Phase 1 — The Moment
OpenAI posted Product Designer, Payments on August 6. $245K–$310K base plus pre-IPO equity at an $852B valuation. The posting language is generic, but the timing around it is specific.
In June, OpenAI shipped enterprise spend controls — project budgets, overage toggles, approval workflows. In July, they published agentic-era investment guidance acknowledging that agents consume resources autonomously and organizations need governance before they'll trust agents to work. The current billing surface is a patchwork across Codex, API, Enterprise, and Business, with documented semantic inconsistencies where "No limit" means different things on different control surfaces.
Real mandate: Design the surfaces where an enterprise administrator sets what an agent may spend, sees what it did spend, and decides whether to let it keep going. Financial trust architecture for autonomous systems.
Posting is 3 days old (6 days remaining at full response value). Company score 11. Role score 15. Highest-ceiling Act-tier opportunity in the current scan.
Phase 2 — Portfolio Mapping
Intelligence layer leads. The admin in this role doesn't control the agent directly. They decide based on what the system surfaces about agent spending behavior: how much was consumed, by which workflow, against what budget, whether to authorize more. The entry point is the moment between what an AI surfaces and what a human decides. Your published essay "Trust Is the New Interface" is the intellectual framework — five handoffs between machine action and human judgment, with the trust ladder (Watch → Verify → Delegate) as the progression model. Lead with that frame verbally. Then prove it with Alibaba.
Lead with Alibaba. The structural analogy is specific. Alibaba's B2B buyers transacted at $50B+ GMV through flows they couldn't fully inspect. You named the structural trust gap, built the cross-functional sprint program, redesigned homepage, search, and PDP to make transaction decisions legible. Result: +20% daily transactions, +2.2pt NPS. OpenAI's enterprise admins face the same problem at a different altitude — agents consume resources through workflows the admin didn't initiate and can't fully predict. The trust gap is structurally identical; the object of trust shifted from a counterparty to an autonomous agent.
Support with Allē. 30M members, dual-surface redesign, 3.2× redemption lift, $42 CAC. Redemption is a spending decision — budget allocation (points vs. cash), authorization logic (eligibility, expiration, provider rules), and behavioral feedback that shapes the next transaction. Those are the structural components this role applies to agent spending. Allē proves you design financial-action surfaces that change behavior at scale.
Five handoffs applied to money. Two controls map directly: how much the system may spend without asking (autonomy scope), and how sharply the admin reacts when spending looks wrong (intervention sensitivity).
- Intent-Setting = budget configuration before the agent runs. Published proof: Carrier IQ's structured intake before quote automation.
- In-Progress = usage visibility during agent work. Published proof: Brand Pulse real-time source-by-source scoring — confidence forms through visibility, not a loading state.
- Output Review = post-run spend attribution and reconciliation. Published proof: Carrier IQ's provenance layer — portal element highlighted, raw value extracted, confidence signal based on prior run consistency. For Payments, this is where the admin sees what the agent consumed, which workflow drove it, and whether charges match expectations.
- Decision Gate = approval thresholds, overage authorization. Published proof: Thermo Fisher's regulatory release gate — five agents running, one irreplaceable human gate.
- Loop Feedback = usage patterns informing future budget allocation. Published proof: Carrier IQ's published control record — wrong output marked, element down-weighted, verification step added by run 5, revised release criterion applied.
TinyFish — strong currency here, clear limits. At an AI-native company, TinyFish is your highest-value credential. You are building enterprise agent governance in production right now — tracing agent runs, designing correction interfaces for misclassified actions, working through auditability and attribution daily. Practitioner depth on the exact problem this role solves. Use it as current role context and technical currency in the message and cover letter.
The gap: you have no published artifact showing agent budgets, spending limits, or cost-attribution surfaces. Carrier IQ's control record is the closest published evidence of designing human oversight for autonomous actions. If asked, name the gap directly. The verbal TinyFish answer covers it. The published portfolio does not.
Hardest question: the title. The posting says "Product Designer." You've held Head of Design and Head of Product. The panel will ask — silently or aloud — whether you'll try to manage when the role is IC.
Your answer here is your strongest asset. You left a Head title at Alibaba. You took Head of Product at TinyFish specifically to build — shipped 3 products in 3 months, hands on agent traces and production deployment daily. You want to make things in a specific high-stakes domain. The TinyFish move is proof of pattern: you choose scope and craft depth over title. OpenAI's comp range and scope language (design direction, roadmap influence, culture-building as the team grows) confirm this is a senior IC role with room to shape the function. Say that directly. Avoid the word "returning" — it concedes a hierarchy nobody applied.
Phase 3 — The Outreach Package
Buyer routing. No public source confirms the direct hiring manager for this role. Micah Sivitz publicly described his OpenAI mandate as leading Work, Enterprise, and API design — the closest named remit to Payments' enterprise billing and API consumption surfaces. Ian Silber is confirmed Head of Design but not confirmed as the operational buyer. Moderate confidence: route first contact to Sivitz. A well-positioned message from a qualified candidate gets forwarded.
First Contact Message — to Micah Sivitz
Micah — your post about Platform and Tools design needing people who combine craft, product thinking, systems thinking, and technical depth is what flagged the Payments role for me.
I led the B2B platform redesign at Alibaba.com as Head of Design — $50B+ GMV, +20% transactions — where the core problem was making complex transactional flows trustworthy at enterprise scale. Now at TinyFish (enterprise web agents, Series A) I'm Head of Product, building agent governance in production: traces, auditability, attribution, correction interfaces for autonomous actions.
The Payments role sits at the intersection — financial trust architecture for systems that act autonomously. I've designed both the transactional trust layer at marketplace scale and the agent oversight layer in production. My published portfolio includes the Alibaba case and three agentic AI projects with live trust-calibration patterns.
Would you have 20 minutes this week to discuss the role?
(158 words)
Resume Framing Note
Lead the summary with trust architecture for autonomous systems — that maps to the Payments mandate without translation. Surface Alibaba transaction metrics first ($50B+ GMV, +20% transactions, +2.2pt NPS) because this panel is evaluating whether you understand financial-system design at scale. Subordinate Equinox+ — strong work, wrong buyer. TinyFish appears as current role context only: "Head of Product, TinyFish (enterprise web agent platform, $47M Series A) — building agent governance, auditability, and trust calibration in production." Frame the trajectory as a maker who has led functions when the work required it. Avoid any language that positions you as a manager who decided to go back to craft.
Cover Letter Hook
OpenAI's June enterprise spend controls and July agentic-investment guidance describe the same design problem I've been solving across two roles: how do you make autonomous system behavior financially legible to the person who has to trust it? I built the transactional trust architecture for Alibaba's $50B+ B2B platform and am now building agent governance controls in production at TinyFish — the Payments role is where those two lines meet.
Phase 4 — Window Summary
| Action | Deadline | What degrades without it |
|---|---|---|
| Tailor resume with framing note applied | Monday August 10 | Message without resume loses credibility if Sivitz forwards internally. 1 day. |
| Send first contact to Sivitz | Tuesday August 11 | Posting enters second week; early-applicant advantage halves. 2 days. |
| Prepare verbal TinyFish agent-budget answers for screen | Friday August 14 | If screen is scheduled within the week, the portfolio gap on agent-financial controls becomes an unforced error without rehearsed specifics. 5 days. |
- OpenAI's Engineering Acceleration role: The companion posting names instrument → observe → evaluate → decide → iterate → rollback as the product loop, which maps directly to the Carrier IQ control record and could become a second Act-tier brief if Payments doesn't convert.
- Gusto's trust vocabulary overlap: Gusto's Head of Design, Unified Service Platform posting uses "ready to ship, needs human review, should not be sent" — nearly identical to the Decision Gate handoff mapped here, making it the strongest fallback if you want to reuse the same evidence spine with a different claim.
- Approval as capability boundary: Anthropic's trustworthy agents research and Vercel's agent implementation both treat permission as action-specific rather than global trust level — prepare for a screen question about when an agent with high confidence should still be blocked from spending.
- The CHI transparency tension: A CHI 2026 study found that participants preferred progressive or on-demand transparency over maximal process visibility, which directly challenges the Trust essay's "confidence through visible work" position — have a verbal answer ready if the panel asks how much of the billing process to expose.