Trigger: Director, Product Design - AI in Experience Design (R249772) reposted August 21. Five weeks live and unclosed as of September 26. The listing is active on Workday and Dohan's hiring post remains up. Confidence: high.
Window: You are mid-window. Five weeks on a Fortune 100 Director search is standard enterprise cadence. The role has existed in some form since May, which means the committee has had time to learn what it actually needs — whoever walks in next faces a more refined evaluation. Act in the next two weeks. Enterprise searches at this level typically run 8–12 weeks per posting cycle before pausing or resetting.
What five weeks tells you about the evaluation gap
The prior battlecard established that this role's mandate is the design org itself — defining how Capital One's hundreds-person XD organization works with AI. The Issue #12 analysis mapped Dohan as the operational buyer and Jorge as the executive sponsor. Both hold.
Here is what the duration adds. A posting that survives multiple requisition cycles with largely unchanged language means the committee agrees on the description but hasn't found someone who proves the capability. The mandate hasn't changed, but the pipeline has turned over.
For a role this organizational, the proof gap is almost certainly institutional. Confidence: moderate — pattern inference from search duration, not direct evidence. The panel has probably seen candidates who can demo AI tools, talk about design systems, present interaction models. What they likely haven't seen is someone who can show how a single AI-enabled design practice moves from a local experiment to an organizational standard, with explicit evidence thresholds, decision rights, and failure boundaries governing that transition.
Build your outreach around that artifact.
Your angle
The Trust essay already contains the framework. Handoff 4 — the Decision Gate — defines the moment where a human reviews AI output against evidence thresholds and decides whether to accept, modify, or reject. The trust ladder (Watch → Verify → Delegate) describes how that gate loosens as confidence accumulates.
Apply the framework to Capital One's problem directly. Replace "AI output" with "AI-enabled design practice." Replace "user" with "XD leadership." The decision gate becomes: what evidence must a local experiment produce before the XD leadership team promotes it to an organizational standard? Who holds the decision right? What is the rollback when a standard doesn't transfer to a different line of business?
Lead with this. Your Agentic Labs apps demonstrate trust architecture in a working system. The Trust essay names the framework. The institutional application — experiment-to-standard — connects both to what Capital One actually needs: someone who treats this as an organizational change management problem where AI is the medium, not an AI problem that happens to involve the org.
The forward-looking visual design artifacts — interaction models and decision flows for human-agent systems — would make the methodology visible and installable if published before outreach. They're not live yet. The window matters more than waiting, but if you can ship one before sending, prioritize the Human-Agent System Design domain. It maps directly to the organizational coordination problem this role exists to solve.
Alibaba and Thermo Fisher are secondary evidence here: you've redesigned enterprise platforms and built operating models in regulated institutional environments. They prove you can work at scale in complex orgs. Lead with the framework.
The category question
The posting reads like it could be three jobs: a methodology leader who redefines design practice, a tool-adoption operator who rolls out AI workflows, or a design technologist who builds AI-powered design infrastructure. Each one calls for a different person, a different first-90-day plan, different evaluation criteria. The committee may not have resolved which one they're hiring for — that ambiguity alone could explain the search duration.
Your outreach to Dohan should surface this directly. You bring a framework for how to move AI practice from experiment to standard. Ask which category of work the team sees as primary. That exchange positions you as someone who has already thought past the job description.
Who to contact
Patrick Dohan, Capital One Design Leadership Team — LinkedIn. His exact title was not confirmed in this research cycle; verify from your authenticated LinkedIn before using a specific title in outreach. No new public posts about AI or design operations in the last 60 days. His hiring post remains the strongest attributable hook — specifically the phrase "experimentation bias at scale." That phrase tells you Dohan values someone who runs experiments and makes them stick.
Warm path: unverified from public sources. Treat as cold unless your LinkedIn shows 3+ mutual connections.
Daniela Jorge, SVP and Chief Design Officer — LinkedIn. Route through Dohan first. Jorge is the executive sponsor, not the operational buyer.
Outreach messages
Warm message (use if your LinkedIn reveals a mutual connection):
Patrick — [name] suggested I reach out about the AI-in-XD Director role. I published a framework for the experiment-to-standard transition in "Trust Is the New Interface" and shipped an enterprise agentic platform from 0→1 at TinyFish — the governance architecture for when AI practices earn organizational adoption is the problem I've been working on. Would you have 20 minutes to compare notes on how the team is drawing the line between methodology, adoption, and infrastructure?
Cold message to Dohan:
Subject: The experiment-to-standard gate in XD — who designs that?
Patrick — "experimentation bias at scale" in the AI-in-XD posting is the most precise framing I've seen of the actual problem. Capital One doesn't lack AI experiments. The gap is the decision architecture for when an experiment earns organizational adoption — evidence thresholds, decision rights, rollback boundaries.
I published a framework for exactly this transition in "Trust Is the New Interface" — evidence thresholds and trust calibration across Watch → Verify → Delegate stages — built from shipping an enterprise agentic platform from 0→1 as Head of Product at TinyFish, where the governance question surfaced daily: when does a human stop checking?
I'm curious whether the team sees this Director role primarily as a methodology question, an adoption question, or an infrastructure question. That distinction shapes the first 90 days entirely. Would you have 20 minutes to compare notes on how you're drawing that line? I'd bring a specific model for the experiment-to-standard gate.
Confirm "experimentation bias at scale" is Dohan's exact language in the linked post. If the wording is close but not exact, paraphrase rather than mirror. A misquote in the opening line is unrecoverable.
What not to say: Don't reference the search duration or imply the role has been hard to fill. Don't position yourself as an AI product designer — this role is organizational, and framing yourself as a product designer signals you've misread the mandate.
Cover letter draft
The hardest design problem in Capital One's Experience Design organization right now is the decision architecture for when an AI-enabled practice stops being one team's experiment and becomes an organizational standard — with evidence thresholds that justify the transition, decision rights that govern it, and failure boundaries that protect the teams it scales to.
Most recently as Head of Product at TinyFish, I shipped an enterprise agentic platform from 0 to 1 in three months, working daily through agent auditability, attribution, reversibility, and the governance challenges that surface when AI operates at organizational scale. That production experience is the foundation for "Trust Is the New Interface", a framework defining the five handoffs where human-AI trust is built or broken and the trust ladder (Watch → Verify → Delegate) that governs how decision rights transfer. The framework applies directly to Capital One's challenge: each AI-enabled design practice in XD needs its own trust ladder, its own evidence thresholds, and its own decision gate before it earns organizational adoption. I moved to product because I wanted to build AI-natively from zero. I'm returning to design because institutional design — the kind that determines whether AI actually changes how a design organization works — is the problem I want to spend the next several years on.
Before TinyFish, I led design at enterprise scale. As Head of Design and Research at Alibaba.com (North America), I redesigned the B2B platform serving $50B+ GMV, driving +20% daily transactions and +2.2pt NPS. As Product Design Director at BCG Digital Ventures, I built design functions from zero in regulated environments: Thermo Fisher's digital supply chain ($20M+ margin recovered, 6 pharma partners, 100% adoption) and the American Red Cross disaster relief platform ($847K disbursed, 6 legacy systems consolidated to one). Each of these was an organizational design problem where the product was the operating model.
My portfolio and framework are at junochen.com.
Resume emphasis guidance
- Trust essay and Agentic Labs lead — the framework and production proof that map directly to the organizational mandate. Link the Trust essay. Link the three lab apps by direct URLs: Brand Pulse, Carrier IQ, Retail Velocity.
- TinyFish second — most recent role context, bridge narrative. Head of Product, enterprise web agent platform, 0→1 in 3 months. Frame as: moved to product for AI-native build, returning to design for institutional impact.
- Alibaba third — Head of Design and Research, $50B+ GMV, enterprise platform redesign. Enterprise-scale credibility anchor.
- Thermo Fisher and Red Cross fourth — regulated 0→1 builds as Product Design Director at BCG Digital Ventures. Prove you can build operating models in complex institutional environments.
- Trim: Equinox+ and Allē are less relevant here. The panel is buying institutional design capability.
Post-outreach signals
Enterprise cadence applies — 5–7 day response is normal, follow up at day 10 and day 20. See prior card for the full enterprise timeline. A reply that engages the category question (methodology vs. adoption vs. infrastructure) is a strong signal the message cut through. Respond within 24 hours and deepen the thread.
The non-obvious thing
Daniela Jorge's July 27 post praised promoted designers for working "across nearly every line of business and horizontal organization." That language tells you what Jorge values in XD leaders: horizontal reach across the org. If you get to a Jorge conversation, frame the experiment-to-standard gate as a horizontal capability — something that works across lines of business, not a deep methodology for one team. Use the word "horizontal" when talking to her.
The Agentic Labs case-page routes on junochen.com still return 404. Use the direct application URLs (Brand Pulse, Carrier IQ, Retail Velocity) in all outreach. A hiring manager who clicks a broken link forms an impression you cannot repair. Fix the case-page routing before sending if possible — but the window matters more than the routing.
- Slack's Principal Architect search: Michael Lenahan publicly confirmed the role reports directly to him and launched alongside Slack Code's human-agent review surfaces — a live trigger with a named buyer and a product event happening now.
- Render's first Staff Agent seat: Head of Design Samia Saleem opened DMs and asked for portfolios for Render's Staff Product Designer, Agent Experience, but the role's CLI/API/MCP substrate demand creates a different evidence-fit question than Capital One's organizational mandate.
- Atlassian's new CPO arrival: Ken Exner started as Chief Product Officer for Enterprise & Emerging on August 6, and Atlassian currently lists a Principal Product Designer for Rovo & AI — though no public source yet connects Exner to that search.
- Forward-looking artifact status: The Human-Agent System Design and Inference-Aware UX domains both returned 404 on September 26 — prioritize the Human-Agent artifact if you're building before Capital One outreach, since it also supports Slack and Atlassian positioning.
