Welcome

Here's everything you need for your day with the Assembly.

The event is designed around conversation, and the room will be set up to facilitate participation and engagement. Please review the context section for an overview of questions submitted by the group when registering. The discussion will be off the record, following the Chatham House Rule.

Arrival

Doors open at 12:30 PM. The event begins at 1:00 PM. Coffee and light refreshments will be available during the event, with reception from 5:30-7:00 PM.

Your onsite contact is Maggi Richmond at +1 (650) 996-7739 if you need anything.

Venue

Four One Nine

419 10th St, San Francisco, CA 94103

When you registered for the Assembly, we asked the following question:

What would you'd like to get out of the conversation, and what can ryou bring to it? What is capturing your imagination at the intersection of design and AI? What are you working on now in this space?

Your answers were thoughtful and a rich source of insight into the issues and opportunities facing designers today. We took those answers, stripped them of any identifiable information, and processed them with ChatGPT 5.5 Pro. The generated output is below.

— Jeff

The Assembly: Attendee analysis

This report summarizes what the registration data suggests about the people in the room and the questions they are bringing into the day. It is intended as a conversation primer, not a survey report. The registration pool contains 96 records; the actual invitation-only room is expected to be closer to 80 people. In the dataset, 84 people supplied a response to the open-ended “Question or project” field, and 84 people selected at least one “Conversations” interest.

No attendee names or attendee organizations are used here. Organization type, sector, and size are inferred from the registration fields and should be read as directional rather than audited classifications. Open-text themes are paraphrased and grouped to preserve privacy.

What kind of room this is

The registration pool is unusually senior and unusually close to the work. It is not only a room of design executives talking about AI from a distance; it includes founders, design and product leaders, design engineers, educators, investors, researchers, and independent practitioners who are actively building, teaching, funding, or reorganizing around AI.

The strongest pattern is a blend of two perspectives that do not always sit together: scaled organization leaders who are trying to bring coherence, governance, and training to large teams, and builders in smaller AI-native environments who are experimenting directly with agents, voice, model behavior, code, and new product forms. That mix should make the day especially useful if the conversation stays specific: what has changed in the work, what remains hard, and what new practices are emerging.

Seniority

A useful shorthand: roughly three quarters of the registration pool are either founders, C-level/VP/head-of-function leaders, partners, principals, or directors. This gives the event a leadership-heavy center of gravity, but the presence of builders and design engineers should keep the discussion connected to actual practice rather than only strategy.

The core audience is design-led, but not narrowly design-department-led. Many people are coming from product, engineering, operations, venture, research, education, and founder roles. That matters because the registration responses repeatedly ask whether design remains a separate discipline, becomes more strategic, merges with product and engineering, or becomes a more fluid capability across builder teams.

Organization type and scale

The sector mix is similarly broad. The largest groups are scaled platforms and enterprise institutions, AI-native startups and new product companies, and design consultancies or independent practices. Smaller but important groups include venture, education, research, publishing, healthcare, finance, climate, accessibility, and other regulated or high-stakes domains.

That mix sets up a useful tension for the day. Startups can often move quickly into AI-native workflows and product forms, but they are still searching for durable patterns. Larger organizations have sharper questions about training, safety, coherence, governance, design systems, brand, and organizational adoption. Educators are asking what must be taught now. Investors and advisors are looking for the shape of the market and the next product categories. Independent practitioners are asking what this means for craft, judgment, and identity.

What people selected for conversation

The “Conversations” field was a multi-select field. Among the 84 people who selected at least one option, interest is broad rather than concentrated. The average respondent selected 3.77 conversation topics, and the median respondent selected 4. Only 9 respondents selected a single topic. Thirteen selected all six.

This matters because it suggests the group is not approaching AI and design as one isolated issue. People are linking role change, creative capability, organization design, new interaction idioms, agents, and open exploration into one connected agenda.

In plain language: attendees are not only asking what AI tools can do. They are asking what designers become, how teams reorganize, and how to talk honestly about uncertainty while the ground is still moving.

Themes from the open-ended responses

The open-text responses were coded into overlapping themes. Because one response can mention several ideas, the counts below are multi-label and should not be added together. They are best read as a map of recurring concerns rather than as mutually exclusive categories.

1. The role question is really a boundary question

The most consistent underlying concern is not whether designers will use AI. The room seems past that. The deeper question is what the boundaries of design become when AI can generate artifacts, write code, create prototypes, produce variations, and participate in research or decision-making.

Several responses ask, in different ways, what happens when old role boundaries collapse. Some frame it as design, product, and engineering blending into a new builder role. Others ask whether design becomes more strategic because execution is cheaper. Others worry that the traditional discipline may become less visible precisely when design judgment is more important.

One recurring distinction is between output and authorship. If AI can produce plausible artifacts, then the designer’s value may move upstream into intent, taste, framing, systems thinking, evaluation, and the design of the process that creates the artifact. A few people describe a shift from critiquing finished work to critiquing the protocols, tools, and workflows that generate work.

This is a good thread to carry through the day: when execution becomes abundant, what becomes scarce? Possible answers in the responses include taste, judgment, coherence, trust, framing, agency, ethics, context, and the ability to decide what should exist in the first place.

2. AI adoption is being experienced as organization design

Many attendees are not just experimenting personally; they are trying to lead teams through the transition. Their questions are about training, management, career paths, team structures, hiring, change management, and the emotional reality of designers who may feel both excited and threatened.

A few responses make the point that AI exposes existing organizational health. In a strong organization, AI may amplify collaboration, learning, and creative output. In an unhealthy one, it may amplify confusion, fear, inconsistency, or low trust. That is a useful lens for the “AI at Scale” conversation: adoption is not just tool deployment. It is a stress test for the operating system of the team.

Attendees seem interested in practical examples: how leaders are training teams, what new rituals or reviews are working, how teams maintain shared standards when output volume rises, and how design managers coach people into new capabilities without reducing the conversation to productivity.

3. The daily workflow is becoming the laboratory

A large share of responses focus on concrete workflows: tool stacks, design-to-code, prototypes, internal tools, agents, AI-assisted production, and the changing path from concept to shipped software. There is clear appetite for honest reports from the field rather than polished predictions.

People want to know which parts of the process are actually better, which parts are only faster, and which parts have become harder. The repeated questions include: how designers get into the codebase; how AI affects design reviews; how teams use AI across product, brand, marketing, research, and operations; how design systems become readable by AI; and how prototypes move closer to production.

A useful framing for the demos and tool conversations: the group is not only evaluating tools as productivity aids. They are asking which new workflows, team shapes, and product categories become possible because the tool exists.

4. People are looking beyond chat toward new interaction idioms

The open responses repeatedly push beyond chat as the default interface. Attendees mention voice, agents, physical interfaces, sensor data, adaptive experiences, dynamically composed interfaces, capability-based products, and model behavior as a design surface.

Voice comes up as a particularly vivid example because it makes familiar interface assumptions less useful. In voice-first systems, pacing, interruption, latency, emotional register, turn-taking, and trust become primary design materials. Physical AI and AI-mediated environments raise a related question: when the system is interpreting the world rather than merely displaying information, how do people understand what it sees, what it knows, and when to act?

This theme aligns closely with the “AI as Material” session. The interesting question is not simply “what should the interface look like?” It is “what are the properties, constraints, and grain of model-shaped systems, and how do we design with them?”

5. Trust, quality, and coherence are the counterweight to speed

A strong thread in the responses is concern about what happens when production becomes cheap. Attendees mention trust, cognitive load, brand consistency, craft, accessibility, human agency, bias, security, and the risk of generic output. The concern is not nostalgia for slower work. It is the practical fear that abundant output can make systems less coherent unless the design function changes accordingly.

Several people are thinking about AI in high-stakes settings: health, education, finance, industrial operations, energy, climate, accessibility, and physical environments. In those contexts, “good enough” output is not enough. The design challenge becomes knowing when AI should step in, when it should ask, when it should stay quiet, how it should reveal uncertainty, and how people can remain in control without being overloaded.

A useful provocation for the day: if AI multiplies design output by an order of magnitude, what mechanisms multiply design judgment by the same amount?

6. Design systems and brand are becoming machine-readable problems

A subset of responses focus on coherence at scale: brand systems, design systems, interface patterns, quality, and consistency. The new question is not only how humans use a design system, but how AI can understand and operate within one.

This shifts the design system from a static library or governance artifact toward something closer to a production grammar. If models are generating prototypes, UI, content, and code, then design systems need to express constraints, intent, usage patterns, accessibility expectations, and brand behavior in ways that machines can act on.

That is likely to be a productive bridge between the scaled-company leaders and the tool builders in the room.

7. Education is both external and internal

Educators are rewriting curricula, and company leaders are building internal AI learning programs. These are two versions of the same problem: what should designers learn now?

The responses suggest several candidate literacies: model behavior, prompt and context design, prototyping with code, systems thinking, evaluation, AI ethics, taste under abundance, designing for uncertainty, and collaboration with non-human actors. There is also an implied question about what to stop teaching, or at least what to teach differently, if execution-oriented skills are being transformed.

The “Training the Next Generation” session should resonate beyond formal education. Many leaders in the room are trying to retrain the current generation at the same time.

Questions to bring into the room

These questions are designed to make the day more concrete and participatory.

1. What has AI made genuinely better in your design process, not merely faster?
2. Which part of your workflow has become more confusing, fragile, or harder to govern?
3. Where are designers moving upstream into strategy, systems, tooling, evaluation, or model behavior?
4. Which role boundary has usefully collapsed, and which one still matters?
5. What should remain deliberately slow, human, or high-friction?
6. How do you maintain taste and coherence when variation and production are cheap?
7. What does a design review look like when the thing being reviewed is a workflow, agent, prompt chain, evaluation, or protocol rather than a static artifact?
8. How should design systems express intent so that AI can generate within them rather than merely imitate their surface?
9. What are the new failure modes of AI-native experiences, especially in high-stakes domains?
10. What should a junior designer learn this year that would not have been on the curriculum three years ago?

The main readout

The registration data points to a group that is optimistic but not naive. People are excited by speed, new tools, and the possibility of entirely new product forms. But the center of gravity is not “AI will automate design.” It is closer to: AI is forcing design to become more explicit about its own value.

The room is asking how design changes when the artifact is no longer the scarce thing. That question touches every part of the agenda: frontier practice, scaled organizations, new tools, education, and AI as a design material. The most useful conversations will probably be the ones that connect practical workflow changes to deeper questions of judgment, trust, taste, organizational health, and what designers are uniquely responsible for next.

Appendix 1: Conversations statistical method

The “Conversations” field was analyzed as a multi-select variable. Each non-empty cell was split on commas, trimmed, deduplicated within a respondent, and one-hot counted by option. Percentages are reported against both conversation respondents and the full registration pool to make missing data visible. Pairwise co-occurrence was analyzed using raw co-selection counts, share of respondents, Jaccard index, and lift against an independence baseline.

Key quality checks:

• Total records: 96
• Non-empty conversation responses: 84
• Missing or blank conversation responses: 12
• Response rate for the field: 87.5%
• Unique options observed: 6
• Mean selections per respondent: 3.77
• Median selections per respondent: 4

Top pairwise co-occurrences:

Option A	Option B	Co-selected count	Share of conversation respondents	Jaccard index	Lift vs. independence
Open exploratory discussion	The evolving role of designers as AI accelerates	43	51.2%	0.61	1.11
AI as a force multiplier for creative capability	The evolving role of designers as AI accelerates	43	51.2%	0.57	1.02
AI as a force multiplier for creative capability	Organizational and team structure shifts	42	50.0%	0.59	1.11
Organizational and team structure shifts	The evolving role of designers as AI accelerates	41	48.8%	0.56	1.06
AI as a force multiplier for creative capability	Open exploratory discussion	38	45.2%	0.51	1.00
New design idioms enabled by model capabilities	The evolving role of designers as AI accelerates	37	44.0%	0.54	1.13
AI as a force multiplier for creative capability	Agent-based products and systems	36	42.9%	0.54	1.16
New design idioms enabled by model capabilities	Organizational and team structure shifts	35	41.7%	0.54	1.18

Appendix 2: The prompt

I'd like for you to do a thorough analysis of the attached data set. It is registration data for an event that I am organizing. I'm leaving the names out to protect personally identifying information, but I have included the job title to give you a sense of seniority, and the company so you can determine sector and size.

First, I'd like for you to write a script that quickly analyzes the "conversations" column, which was a multi-select field that gauges interest in the session at the event. your script should be a statistical summary of those selections using proper data analytics best practices.

Then, I'd like you to conduct a semantic analysis of the "Question or project" column. You can frame your analysis in both the overall description of the event and the agenda, both of which I'll paste below.

The resulting report is going to be distributed to the attendees prior to the event, to give them context on what the group is thinking right now. Note that this is an invitation only event that will only have about 80 in attendance. The intent is to facilitate discussion and sharing rather than simply watching presentations. You should write to help prepare them for the day.

Let's start with an overview of the attendees in broad categories: seniority, roles, sectors of companies, etc. then, let's get into the themes you pick out of their responses. Paraphrase when you want to quote, and do not use the names of companies, as we do not have permission to share their responses.

Please format the result in a markdown file I can download.