PHASE 1 — THE MOMENT
Read the posting language carefully. "How users delegate work to AI, supervise it in flight, review and correct outputs, and build trust in automation that touches money." That is not the usual "design AI-powered experiences" filler that means whatever the hiring manager needs it to mean next Tuesday. That is a design brief. Bonus qualifications sharpen it further: "high-stakes human-in-the-loop AI experience in domains such as fintech/healthcare/security." They know what they need. They wrote it down. This almost never happens.
The trigger. Brex already shipped the agents. Multi-agent onboarding cut manual identity reviews by 70% and hit 40% auto-approval on card applications. AI assistants inside ChatGPT and Claude are live, with MCP connections to Claude Code, Codex, and Cursor updated as recently as May 2026. AI-driven expense routing is in production. The infrastructure works. What doesn't exist yet: the trust layer. The surfaces that let a finance team watch an agent move money and decide whether to keep letting it. That gap is the hire.
Window indicators. First published July 8, 2026 — 16 days old, actionable. Greenhouse metadata shows a minor update July 13, no meaningful language revision. This is the only AI-labeled design req on the board. But sixteen AI-labeled roles span Engineering, Product, and CX. Active AI hiring wave, not an isolated req.
The Capital One question. Address this directly because it changes the math. Brex LLC is a wholly owned subsidiary of Capital One, N.A. Acquisition closed April 7, 2026, approximately $5.15 billion. CEO Pedro Franceschi continues leading Brex with the existing leadership team, reporting to Frank LaPrade. Separate careers page, separate Greenhouse board, separate brand. SF Chronicle reported (July 21, 2026) that Brex continues operating as a standalone business.
None of that changes the core fact. Pre-IPO equity is dead. The posting says "equity or other compensation may be provided" — under Capital One ownership, that likely means COF stock or a subsidiary-linked instrument. Incremental, not transformative. Cash range of $240K–$300K is within your floor.
Rubric recalculation:
| Dimension | Score | Rationale |
|---|---|---|
| AI centrality | 2 | AI is a primary strategic bet across every product surface; Brex is fundamentally a finance platform, not an AI company |
| Stage / equity window | 1 | Post-acquisition subsidiary; no pre-IPO equity remaining |
| Design influence ceiling | 3 | CDO reports to CEO (pre-acquisition org structure; monitor for post-integration changes) |
| Trajectory | 2 | Backed by $475B-deposit bank; expanding SF footprint to full 200K-sq-ft building; growth trajectory as subsidiary is opaque |
| Company total | 8 | |
| Total comp potential | 2 | $240K–$300K cash (within floor); equity upside limited under subsidiary structure |
| Scope expandability | 2 | Staff IC with leadership influence; subsidiary ceiling is real but standalone operations create room |
| Craft depth | 3 | Hands-on IC with strategic seat, explicitly stated |
| AI exposure quality | 3 | Delegation, supervision, correction, trust in financial automation — named in the posting, not implied |
| Portfolio value | 3 | Publishable, differentiated: agentic trust surfaces in fintech |
| Role total | 13 |
Company score (8) clears Watch but not Act. Role score (13) clears Act. The company is not the reason to move. The posting's design problem is. It maps to your published work with near-verbatim language convergence, and no other candidate in this search likely has a live agentic system in regulated fintech sitting at a public URL. Move on craft merit. Expect 4–6 weeks minimum — committee-stage approvals under Capital One's enterprise cadence.
PHASE 2 — PORTFOLIO MAPPING
The posting describes a specific boundary: what an AI system decided versus what a human is willing to stake money on. The Trust essay's five handoffs are the through-line.
Lead with Carrier IQ (Agentic Lab 04). Regulated workflow automation in insurance/fintech. Quote automation, coverage selection, confidence surfaces. The design problem is identical: an agent performs financial work, the human needs to know when to trust it, when to intervene, how to correct it without restarting. This is your most direct proof and no other candidate has it live at a public URL.
Second: Alibaba.com B2B platform redesign. $50B+ GMV, cross-border procurement, single orders running six figures. The trust architecture — inspectable processes, reversible decisions, confidence at the moment before commitment — maps to Brex's problem at the structural level. Different mechanism, same stakes: a buyer deciding whether to trust what the system is showing them before authorizing a large financial action.
Supporting: Thermo Fisher mySupply. The posting's bonus qualifications name healthcare as a qualifying domain. $20M-margin pharma supply chain, exception-first design (197 orders, surface the 3 that matter), a wrong batch means a compliance event. Strengthens the multi-domain regulated-systems argument without diluting the Carrier IQ lead.
Framework: "Trust Is the New Interface." The posting's language about delegation, supervision, correction, and trust reads like a product brief someone derived from your essay. The Trust Ladder (Watch → Verify → Delegate) is the conceptual architecture this role needs to operationalize. Cite as published thinking, not portfolio.
TinyFish as verbal bridge. Head of Product at an enterprise web agent platform. Daily work with agent traces and reversibility in production deployments. Three products shipped in three months. Establishes that you are building production AI agents right now, commercially, at scale. Do not cite as design portfolio proof. Use in outreach and conversation only.
The objection. The hardest question this buyer will raise: Your current title is Head of Product. Are you still a design practitioner? The title gap is 18+ months, and the posting requires 10+ years of product design experience. Answer with the work, not the explanation. During the same period you held the product title, you designed and shipped three live agentic products at junochen.com. Carrier IQ is a functioning agentic system in a regulated industry, built solo, live in production. The product title gave you a production-engineering vocabulary no pure-design candidate in this search brings.
PHASE 3 — THE OUTREACH PACKAGE
Buyer identification. Hiring manager is not named in the posting or Greenhouse metadata. I cannot confirm who owns this req from public sources. The strongest buyer-adjacent contact is Matt Bango, Chief Design Officer, who wrote publicly about restructuring Brex so design reports directly to the CEO and co-hosted Brex Design Config After Hours in May 2025. His article is the outreach hook. It names the structural bet that this hire either validates or doesn't. A secondary contact is Maykel Loomans, who co-hosted the same event, but Loomans has no confirmed connection to this specific req.
FIRST CONTACT MESSAGE
Matt — your "Design as a catalyst" piece made a structural argument: design reports to the CEO, CDO sits on the leadership team, and the function earns its seat by driving measurable outcomes. The Staff Product Designer role on your AI team is where that argument gets tested. Every delegation surface, every confidence score, every correction flow either earns or erodes the trust that keeps customers automating their finances.
I've spent the past two years designing exactly that boundary. At Agentic Labs I built Carrier IQ — a live agentic system that automates insurance quoting in a regulated workflow, with the confidence surfaces and human-override patterns your posting describes. Before that I led the Alibaba.com B2B redesign where $50B+ in GMV depended on trust architecture at the moment before a large cross-border order.
Currently I'm Head of Product at TinyFish, an enterprise web agent platform, building and deploying AI agents in production daily — agent traces, auditability, reversibility. I'm returning to design to apply that technical depth to a specific high-stakes domain. Financial automation where AI touches money is that domain.
Would 20 minutes make sense to explore the fit?
RESUME FRAMING NOTE
Lead the summary with the intersection: agentic AI design in regulated financial systems, trust architecture at transactional scale. Surface Carrier IQ and the Trust essay before Alibaba. Position TinyFish as current role context — "Head of Product, enterprise web agent platform" — with one line on production AI credibility, nothing more. Drop Equinox+ and Allē entirely. Use "confidence surfaces," "human-in-the-loop," and "delegation patterns" in the summary. The posting uses all three concepts and the ATS will pattern-match. Do not use "design system" or "design ops" language. The posting never mentions either.
COVER LETTER HOOK
"Brex's AI team is designing the moment where automation asks for permission to move money — and the interface that earns or loses that permission. I've designed the confidence surfaces and correction flows that earn it in regulated fintech automation where a wrong delegation means a wrong transaction, and I wrote the framework for why users anchor to worst outcome, not average accuracy, in agentic systems."
PHASE 4 — WINDOW SUMMARY
| Action | Deadline | What degrades without it |
|---|---|---|
| Send first contact to Matt Bango + direct apply via Greenhouse | July 28 | Posting enters day 20; post-acquisition hiring cadence likely involves committee-stage approvals — early entry matters because the timeline stretches long, not because it's closing soon |
| Prepare Carrier IQ + Alibaba portfolio walkthrough tailored to financial delegation/trust framing | July 31 | If Bango or AI team lead responds, you need the presentation ready before they offer a slot, not after |
| Monitor Brex Greenhouse board for AI design-manager or design-director posting | August 15 | A leadership posting after this Staff IC req signals expanded AI design investment and a second entry point at higher scope |
August 15, 2026.
- Capital One's design cluster: Two Director-level design roles posted July 20–22 at Capital One proper — one covering AI in Experience Design and one covering Card Servicing platform governance — which together reveal how the parent company thinks about design systems of control, useful context if Brex's hiring process routes through Capital One leadership.
- Brex's onboarding agent architecture: The January 2026 post on rebuilding onboarding as a multi-agent system details confidence scores, human deferral under conflict, and replayability patterns that are likely the internal precedent this Staff Designer role will extend across other product surfaces.
- Amplitude's parallel mandate: Amplitude's Head of Product Design posting asks the leader to define how agents and humans work together in shared workflows, making it the closest parallel mandate to Brex's AI trust problem but at the leadership level rather than Staff IC.
- Brex's AI-app integrations scope: The support documentation for using Brex in AI apps reveals which actions users can already delegate through ChatGPT and Claude connectors versus which still require the Brex dashboard, mapping the current trust boundary this role will likely push outward.