Scaling for AI
Twenty years ago companies were chasing design. The iPhone had just come out and Apple was showing the world how design drives growth. How design makes people desire your product. But most organizations had no idea what made Apple's approach to design so effective. Today companies find themselves in a race again, this time it's chasing AI.
According to a recent report from McKinsey, while over three-quarters of organizations surveyed utilize AI in at least one function, only 1% report having integrated AI well enough to generate enterprise-wide value¹. In a subsequent report² McKinsey goes on to say that the ROI for AI reveals a clear productivity paradox: while adoption of AI is near-universal across organizations of all types, the bottom-line impact remains stifled: "Most firms see minor or localized cost-and-revenue benefits rather than broad profitability shifts."² From that same report McKinsey also notes that over 90% of those organizations exceeded their budgets for AI due to increasing operating costs related to scaling AI. Their research indicates that pursuing the newest model, or spending more money, is less effective than fundamental process changes. These findings parallel how companies approached design; wild investments with disappointing outcomes.
I was recruited by SAP to lead an organizational transformation program focused on design, specifically returning SAP to being passionately focused on their customers and using that passion to shape a new generation of SAP products that were intentionally designed for engagement. Over the course of 4 years my team delivered 100+ projects, trained 3000 coaches, and redesigned SAP's product development lifecycle end-to-end. Today more than a decade later, SAP's design organization is globally recognized and design continues to drive their innovation and market leadership.
Today companies are facing a nearly identical challenge with the adoption of AI.
Using a federated coaching network is a proven way to ensure success.
AI is repeating design's fifteen minutes of fame
For design rather than throwing money at the problem, companies threw headcount. However there were not enough professional designers so recruiters had to settle for bootcamp alumni or even self-taught "designers." Because it's so easy, anyone can design. The results speak for themselves.
In short, AI has yet to prove its value because organizations keep throwing AI at everything instead of making well-considered investments. Real investment means scaling up teams' mindsets, skills, tools, and shared best practices, all reinforced with outcomes-based incentives.
Those 90% of companies outspending their budgets are spraying AI at their products like a firehose. Their teams are left to their own devices to learn what AI is good at and what it's not. They're incentivized by the percentage of their work generated by AI, not the outcomes their work delivers or fails to deliver. There are no shared standards for assessing generated outputs, so individuals are left to apply their personal judgment.
People talk about humans in the loop, but what does that mean if they start by asking AI what problem they should solve? Followed by prompts to generate ideas for how it could be solved? And then prompts to generate and test the code? The human isn't in the loop; they're at the mercy of it. How can you expect the people on your team to have any basis on which to determine if what's being generated is in fact good? If it's really solving the right problem? If the solution is actually meaningful to your customers?
The human isn't in the loop; they're at the mercy of it.
How can you expect the people on your team to determine if what's being generated is in fact good if they have no basis on which to make that assessment?
This is the same problem organizations had in adopting a design-led approach. Design was not unknown to SAP, but they were using procurement as a proxy for users, and sales as user research. Rather than treating design as a way to understand and solve the people's actual problems, they were using it to simply paint the screen.
Don't spray and pray
At SAP we built a federated coaching network. Rather than simply tell everyone to go read a couple of design books, or check out the latest videos from IDEO, or just go design something, we assembled a team intentionally focused on organizational transformation. A team that could led by example, demonstrating design’s full potential. And who could provide the training and tools so others could understand design as problem solving not aesthetics. A team who could provide ongoing support to a community of design coaches.
The core team itself was comprised of people from a range of disciplines (engineering, program management, user research, UX, visual design, etc.), but more importantly it included both long-term SAP employees and experts from outside the company. Having people on the team who knew SAP's organization, understood its politics and its players, and had deep connections to its culture, was critical for our ability to effect meaningful change. Externally we recruited people who understood design at a systems level, and who could effectively coach others to adopt design as a problem-solving approach rather than "screen painting."
We leveraged data-driven, creative problem-solving methodologies to ensure the right problems were being solved in the most meaningful ways possible. While we called the method design thinking it was not the IDEO version; it was a modified approach that was scaled to address the needs of enterprise and tailored for the culture inside SAP.
Importantly, the team did not operate in a silo: they were constantly seeking out and engaging with teams across SAP. We took a three-pronged approach:
Lighthouse Projects: high-profile projects that would demonstrate design's impact on the business by delivering greater customer value.
Training and Tools: series of focused workshops to demonstrate how to use different design tools to cover everything from discovery to prototyping.
Coaching Network: following models from Procter & Gamble, Steelcase, Kaiser Permanente, and others, our real goal was to scale design by establishing a federated network of coaches that would support and reinforce the change management program.
Lighthouse Projects
I was engaging with the Executive Board and various GMs to identify high-profile initiatives that we could use as lighthouse projects. As a result, we redesigned contracts, performance assessments, and compensation policies. We redesigned how recruiting was managed. We also redesigned product suites, go-to-market programs, etc. We also redesigned the product development lifecycle, the customer sales centers, and the physical office spaces where teams worked. We prototyped SAP AppHaus and paved the way for the establishment of a Chief Design Officer, helping embed design-led, customer-focused development into every product group globally.
Training and Tools
The training started with a two-day workshop that ran teams through process, learned the vocabulary, and basic flow (discovery, synthesis, ideation, prototyping, validation, etc.). This gave them both the foundation and gave my team the ability to spot potential issues (one engineer refused to come out of the bathroom at a customer interview). We would then assign 1-3 members of my team to a project team for its duration. They would coach the project teams, helping plan, prep and run the projects. Coaching them through the project lifecycle, to be clear these transformations are behavioral changes requiring support, inspiration, accountability, and encouragement.
The toolkit was a constantly expanding resource, covering:
Inquiry frameworks for framing and reframing the problem
Facilitation techniques for seamless collaboration within the development team
Participatory design practices for including customers and partners
Prototyping tools, techniques, and storytelling best practices
As people discovered new tools, we would collect and refine them for inclusion in the toolkit.
Federated Coaching Network
During the course of coaching the project teams, we would identify and recruit individuals with a strong passion and aptitude for design-led customer-focused development. My team would work with these people to gain deeper knowledge of the methods but also facilitation and coaching skills. These people would then be certified to run their own design-thinking workshops and to help project teams on their own, with support as needed from my team members. And most importantly, this first set of coaches was tasked with identifying the next cohort. By repeating this approach, we were able to train 3000+ of these coaches within SAP in 18 months.
Outcomes
AI adoption needs to be guided if it's to be effective. Companies should create their own transformation team to build a federated coaching network that is focused on demonstrating, refining, and developing the skills for using AI effectively. It's simply not enough to let people experiment on their own. AI is a paradigm shift that requires guidance and coaching. More specifically, that guidance needs to be scaled to fit the organization both in regard to its objectives and its culture.
Understand the culture and its willingness to adopt meaningful changes
Find an executive champion
Intentionally reflect on the type of people you need on your core team
Start building your living toolkit
Define and get alignment on your metrics
Define a plan for sharing your progress
Note: Focus the messaging on others' success not your team’s
Identify key lighthouse projects
Tap first cohort of coaches
Set up community services for the coaches (comms, resources, etc.)
Building a federated coaching network doesn't require a heavy investment. In a short time you will have your first cohort that can work with other teams. But it will help ensure you have the means to measure your investment in AI, and be able to effectively refine and improve your organization's utilization of AI over time.
1.) “Superagency in the workplace: Empowering people to unlock AI’s full potential”, McKinsey, Jan 2025
2.) “The state of AI in 2025: Agents, innovation, and transformation”, McKinsey, Nov 2025.
