
Does the output need refining? Always. But the main structural elements are already there. What would have taken days of wireframing now happens in hours. More importantly, I used that same DataStage prototype to get the whole team aligned around a shared vision. There’s no substitute for showing something tangible—describing an interaction isn’t as impactful as letting someone experience it.
The gap between concept and execution has never felt smaller. Honestly, it has made design fun again.
Image generation & visual assets
I’ve worked on projects requiring custom character and asset generation—work that used to mean hiring illustrators or using generic stock images. Image generation tools are now very powerful, but they have one frustrating weakness: consistency across multiple images. You generate a character, love it, then it looks completely different in the next frame.
I’ve come up with a workaround. After extensive prompting to find the ideal asset, I generate multiple views of the same subject—front, side, and three-quarter angles. These multi-angle references, combined with contextual information for each new generation, produce a remarkably consistent set of images.
Using Nano Banana Pro, I’ve created character assets that maintain visual coherence across entire projects. Going further, I’ve fed these multi-angle images into Meshy.ai to produce 3D assets, which I can then animate. What would have required a 3D artist and a substantial budget now happens within my own workflow.
This isn’t about taking the place of specialized creative professionals. It’s about expanding what’s possible during exploration and concept phases—creating assets for internal presentations and prototypes that otherwise wouldn’t exist.
So, what's changed?
The new challenge isn't about mastering any single tool. It’s about adjusting to a changing environment quickly—even when that involves using tools that don’t come naturally. Being curious and genuinely open to trying new things has helped me more than any specific technique.
But here’s what I didn’t expect: sharing ideas with friends and colleagues has sped everything up. Because creating something no longer feels as tedious, people are more willing to actually try out ideas instead of just discussing them. There’s almost no reason to keep great ideas to yourself anymore. Just share them. Start building together. See what happens.
That shift—moving from protecting ideas to openly exploring them—might be the biggest change AI has brought to how I work.