Every published case in your portfolio solves one problem at a different altitude: what happens at the seam between what an agent produces and what a human needs to trust, verify, and act on. Alibaba's trust architecture moved buyer security concerns -47% across a $50B+ GMV marketplace because you designed the verification layer, not the interface. The Red Cross platform replaced six legacy systems with a single national surface that processed 1,689 cases in its first two weeks under 10× surge conditions. The interface was fine. What mattered was that the verification layer worked for untrained volunteers under federal oversight, not just experienced caseworkers in normal operations. Your Agentic Labs systems are live products where the handoff between machine output and human judgment is the entire surface.
That coherence problem is the positioning spine. What follows are four frameworks showing how to modulate it per buyer. Same cases, same metrics. Different frame, different buyer, different interview.
Suno — Frontier AI
Lead with the Trust essay. The five-handoff framework IS their product problem.
Live Head of Product Design posting. Their product generates a complete song from a text prompt in under a minute, then hands the creator a multitrack workspace to regenerate, refine, extract stems, and remix. The hard design problem is not generation. It is the communication layer between what the agent interpreted from the prompt and what the creator intended. Your Trust essay maps that layer across five handoff points. Open with it.
Your product gives a creator a complete song in under a minute. The design challenge that follows is harder than the one that preceded it: the creator must read the agent's interpretation of their prompt, decide whether it matches their intent, and choose whether to refine, regenerate, or carry the output forward. That communication layer between agent interpretation and human evaluation is the design problem I've spent the last two years defining.
In "Trust Is the New Interface" I mapped five handoff points in agentic workflows — Intent-Setting, In-Progress, Output Review, Decision Gate, and Loop Feedback — that correspond directly to Suno's creator workflow from prompt to Custom Model. When I applied this framework to trust architecture at Alibaba.com, it moved buyer-reported security concerns -47% across $50B+ GMV, because designing the handoff deliberately produces measurable trust outcomes, not just analytical clarity.
I'm currently Head of Product at TinyFish, an enterprise web agent platform (Series A), where I work daily at the boundary between agent output and human oversight in production systems.
I'd welcome 30 minutes to discuss how the five-handoff framework applies to Suno's creator experience and where the highest-leverage coherence gaps live today.
Headway — Regulated Platform
Lead with Thermo Fisher for mission-critical verification. Red Cross for surge-scale proof. Allē is available if the conversation turns to multi-sided platform dynamics, but don't open with it.
Headway's Scribe generates insurance-compliant progress notes from session audio for provider review and signature. A hallucinated clinical detail in that note is not a UX error. It is a care failure that affects treatment continuity, claims accuracy, and audit exposure. Your Thermo Fisher and Red Cross work prove you've designed verification layers under exactly those stakes.
Scribe drafts progress notes from session audio. When it hallucinates a clinical detail and the provider signs without catching it, that's not a bug report. That's a care failure with downstream consequences for treatment records, insurance claims, and audit exposure. Designing the review layer so providers reliably catch what needs catching without turning every note into a line-by-line audit is a coherence problem I've solved in adjacent high-stakes domains.
At Thermo Fisher, I built a pharma supply-chain platform across six partners and nine sites where late exceptions at the delivery gate cost 3–5x more to fix. The system recovered $20M+ in annual margin and reduced overhead 42% by surfacing the right verification moment at the right decision point. At the American Red Cross, I replaced six legacy systems with a single national platform that disbursed $847K and processed 1,689 cases in its first two weeks at 10x surge volume with zero retraining.
I'm currently Head of Product at TinyFish, an enterprise web agent platform (Series A), where I work daily at the boundary between agent automation and human verification in production systems.
I'd welcome a conversation about how Headway is thinking about the verification layer in Scribe's clinical workflow and where design can reduce provider cognitive load without reducing clinical accuracy.
Adobe GenStudio — Enterprise
Lead with Alibaba. The -47% buyer security concerns metric proves you move trust outcomes at scale, not usability metrics.
GenStudio's Brand Intelligence captures team judgment in a structured brand ontology, then validates AI-generated content variants across email, social, display, and CTV at volumes where manual review of every asset is impossible. The coherence challenge here: the agent produces more content than humans can verify, so the layer between what the agent generated and what the brand actually intended becomes the product. Alibaba is the analog. Use it.
Brand Intelligence captures team judgment in a structured ontology and validates AI-generated variants against brand guidelines, channel requirements, and accessibility standards across every surface from Meta ads to Connected TV. At enterprise scale, the agent produces more content than humans can manually verify. The coherence layer between what the agent generated and what the brand actually intended becomes the product itself.
I designed that trust infrastructure for Alibaba.com at $50B+ GMV scale, serving 200K+ suppliers and 25M+ desktop sessions. Through 32 stakeholder interviews, gaze-tracking studies, and a Baymard audit across three cross-functional sprints, I rebuilt the platform's trust architecture: +20% daily transactions, +2.2pt NPS, and -47% buyer-reported security concerns. That last metric matters most here. Moving trust perception at marketplace scale required designing systems where users could verify what they needed to verify without being forced to verify everything.
I'm currently Head of Product at TinyFish, an enterprise web agent platform (Series A), where I lead product strategy for agent-mediated enterprise content workflows.
I'd welcome 30 minutes to discuss how GenStudio's brand-governance layer handles the verification-at-scale problem as content volume grows.
Capital One — Enterprise Design Org Transformation
Invert everything. Lead with the Trust essay and live Agentic Labs systems, NOT case studies.
Capital One's buyer is not evaluating a product surface. They are trying to change how a design organization integrates agentic systems at enterprise scale. Lead with the fact that you've already built the agentic workflows they're trying to install and written the framework for how humans and agents hand off to each other. Case studies appear only as secondary enterprise credibility. The closing ask is a conversation about operating model, not a portfolio walkthrough.
The role you're filling is about changing how a design organization works with agentic systems at enterprise scale. I've already built the systems and written the framework.
"Trust Is the New Interface" is my published architecture for how humans and AI agents hand off across five critical moments: Intent-Setting, In-Progress, Output Review, Decision Gate, and Loop Feedback. It is the operating framework behind three live agentic systems in my Agentic Labs practice — Brand Pulse for real-time brand signal interpretation, Carrier IQ for logistics decision intelligence, and Retail Velocity for retail performance pattern detection. Each required designing the exact boundary you're trying to define at organizational scale: where does the agent's output end and the human's judgment begin, and how do you make that boundary legible across teams that didn't build the system?
The enterprise proof behind this work: trust architecture at Alibaba.com across $50B+ GMV and 200K+ suppliers, and a mission-critical supply-chain platform at Thermo Fisher recovering $20M+ in annual margin across six pharma partners. What I'd bring to Capital One goes past the case studies. It's an operational model for how design teams integrate agentic workflows without losing craft quality or human judgment.
I'm currently Head of Product at TinyFish, an enterprise web agent platform (Series A), where I lead product strategy at the intersection of agent automation and enterprise-scale design practice.
I'd welcome a conversation about how Capital One's design organization is approaching the agent-integration challenge and what operating model you're building toward.
Deployment Notes
Each letter runs under 250 words. That constraint is doing work for you. At this level, the cover letter is not the case. It is the reason someone opens the portfolio link. Every sentence that does not earn a click or a reply gets cut.
Customize the first sentence of each hook. A product launch, a leadership change, a feature announcement. That sentence proves you did the research. The frameworks above prove you can solve the problem.
The through-line holds because the thesis is real. You designed trust architecture before the industry had language for it. You built verification layers in regulated environments when "human-in-the-loop" was still an academic phrase. The positioning work here is selection, not invention. You are choosing which version of a true story matches what this specific buyer needs to hear this week.
Priority order if you're sending this week:
- Suno first. Live posting, clear mandate, direct match to your published framework. This is the closest thing to a clean shot you have right now.
- Headway second. The regulated-platform coherence problem is acute and your mission-critical proof is the strongest evidence in your portfolio. Don't let this one sit.
- Adobe third. Enterprise cycles are longer. The timing is less urgent. You can afford a few days.
- Capital One last. Org-transformation conversations take longer to convert and the entry point is a working session, not a portfolio review. Start the relationship, but calibrate your expectations on timeline.
Act on the first two before the week is out. The others can hold.
- Amplitude's Head of Design: Their live posting explicitly asks for someone who can define how agents and humans work together in the same workflows — language that maps cleanly to the coherence thesis and may warrant a fifth framework.
- Vanta's VP-level design role: Their Head of Design posting spans GRC, Trust, Platform, and AI across a ~40-person org, combining the regulated-platform and org-transformation archetypes in a single mandate worth watching.
- Gusto's agentic service design language: Their Senior Staff Service Designer posting names AI wrongness, escalation ownership, and coherent human-AI-service handoffs as explicit design problems — strong vocabulary to harvest even if the role level doesn't match.
- Adobe's horizontal agentic team: A Senior Design Program Manager posting confirms Adobe has launched a cross-product agentic experience team that now needs operational infrastructure to scale, which changes the context for how GenStudio's design leadership role connects to the broader org.