Pioneering Design for AI
Driving human-centered artificial intelligence design for 350,000 IBMers—years before the LLM explosion.
In 2017, IBM was in the midst of multiple transformation initiatives, including design and AI. Alongside design executive, Adam Cutler, my team was the tip of the spear, working cross-functionally across the 350k enterprise to synthesize best practices, create net new artifacts, and drive adoption. We accelerated 60+ AI offerings, badged and upskilled 125,000 practitioners, 2x time-to-market, and provided a 301% return on investment.
As a designer with a technical background, my responsibilities spanned co-creating our unifying framework, user research, facilitation, content design, creating prototypes and design patterns with IBM's tools, taking ownership of our conversation and data literacy bodies of work, and more.
Background
IBM Design for AI sat within IBM Design. Accordingly, our mission was an extension of our parent team's. As a team of three, we all wore many hats; as the only technical one, I inherited many of the technical tasks.
Business & User Problem
Our business objective was to unlock the following proven benefits of Enterprise Design Thinking for AI solutions.
For many cross-functional teams, they lacked the requisite understanding of what AI was, and how they might use it to create solutions to address real user problems. It wasn't uncommon for a project of ours to begin with some variation of this quote.
We were told we need AI by next quarter. 😳
How might we...
Equip 350,000 IBMers, regardless of technical background, to design human-centered AI experiences—as a team of three.
Our Approach
For both data and conversation, I collaborated with experts across the enterprise, including:
- IBM Software Teams (Watson Conversation, Cloud, Security, Design)
- IBM Services Teams (Global Business Services, iX)
- IBM HQ (Finance & Operations)
- IBM Research
What We Created
To help users navigate this uncharted territory, we created three core bodies of work:
- One single unifying framework, called the Human/ Machine Interaction Model
- Practical, quick-start guides for designers (e.g. in conversation, data, Watson tools)
- AI ethical guardrails
My Personal Initiatives: Data, Conversation, & Prototypes
During my time on the team, I noticed two emerging themes where designers were concentrating their attention: data-rich applications and conversation. Designers needed a baseline understanding of these subjects, a shared vocabulary to use with their technical peers, and practical tools to get started.
For both data and conversation, I collaborated with experts across the enterprise, including:
- IBM Software Teams (Watson Conversation, Cloud, Security, Design)
- IBM Services Teams (Global Business Services, iX)
- IBM HQ (Finance & Operations)
- IBM Research
The Outcomes
- 60+ AI offerings accelerated
- 125k badged (2019-2024)
- 2x faster time to market
- 75% decrease in time for initial design and alignment
- Nice messages from strangers on LinkedIn (like Jeroen's below)
.svg)
.png)
In my 3 years on the team, the most salient takeaways were:
- The best way to learn is by teaching.
- Giving ownership builds buy-in.
- The power of synthesis.
.svg){kind=icon}
Let's talk details.
To save you from more of my yapping, let's make this a two-way conversation. Happy to share more details, links, assets, and behind the scenes info over a call.