Full Steam 04: When to trust AI design (and when not to)
Full Steam is an ongoing conversation about the craft of B2B marketing. In each episode, experts from Iron Horse explore a single topic in depth, giving you practical takeaways and a fresh perspective on the challenges you're navigating every day. Catch up on previous episodes about AI in B2B marketing, Claude Cowork, and AI for content marketers.
The real skill in AI-assisted design is knowing where to use it
Design trends are evolving faster than ever, and so are the AI tools that emulate them. Now B2B marketing teams can generate wireframes, slide decks, and images with minimal effort. But where does that leave the designer?
In this episode of Full Steam, our principal content strategist Alex Jonathan Brown sits down with associate creative director Jenny Creed to talk about where AI-generated design works best, and where the work suffers without a human touch.
They discuss how to enable your non-designers to create assets without having to be the design police, why large B2B projects stretch the limits of AI tools’ capabilities, and why Jenny thinks the next big design trend will be touching grass.
Featuring
- AI speeds up design, in a good way. Today’s designers should embrace AI for day-to-day tasks like building slides or generating quick ideas for wireframes or persona journeys.
- But humans still need to do the thinking and verifying. A lot of AI-generated content is still “good from far, far from good.” Check for issues like AI hallucinations and poorly resized banner ads.
- The shelf life of websites has shrunk. AI tools and design trends move faster than ever, meaning today’s trends age quickly. Redesign your website every two to three years to stay current.
How should B2B companies use AI in their design process?
Use AI to generate starting points and handle repetitive production work, then have a designer review and refine the output.
For assets like slide decks, ads, and social graphics, Jenny recommends writing detailed instructions up front. Make sure to focus on accessibility, paying attention to font sizes, color contrast, and hierarchy so the output is usable without a designer touching every file.
Keep in mind that the ultimate goal is consistency, not perfection. “If someone is going to be looking at an asset for two seconds, we’re not going to slow down the process because the right corner radius isn’t the right width,” Jenny says. “If it reflects the brand, it’s accessible, it has the right logo, it has the right look and feel, it’s good enough.”
AI can also be valuable in the early stages of a project, where you can quickly generate multiple wireframe options from a single content document. But that output still needs to be checked against real user journeys, business requirements, and brand standards that AI doesn't have full context on.
What should designers let AI do vs. do themselves?
Use AI in web design for the production work, and keep the tasks that require human judgment.
If it’s repetitive, time-consuming, and has clear parameters, let AI take it. If it requires understanding context, brand nuance, or real user behavior, the designer should handle it.
On the production side, Jenny advises using AI for building out design systems and generating first-draft wireframes. The time you save can go toward work that requires human thinking, like creating better branding, more considered layouts, and more functional sections.
How is AI changing the B2B website redesign process?
AI is speeding up certain steps, but for complex B2B sites, it hasn’t replaced the fundamentals.
For simple sites with a straightforward purpose, AI can handle a lot of the heavy lifting. For enterprise B2B sites with multiple user journeys and conversion goals, the general process remains the same: content, wireframes, design, and development, in that order.
AI helps you move faster within a stage, especially with previously discussed tasks like generating wireframe variations, design systems, or persona journeys. But skipping entire steps will create more work (and take more time) in the long run.