Allocation Map
| Company | Lead Proof | Why This One |
|---|---|---|
| Amplitude | Retail Velocity (Lab) | Wave's core flow is agent-surfaces-signal, human-decides-to-act. Retail Velocity is that pattern, live. |
| Ramp | Carrier IQ (Lab) | Ramp's designers ship PRs and build financial agents. Carrier IQ is a built financial decision surface, not a mock. |
| Brex | Trust essay (Decision-gate section) | Staff IC role. They're evaluating thinking-in-the-medium, not leadership scale. The essay proves philosophy for agentic finance. |
| Maven | Thermo Fisher (Case study) | VP Design in healthcare AI. Trust has clinical stakes. Thermo Fisher is that proof. |
Each pairing follows the posting's actual problem. In Issue #4's room-level routing, Alibaba was the default lead for CPO/VP Product rooms. Correct first pass. This is the company-specific layer: no two buyers see the same lead, and each assignment reflects what the posting reveals about what they're solving for right now.
Amplitude
(1) The design problem. Amplitude's Wave ships agent-surfaced product opportunities to human teams who then decide whether to commit resources. The problem is specific: what does a human need to see, in what sequence, at what confidence level, to trust an agent's recommendation enough to act on it? The CPO resonates with "intelligence layer" language. Use that phrase. Not "agentic products."
(2) Lead with Retail Velocity. An agent audits accounts daily, surfaces placement gaps with opportunity scores, produces a ranked action list for a human field rep. The architecture maps to Wave directly: continuous signal detection → ranked surfacing → human commitment point.
"I built an intelligence layer that turns continuous field data into a daily ranked action surface for human operators. That's the same design problem Wave solves for product teams."
(3) Metric. TinyFish velocity: current AI-production shipping speed. Confirm the exact product count and timeline from Juno's own records before citing. Frame as current operating speed, not past case study output. This answers the CPO's top anxiety, will design slow the pod, without positioning TinyFish as portfolio proof. TinyFish proves currency. It does not prove craft.
(4) Competitive gap. Other candidates will show AI interface work. You show a working intelligence-layer product you designed and built, with the same signal-to-action architecture Wave uses. The gap: you've already solved the specific problem of making agent-surfaced signals actionable for a human operator, and the live artifact exists.
(5) Do not lead with the Trust essay as an abstract framework. The CPO buyer evaluates cognitive load reduction, not design philosophy. Essay sections support the conversation if it goes deep. They don't open it. And don't say "agentic products." Say "intelligence layer." Confirmed.
(6) Trust proof note. The portfolio's progressive disclosure mirrors the intelligence-layer philosophy Amplitude is hiring for: surface the signal first, let the human pull deeper when ready.
Ramp
(1) The design problem. Ramp's Director posting says design is "building tools and agents, designing memory, shipping PRs, and rethinking process as new capabilities unlock." Read that verb list. Every word is a builder-velocity signal. The role reports into an existing structure under VP Design Diego Zaks. You start by owning a product area and team, with room to grow scope.
Can you lead a team of builders at the speed this company ships, while the domain constraint of real-money financial agents makes every trust failure expensive? Velocity is the primary anxiety. Trust is what makes velocity hard.
(2) Lead with Carrier IQ. A financial decision surface where an agent runs carrier workflows in parallel, returns quotes with confidence scores, presents a side-by-side evaluation surface. The broker evaluates. The agent assembles. Built artifact, exact domain Ramp operates in, and it demonstrates the trust architecture financial agents require without sacrificing the speed signal.
"I designed and built a financial decision surface where an agent assembles quotes and a human broker evaluates them with confidence scores. That's the trust layer your financial agents need, and I built it at the speed your team ships."
(3) Metric. Carrier IQ opens the door but creates an objection: lab demos aren't production scale. Neutralize with a Thermo Fisher metric proving you've shipped at enterprise scale. The lead-and-neutralize pattern from Issue #4 applies here. I cannot confirm the exact Thermo Fisher figure from public sources. Pull the specific number from the live case study before using. You want the metric showing volume of decisions designed for. Not team size. Not org scope. One number that says "I build fast AND at scale."
(4) Competitive gap. The posting uses "shipping PRs" and "designing memory." That's verb-level screening for whether you work in the medium or design interfaces for it. Your labs are built artifacts, not Figma mocks. Other Director-level candidates will present design process and team leadership. You present working financial-agent surfaces you made. At a company where designers ship PRs, that is the gap.
(5) Do not lead with org-building narrative. The posting says "start by owning a product area and team, with room to take on broader leadership over time." They're hiring a builder who leads, not a function-builder. Process-heavy, design-maturity framing will read as enterprise drag in this room. Do not lead with Alibaba. The scale signal is wrong for Ramp's builder culture.
(6) Trust proof note. The portfolio's gated architecture demonstrates the same progressive-trust pattern Ramp's financial agents need: show the reasoning before asking for the commitment.
Brex
Brex's Staff Product Designer, AI role is a senior IC seat inside an existing design org. If the priority is a Head of Design role, deprioritize Brex. If the priority is getting inside a company where design has C-level representation and the AI surface is the most consequential product work, the conversation is worth having. Position as a maker who chose this, not a leader who settled.
(1) The design problem. The posting says the hire designs "agentic experiences at the heart of Brex's intelligent finance platform." The design org is led by CDO Matt Bango, who reports to the CEO. Seven design postings right now, all Staff or Senior. No leadership openings.
What does the interaction layer look like when a financial agent acts on behalf of a business, and the user needs to trust it with real spend decisions? Craft problem. The evaluator will assess depth of thinking about agentic design patterns. Not org-building capability.
(2) Lead with the Trust essay. The decision-gate section specifically: the framework for designing the moment where a human decides whether to accept, modify, or reject what an agent produced. For a Staff IC role, the essay proves something more valuable than any single case study. You have a design philosophy for the exact problem they're solving, articulated at a level that raises the team's thinking.
"I published a design framework for the problem your AI team is solving — how humans decide to trust, override, or act on what an agent produces. Here's how I'd apply it to Brex's agentic finance surface."
(3) Metric. The essay opens the door. A number proves you ship, not just think. Pull the Carrier IQ metric showing confidence-scored outputs evaluated by a human operator. If the live case study surfaces a task-completion or decision-accuracy figure from the broker evaluation flow, use that. Maps directly to Brex's agentic finance surface and proves maker output alongside conceptual depth. Verify the exact figure against the live case study before using.
(4) Competitive gap. Other Staff-level candidates will show AI interface work. Few will have a published design framework for human-agent trust that maps to financial agentic products AND working lab demos proving they build in the medium. At Staff level, most candidates bring depth or output. You bring both.
(5) Do not lead with leadership narrative, team-building experience, or org-design language. The posting never mentions hiring, team building, or design operations. Leading with those signals you see this role as beneath you. Do not lead with Thermo Fisher's enterprise scale. Wrong register for a Staff IC conversation.
(6) Trust proof note. The essay's own structure, each section building on the last before revealing the next layer of complexity, demonstrates the interaction design philosophy Brex's agentic products need: earn trust at each layer before asking for more.
Maven
(1) The design problem. Maven's VP of Design reports to CPO & COO Jason Lee, who oversees design and development of Maven's virtual platform. Lee has described Maven's product mission as connecting patient experience with the business while building "safe spaces for deep dialogue." He has led a cross-functional team of designers, product managers, and ops specialists, and emphasized quick feedback loops with internal care teams. Maven launched Maven Intelligence in March 2026: an AI orchestration layer using OpenAI and Google LLMs with HIPAA compliance, multidisciplinary review, and ongoing human review of conversation samples.
The problem: build a design function that can ship AI-powered clinical experiences where a wrong output is a health risk, under a CPO who has strong product instincts and has been operating without senior design leadership. Trust here is a clinical requirement.
The CPO buyer pattern matters. Lee owns design. He's been running the function directly. His bio says he oversees design and development; the VP Design seat is the role being filled. His public language shows strong opinions about patient engagement and provider tools. Expect the peer-evaluator dynamic, not a blank-slate mandate. He's hiring someone to take the cognitive load of design decisions off his plate, but he will have views on how those decisions get made.
(2) Lead with Thermo Fisher. Proves design leadership in a regulated environment where human oversight of system outputs is non-negotiable. The parallel to Maven Intelligence, AI-surfaced clinical guidance requiring human review before reaching patients, is direct.
"I led design in a regulated environment where every AI-surfaced recommendation required human validation before reaching the end user. That's the trust architecture Maven Intelligence demands."
(3) Metric. The Thermo Fisher metric that speaks here is the one demonstrating scale of human-in-the-loop design across a regulated product surface. Lee's public statements emphasize quick feedback loops and provider tools. Pull the specific figure showing volume or scale of human-validated decision points designed across the regulated product. If the case study surfaces a number tied to workflow coverage, error-reduction, or review-loop scale, that's the one. I cannot confirm the exact figure from public sources; verify against the live case study before using. You want the number proving you designed those loops at scale.
(4) Competitive gap. Other VP Design candidates will have healthcare experience or AI-product experience. Few will have all three layers: regulated-environment leadership AND a published trust framework for human-agent interaction AND working AI lab demos showing current technical fluency. Maven Intelligence launched four months ago. They need a design leader who doesn't need to learn why trust matters in healthcare AI. You don't.
(5) Do not lead with speed or velocity framing. Lee's public language emphasizes "safe spaces," patient engagement, provider tools. Healthcare AI buyers are anxious about moving too fast, not too slow. The Maven Intelligence announcement emphasizes HIPAA compliance, human review, and multidisciplinary oversight before it mentions any capability. Match that register. Do not lead with the labs. VP role. The evaluator needs leadership proof first, maker proof second.
(6) Trust proof note. The portfolio's gated-case architecture, where sensitive work sits behind access controls and the public layer earns trust before revealing depth, mirrors the information-architecture problem Maven Intelligence faces: how much does the AI surface to the patient, when, and what stays behind the clinical review gate?
- Vanta's 40-person org: Vanta's Head of Design role leads roughly 40 designers across GRC, Trust, Platform, Self-serve, and AI, with agentic compliance workflows shipping now — a potential Act-tier addition if the current four don't convert.
- Ramp's Q2 agent velocity: Ramp's latest product release frames finance AI as "velocity where you want it, control where you need it," with agents now handling reconciliations, procurement, and compliance checks — useful language to mirror back in outreach.
- Maven Intelligence's review architecture: Fierce Healthcare reports that Maven Intelligence includes automated evaluations and ongoing human review of conversation samples, which maps directly to the decision-gate framework and is worth referencing if the conversation goes deep with Lee.
- Gusto's AI-native design reorg: Gusto published how they shifted from a traditional design team to an AI-native one within a quarter, including MCP-connected design-system workflows so AI tools use real components — relevant if Essay 03 on AI-native operating model needs a market-evidence anchor.