Practices from the field.

A full-day workshop I designed and ran for a 130-person design organization at JPMorgan Chase. Designers worked in teams to generate AI-assisted solutions to four real financial use cases — not hypothetical scenarios, but live problems the Payments organization was grappling with. The core method was capability matchmaking: rather than asking designers to brainstorm freely, we gave each team a set of AI capability cards drawn from the AI Design Kit, developed with my collaborators' research on matching AI capabilities to human needs. The cards shift the conversation from "what can AI do" to "which capability fits this human need at this step." The outcomes from that day are now informing the future vision for the Payments organization.

A workshop I've run for 200+ people across JPMorgan Chase and Stanford d.school, built around the idea that the design decisions that matter most about AI aren't technical — they're about role, relationship, and trust. Participants don't learn how to prompt; they work through how to define what a good AI teammate looks like for their specific context: what it should own, what it should hand back, how it should communicate, and where the human stays in the loop. The session reframes AI from a tool you use to a collaborator you design. It's the workshop this site is named after.

A workshop I ran with our internal Community of Practice at JPMorgan Chase, drawing on the ritual design methods from my books (Rituals for Work, Rituals for Virtual Meetings). The premise: AI adoption fails not because of capability gaps but because teams don't build the shared habits and repeated structures that make new behavior stick. A ritual isn't a policy or a process — it's a meaningful, repeated practice that a group does together and makes their own. In this workshop, teams design their own ritual for integrating AI into their work: when they'll reflect on it, how they'll share what they've learned, and what signals will tell them it's working.

The simplest practice on this list: I pair up with a teammate and we build something together using AI — no agenda, no required deliverable, just a real project we're curious about and two people taking turns at the keyboard. One person drives (writes prompts, reviews output, makes decisions), the other watches and asks questions — then we switch. The pair dynamic matters: building alongside someone else makes invisible reasoning visible, surfaces assumptions you wouldn't catch solo, and is genuinely more fun. I do this regularly with a teammate at JPMC. Nothing built in these sessions has to ship.

An every-other-Friday gathering I now lead within JPMorgan Chase's design organization, after the original lead left the company. Three designers stepped up to keep it going, and we restructured it around one core format: people share inflight work — something they're actively building or thinking through with AI, not polished case studies or finished projects. Inflight work is harder to share but more useful to watch. The sessions create a low-stakes space where designers can say "I'm not sure this is working" and get real-time reactions from peers navigating the same territory. What started as a knowledge-sharing group has become one of the most honest spaces in the org for talking about what AI actually looks like in practice.