Seven AI Design Roles, Three Different Jobs

A sweep of buyer language across active targets — and which evidence to lead with where

Seven of your active targets have AI design roles open. All seven use "AI" in the title or description. They are hiring for three different jobs.

I swept recent public statements from the decision-makers at these companies — podcasts, blog posts, LinkedIn — and sorted by what they're describing, not what they're calling it. The pattern tells you which evidence to lead with at each company. The freshness of each statement tells you who to contact first.

Midjourney isn't on your original target list, but Banu Guler's July appointment as CDO — unifying product, design, and frontend — plus a Design Lead role reporting to her clears the rubric. It belongs in this sweep.

Three different design problems sharing one label

Capability adoption — OpenAI, Midjourney

The design problem: making an expanding system legible to its users. Ian Silber at OpenAI, in his August 16 podcast appearance, talks about finding "underlying primitives" that compose into "one cohesive, simple thing" — a single interface that can "shape shift into anything" while serving novices and power users simultaneously. His stated candidate filter: high curiosity, prototyping ability, a point of view about where the medium is going. He did not mention AI experience or enterprise scale.

Banu Guler at Midjourney, in her July 24 announcement, describes computers as tools that "reveal our humanity to ourselves" before people can articulate meaning, feeling, or imagination. She's leading a unified product, design, and frontend team across all Midjourney projects.

The question they're hiring someone to answer: did the user discover what this thing can do, and did they use it?

Your lead: Forward-looking artifacts — intent-based interaction, inference-aware UX. The Trust essay's intent-setting handoff maps directly to Silber's blank-input problem. Agentic Labs supports but doesn't lead here. Case studies are background credibility. Artifact-status note: the intent-setting artifact is conceptual, not yet published at junochen.com. Given Silber's freshness (six days), move now with the Trust essay and Agentic Labs. If the conversation reaches portfolio depth, name the artifact as in-progress work. Don't wait for it to ship before reaching out.

Organizational capability — Capital One, Gusto

The design problem: changing how an existing design organization works with AI. Daniela Jorge at Capital One describes a several-hundred-person XD org that must "mind the seams" among design, product, engineering, data, risk, enterprise platforms, and AI/ML. Patrick Dohan, the operational buyer, frames the mandate as "architect how designers work in the future" with "deep technical expertise" and an "experimentation bias at scale."

Amy Thibodeau at Gusto writes that "AI doesn't get to lead — our judgement about the customer problem does" and describes building an "AI-first design team" while retaining "human taste, judgement, and customer focus." Worth knowing: the "designers to builders" framing you may have seen comes from Katie Kovalcin's account of a transformation Thibodeau sponsored. Useful context, but not Amy's vocabulary. Mirror Amy's words in outreach, not Katie's.

The question: can this function absorb AI without losing craft quality or customer judgment?

Your lead: Trust essay as methodology you'd install at scale, not a framework you wrote once. Alibaba and Thermo Fisher (as Product Design Director at BCG Digital Ventures) as proof you've operated inside large, regulated organizations and changed how design functions. Forward-looking artifacts demonstrate what the methodology produces when applied. The Capital One battlecard from Issue #5 already established this positioning; the August 21 requisition renewal confirms the mandate is still live.

Delegated action in consequential domains — Headway, Brex, Stripe

The design problem: the trust boundary around AI acting on behalf of humans where mistakes cost money, health, or safety. Headway's three Staff Product Designer postings span provider onboarding, provider CRM, and group practices — each with its own consequential AI behavior. The CRM role designs an AI assistant that communicates and schedules on a provider's behalf, where uncertainty or incorrect behavior reaches a real patient. Brex's Staff Product Designer, AI posting assigns agentic financial workflows — delegation, in-flight supervision, correction involving money. Stripe's Senior Staff Risk posting centers compliance, risk control, and trust at transaction scale.

The question: did the system act within its authority, and could the human intervene before irreversible harm?

Within this cluster, Headway and Brex both have named buyers with public language you can hook on (Poses, Loomans — more on their vocabulary in the gap section below). Stripe's design leadership returned no comparable public statements in this sweep. The role posting is clear about the design problem, but without buyer language, your Stripe outreach has to lead on the problem itself rather than mirroring a decision-maker's framing. That's a harder message to write and a colder entry point.

Your lead: Carrier IQ (regulated automation, all five handoffs) and the Trust essay's decision-gate framework. Thermo Fisher (Product Design Director, BCG Digital Ventures) for regulated-domain credibility. Gap: the correction-lineage artifact — showing a disputed output producing a versioned correction that improves a later run — is still conceptual, not published at junochen.com. You can demonstrate the framework and the production foundation. You cannot yet show the full correction loop as a visual deliverable. If one of these three companies is your top priority this week, that artifact moves to the front of the build queue.

Contact order by language freshness

Buyer language decays. What someone said in January tells you what they were thinking in January.

• Ian Silber, OpenAI — August 16. Six days old. The freshest buyer language across all active targets. Dropped two days before OpenAI posted the Growth design leadership role. He described what he evaluates in candidates and what the design problem is, in his own words. Use them.
• Banu Guler, Midjourney — July 24. Four weeks old, still hookable. Her language is philosophical rather than operational — she hasn't described the design problems in systems terms. That's a definition window: your outreach can help frame the operational challenge rather than mirror an existing frame. The Design Lead role reports to her.
• Amy Thibodeau, Gusto — June 4/11. Eleven weeks old. Still usable because the principles she published are durable, not event-driven. But you're referencing a position paper, not a recent conversation. Engage the principle, don't treat it as news.
• Patrick Dohan, Capital One — February 26. Six months old. The role just renewed, but Dohan's language hasn't refreshed. Lead with Jorge's organizational framing instead — it's standing language, not dated language.
• Maykel Loomans, Brex — January 14. Seven months old. And this is where the next pattern matters.

The gap between what buyers say and what roles describe

Loomans's public language is about builders, product quality, flat organizations, hands-on leadership. The Brex role he's hiring for describes agentic financial workflows — delegation, supervision, and correction involving money. Different vocabularies describing different concerns.

The same gap at Headway. Jake Poses talks publicly about team expansion, raising the bar, building a new mental healthcare system. The role posting talks about provider economics, automation boundaries, patient safety.

Mirror the posting language to a buyer who thinks in builder language, and you sound like you read the job description and nothing else. Mirror the buyer's language without connecting it to the role's actual design problem, and you sound like you didn't read it.

Where buyer language and role language already align — Silber at OpenAI, Thibodeau at Gusto — your outreach is simpler. Where they diverge, the bridge sentence is what separates your message from every candidate who copied phrases from the posting.

Where the sweep found nothing

I scanned the rest of your named target list for comparable buyer language. Anthropic, Amplitude, Ramp, Suno, and Abridge returned no first-person design-leadership statements within the scan window. Stripe's design leadership — as distinct from the role posting discussed above — is similarly quiet in public channels.

At these companies, there is no buyer hook to grab. Your outreach leads on the product's design problem directly, which is a colder entry and requires sharper portfolio-to-problem mapping to compensate. The targets that did return signal — particularly OpenAI and Midjourney — carry disproportionate outreach urgency right now.

Act on this first

OpenAI's Growth role and Silber's podcast dropped within 48 hours of each other. He described what he evaluates — curiosity, prototyping, a point of view — and what the design problem is: composable primitives, a shape-shifting interface, capability discovery for users who don't yet know what the system can do. No other active target gives you buyer language this fresh or this specific. Contact OpenAI first. Lead with the intent-setting problem he named and the Trust essay's first handoff. Everything else in your queue can wait until Wednesday.