AI & I: The Architecture of Our Identity – Xin Yao | Craft 2026

June 04, 2026

AI & I: The Architecture of Our Identity – Xin Yao (Workshop Outline)

A Craft 2026 interactive workshop (~20–30 people, 90 minutes) by Xin Yao, a socio-technical architect who “goes around highlighting connections” (the yellow connector pen) and facilitates the first phase of transformations at companies “with a lot of people, a lot of software, and a lot of complexity.” The inquiry: who am I when knowledge — our former core competency — is being commoditized by AI? Rather than a talk, she models “unlikely, awkward, high-stakes conversations” experientially — “making knowledge relational, not transactional.” The workshop’s own arc: a fist-finger check-in → connection before content (the VIEW model) → a head/heart/gut identity architecture → a possibles / lovables / undiscussables conversation café → a themes-and-tensions harvest → integration. Participants have “100% choice and control over the level of vulnerability” throughout.

1. Opening — The Identity Inquiry

1.1 An experiment for contribution

Framed as a discovery experience plus a learning experiment; the design “could work for you or work poorly for this room.”
Everyone agrees to be “up for experiments today.”

1.2 Identity as the theme

In IT “we all work with identifiers” — identity is the distinguishing characteristic that “makes me me and Simon Simon and you you.”

1.3 Knowledge being commoditized

“We are all knowledge professionals” — we work with knowledge.
Per Gergely’s keynote and Wardley maps, knowledge is being pushed into the commodity range — including analyzing, rating, even creative knowledge activities.
The question: if these former core competencies / expertise are commoditized, “what is our defining characteristic tomorrow?”

2. Check-in — The “Collaborative Calculator”

Participants raise a fist and put up one finger per question they identify with.

2.1 Q1 — Daily AI use

Who uses AI on a daily / almost daily basis? (First finger.)

2.2 Q2 — Efficiency excitement vs. time-deprivation/addiction

Excited that AI saves time / efficiency gains, yet still feeling time-deprived and addicted.

2.3 Q3 — Cognitive boost vs. dependency/erosion

Excited about AI’s cognitive-boosting effects (smarter, more creative), yet worried about dependency and cognitive erosion.

Excited by solo efficiency (no longer depending on others for knowledge), yet missing the social grounding of talking it through with someone.

2.5 Q5 — Role/responsibility clarity

Feeling a lack of clarity about roles and responsibilities.

2.6 The result

Most people raise multiple fingers (some five, four, a few threes) — “thank you for your honesty.”
These ambivalences are exactly what the session unpacks.

3. Session Design and Logistics

3.1 Relational, not transactional

The goal: make knowledge relational, not transactional — the room’s configuration (identities, desires to know, ambivalence, willingness) will define the emerging experience.

3.2 The teaching aim

“Teach” (in quotes) how to model unlikely, awkward, high-stakes, high-sensitivity conversations — the elephants in the room — experientially, not theoretically.

3.3 The facilitator’s self-introduction

She sees herself holding a yellow connector pen, highlighting connections; she facilitates the first phase of transformation, then leaves the space to the company’s own people.

3.4 Logistics — bumblebees, presence, and decibel zero

90 minutes with a 5–10-minute break; “bumblebee” participation allowed (the slot bridges two talk slots, so leaving for Kent Beck’s podcast is fine) — but engage when present.
Recurring check: “Are you present?” — participants wave to confirm.
Signal: when her two hands meet, bring the room to decibel zero.
Pens and stickies are hidden under chairs; the room has no tables (organizers maximized capacity).

4. Connection Before Content — The VIEW Model

4.1 The first connection is to ourselves

“Connection” first evokes people, but the first connection we need is to ourselves — often lost even in keynotes.

4.2 The intention compass

Draw a compass on a sticky, put one word in the center for your intention in the session — how you want to participate/contribute.
Xin Yao’s word: “courage” — a subtle, vulnerable topic in a compact, complex format.

4.3 VIEW — introduced by Joe Hudson

VIEW (Joe Hudson) models our connectedness to ourselves; score your intention (1–10) on each — “want to be” is an intention, not a guarantee (like a Christmas wish list).

4.4 V — Vulnerability

Speak my truth even though it’s scary — “say the scary thing because it is my truth.”
Question: How willing am I to be seen in the room?

4.5 I — Impartiality

Have a clear want for the session without an agenda — not forcing it on yourself or others.
Question: How willing am I to be influenced / moved?

4.6 E — Empathy

Be present with another’s emotion without taking it on.
Question: How willing am I to listen just to get the other person?

4.7 W — Wonder

Curiosity without needing to know the answer.
Question: How willing am I to be in wonder and not “in the know”?

4.8 The trio VIEW conversation

Groups of three explore: “What about AI is making you feel a little weird or wild about being you right now?”
15 seconds of productive silence first; the person with the longest hair starts by asking the question; each person gets ~1.5 minutes while others practice listening.

5. Popcorn Debrief — “What Struck You?”

Rapid-fire (“microwave popcorn”) reflections.

5.1 The human core of fear, purpose, and anger

How ”humanly normal” it is to feel fear, purpose, and competing feelings — AI “strikes at the heart of who we are and why we do this work.”
For one participant it registered as anger — seeing AI as “an incursion of the things we are here to do.”

5.2 Not afraid of replacement (“maybe that’s foolish”)

Mati: “I’m not being afraid of being replaced by AI, but maybe that’s foolish.” (Short and sweet — “with a lot of butter.“)

5.3 Suspicion and the critical-thinking edge

A group’s shared feeling: not anxiety but suspicion — critical thinking should not be outsourced/overtaken or “you’re losing your edge.”
The fundamental question: should AI be a companion, or should you let it replace you?

5.4 Nostalgia for the flow of creating

A ”common sense of nostalgia” for “the golden age of Soviet development where you had the flow of creating something — now it’s gone.”
Xin Yao echoes her own shame/nostalgia for the good old days of focusing on one problem, thinking in the shower, taking a walk and returning with clarity — “those days are far and apart right now.”

5.5 There is no safe space until we make it

Facilitators often declare “this is a safe space,” but ”there is no safe space until we make it safe.”
What makes it safe: the first person who takes a social risk to be seen despite vulnerability — and the room scored high on vulnerability.
The VIEW sticky becomes the intention compass for the rest of the session.

6. Identity Architecture — Head / Heart / Gut

6.1 The tug-of-war in the mud

A tug-of-war: on one side, knowledge professionals want self-expression, creativity, and work-life balance; on the other, the world (per the OpenAI-flavored keynote) wants ruthless productivity, endless output, and endless adaptation.
We stand “in the mud holding the rope while AI shifts the ground underneath us.”

6.2 “Who are we when the machine learns our craft?”

Left in the air at a conference called Craft.

6.3 The Matrix — “Am I the one?”

The famous Oracle’s-kitchen scene where Neo asks ”Am I the one?” — a binary question with only two answers.
When identity is threatened, thinking regresses to binary (“will I stay relevant in the age of AI?”).
Neo lived a double life (programmer by day, hacker by night); Morpheus told him he’d save the world, but he’s anxious to leave his identity behind.

6.4 Grieving the old identity

Leaving behind an old identity of self-expression/creativity requires grieving — Neo won’t grieve “unless you tell me I’m the one.”

6.5 The four F’s — including “feed”

In fear we get fight, flight, freeze, and a fourth added for the age of AI: feed — instead of stopping when tired, “just one more prompt… that dopamine fix around the corner.”

6.6 The three-circle learning-loop trick

She invites imagining three empty circles (Venn diagram) — as knowledge professionals, “your brain does not want to leave those circles empty,” hooking attention (a learning trick).

6.7 Head — Capability

Head = capability: what I know, my role, craft, skill.
This is where fear-based marketing targets us — the flood of podcasts, AI newsletters, and courses you feel you must click “or my craft won’t develop.”

6.8 Heart — Capacity / Connection

Heart = capacity/connection: connection to self, others, and the bigger picture — impact, resources, support system, agency.
Companies with the best-skilled people can still deliver mediocre performance (she’s left such companies); startups with less money but real collaboration “move the needle.”

6.9 Gut — Core (safety and pleasure)

Gut = core: both safety and pleasure.
A coach’s line: ”you can only rise to the height of your lowest system” — if safety is threatened, the frontal cortex won’t activate.
Also pleasure: which parts (prompting, context engineering, linting, harnessing) do you do out of fear vs. enjoyment? Enjoyment feels light, soft, gentle (“I don’t want it to end”); non-enjoyment feels tense, heavy, weighty.

6.10 Craft is only ~1/3 of our humanity

The pulsating “who am I?” sits in the middle; craft/ability is only about one-third of our humanity — “that’s not the whole story.”

6.11 Identity as never-ending recalibration

Success with expertise (she teaches DDD / domain-driven design and architecture) tempts us to fix identity to expertise — but ”life is a never-ending process of identity recalibration,” painful on a visceral level.

6.12 Emotional fluidity underpins identity fluidity

Shedding our “snake skin” of past success requires emotional fluidity — the capacity to feel all emotions and let them flow.
Neuroscientifically, we make decisions to avoid uncomfortable emotions (the FOMO AI-course signup) — these are ”kinks in the water flow” that block us from seeking what we actually want, making us follow someone else’s model.

7. Room Configuration — Grouping by Role

7.1 Sorting into role tribes

No Slido/phones — done physically by moving to corners: technical roles, technical leadership roles, product and business roles, and coach/facilitator/organizational-development roles.
Technical roles dominate the room; there are quite a few product/business roles.

7.2 Self-organizing into trios

Self-organize into groups of three; there are 15 flip charts; claim a modeling space and wave.

8. Break Rituals

8.1 The star pose and golden-retriever shake

Star pose and “feel energy running down your body”; then shake like “a golden retriever coming out of the water” (taken metaphorically).
A note on the ephemerality of the “bumblebees” — “people change, we all evolve — identity recalibration.”

9. Conversation Café — Possibles / Lovables / Undiscussables

9.1 Framing — be selfish, no correct answers

Model your group’s exact configuration — “be super selfish”; don’t model roles not in your group.
“There is no correct answer” — feel the lightness/weirdness and enjoy it.

9.2 Possibles (blue) — roles and AI use cases

Like event storming / command storming: an event is “agent-prompted”; a command/use case is “prompt agent” — a verb in active imperative form.
Format: ”I, the [role], use AI to [verb] [object]” (e.g., “I, the PM, use AI to analyze user research”).
Process: 1 minute silent individual brainstorming (yellow = roles, blue = use cases), then 5 minutes group modeling; then mark use cases that can be collaborative (done as a team with the LLMs).

9.3 Lovables (green) — deeper need + emotion

A combination of user-need mapping and emotional inquiry.
L = logical: the deeper need beneath a use case, found via the five whys (e.g., “I use LLMs to answer emails” → why? → the real need).
E = emotional: ”I feel ___ when this need is met” (adventurous, accomplished, safe, boosted…), building an emotional vocabulary.
At least two greens per blue (one logical, one emotional).

9.4 Undiscussables (red) — the elephants

Pick use cases that evoke a worry, dilemma, or uncomfortable feeling.
Go beyond the socially acceptable to your authentic truth: “what makes you uncomfortable that you haven’t shared — maybe not even with yourself?”
30 seconds silent reflection, then a group conversation with curiosity and empathy switches on.

10. Themes & Tensions — Hypothesis-Based Harvest

10.1 The tension masters

Instead of a gallery walk, a hypothesis-based harvest, documented live by two volunteer co-facilitators: Simon (themes) and Alex (tensions) — “with legible handwriting.”

10.2 Hypothesis 1 — Productivity use cases

Increase productivity (product designers/managers — noted as sounding more like a need).
Documentation (assembling docs with AI).
Research / searching (user research, “finding things faster”).
Writing code and creating tests (a teacher/mentor role — for speed).
Debugging issues / incidents — faster resolution pulling data from graphs and analytics.
Break the effort wall / make people commit (a solutions architect) — user need: prevent procrastination; “excited to see movement.”

10.3 Productivity — lovable

Use case: find/search/analyze documents, emails, chats.
Deeper need: reduce time spent searching.
Emotion when met: feel effective, empowered, and less annoyed (“I’m annoyed spending half a day collecting things just to answer an email”).

10.4 Productivity — undiscussable: addiction

Fear that AI makes me addicted — using it even when not needed, ”let’s ask AI” as the default.
The tension: ”if you’re addicted and use more time, how did it save you time?”

10.5 Productivity — undiscussable: environmental impact

A PM raises environmental impact as “a really big debate” — conscious of it even for small tasks: “to save you two hours… says a PM.”

10.6 Hypothesis 2 — Cognition / creativity use cases

Planning features and ideas for automation (a tech lead).
Discovering different solutions.
Discover new architecture options (a software engineer — permitting easier implementation / bug-fixing).
Substitute other roles for creating POCs — outsource “less creative but still essential work” (e.g., UX text) to get there quickly.
Identify the proper problem statement / get inspired — deeper need: focus on the right thing (“doing the right thing vs. doing the thing right”).

10.7 Cognition — lovable feelings

When you “do the right thing,” the feeling is satisfying and confident (“my teammate always says ‘satisfying’”).
AI acts as a brain partner giving more options — but ”not embodied.”

10.8 Cognition — undiscussable: wasted effort and quality doubt

Fear of ”spending time for nothing and then regretting it” — that the output ”won’t be relevant” and “I should have done it myself without the machine.”
Reframed as insecurity/doubt about quality and inability to trust the output.

10.9 Cognition — undiscussable: the hollow feeling / cognitive erosion

More options first felt complete and confident, then hollow: ”am I losing my critical thinking capabilities?”
The recurring tension: “Am I getting dumber or smarter by being helped by the LLMs?” — cognitive erosion — “a tension I always experience.”

10.10 The secret agenda — AI-generated session notes

Xin Yao reveals she’ll photograph the sticky notes, feed them to AI, and send session notes so participants needn’t wonder what the stickies meant.

10.11 Hypotheses with no coverage

Decisions / agreement (agents have agency = “acting/deciding independently”) — no coverage.
Responsibility / ownership (trust often comes up in AI debates) — no coverage.
Teamwork / collaboration — no coverage.

10.12 Shared understanding — clarifying user intent

Architects: interview the client to clarify user intent and reach shared understanding; they use LLMs to prepare and filter interview questions.
Contrast: another group (in a different session) does all their user research with LLMs — “they’re not talking with the customers anymore.”

10.13 The isolation theme

Substituting other roles for POCs can lead to potential isolation — the industry requires working with product, designers, developers.
With AI you get reactive answers, not proactive opinions; ”opinions matter — it’s not facts,” there can be multiple truths and an array of perspectives.
You get “the LLM’s summary of diverse perspectives, but not from biologically embodied beings” you can debate — a clear social isolation risk.

10.14 Learning to stay relevant vs. cognitive decline

Multiple roles use AI for learning and discovery “to stay relevant” — which sits in contradiction with cognitive decline.

11. Integration and Close

11.1 The food metaphor

The session structure: consume (her short framing/model) → digest (modeling our “rich, messy reality”) → integrate (breaking nutrients into parts, then absorbing).
Thanks to Alex and Simon (tension masters) and Mati and helpers who set up the room.

11.2 Integration via silent reflection

No time for a full application-integration portion, so it’s done as silent reflection in a Google form / feedback form (QR code) with integration questions.
Those who leave contact info get the AI-generated session notes — an “Amazon-style narrative brief” (a four-pager, not a six-pager).

11.3 The closing question

”In the age of AI, answers are cheap — questions are becoming the differentiator.”
Parting question: ”If productivity is the trees the AI friends enable, what is the forest for you?”

People & References Cited

Xin Yao — speaker/facilitator; socio-technical architect; teaches DDD/domain-driven design and architecture; “the yellow connector pen.”
Joe Hudson — creator of the VIEW model (Vulnerability, Impartiality, Empathy, Wonder).
Gergely — keynote speaker referenced for knowledge being commoditized.
Simon Wardley (Wardley maps) — referenced for the commodity range of knowledge.
Kent Beck — the concurrent podcast slot participants could “bumblebee” to.
Mati / Marty — organizer/helper who introduced the session and set up the room.
Simon — volunteer “themes master”; Alex — volunteer “tension master.”
Named participants: Sophie, Augie (a participant whose time-saving need was referenced), plus the anonymous PM, tech lead, software engineer, and solutions architect.
The Matrix — Neo, Morpheus, the Oracle (“Am I the one?”), used as the binary-identity metaphor.
Concepts: identity architecture (head/heart/gut = capability/capacity/core); the collaborative calculator check-in; connection before content; the intention compass; the four F’s (fight/flight/freeze/feed); event storming / command storming; five whys; possibles / lovables / undiscussables; psychological safety (“no safe space until we make it”); emotional fluidity vs. emotional avoidance; cognitive erosion; agency; social isolation; AI-generated Amazon-style narrative brief; “answers are cheap, questions are the differentiator.”

Video: https://www.youtube.com/watch?v=rmuDjlZqVYo — Transcript via yt-transcript.sh; outline generated from the transcript.