Phase 1 — The Moment
Webflow shipped Source on September 2: a governed agent workspace with role-based access, approval paths, and audit trails for marketing teams. Sixteen days later they posted a Staff Product Designer for Agentic CMS. They built the governance layer first, then opened a design seat to own the content those agents will actually touch.
Real mandate: design the surfaces where humans and agents co-author, govern, and publish structured content — content models, API affordances, localization, publishing workflows — and own the controls and trust signals that make agent-authored content safe at enterprise scale. The posting assigns upstream work on content architecture and data modeling alongside downstream craft on agent-facing trust surfaces. Code-based prototyping required. No people management.
Rubric:
| Dimension | Score | Basis |
|---|---|---|
| AI centrality (bonus) | +2 | Source is a primary strategic bet; core product remains the website builder |
| Stage & equity window | 3 | Pre-IPO, Series C at $4B valuation |
| Design influence ceiling | 3 | VP Design leads AI vision; posting assigns influence over product strategy and technical architecture |
| Trajectory | 3 | Source launch, CPO and SVP AI hired July 7, active design hiring |
| Company total | 11 | |
| Total comp potential | 2 | Staff IC with pre-IPO equity; cash tops out at $230K per the posting, before bonus and equity |
| Scope expandability | 3 | Agentic CMS is a new product surface with no structural ceiling |
| Craft depth | 3 | Hands-on IC, code-based prototyping, no management layer |
| AI exposure quality | 2 | Agent trust surfaces for the flagship agentic product, but Webflow is not AI-native; AI is a strategic bet, not the core product |
| Portfolio value | 3 | Agentic CMS at a design-tool company — differentiated, publishable |
| Role total | 13 |
Posting is 1 day old (posted September 18). First-week applicants set the anchor the rest of the pipeline gets measured against. The Source-to-hire sequence is recent enough that referencing it reads as attention, not research.
Phase 2 — Portfolio Mapping
Intelligence layer check applies. Source pushes agent-generated work into a human inbox for inspection and approval. The design problem is the moment between what an agent produces and what a human publishes. The Trust essay's five handoffs and the Watch→Verify→Delegate ladder are your entry vocabulary here.
Lead with: Trust Is the New Interface and Brand Pulse. The essay gives you the framework for governed handoffs: where the human inspects, where they approve, where they delegate, what happens when the agent is confidently wrong. Brand Pulse is the running version — an agent generating sentiment scores and written narrative, a human watching, evaluating, and deciding what to act on. Source's inbox-and-approval pattern is that same structure applied to CMS content at enterprise scale.
Support with: Alibaba.com. The posting wants coherence across interlocking surfaces: authoring, governance, localization, publishing, delivery to agents. Alibaba is the proof that you can name a structural gap in how a large platform holds together, build the mandate to fix it, and execute across interdependent surfaces at $50B+ GMV (+20% transactions, +2.2pt NPS).
TinyFish in conversation only. When production agent experience comes up: "Most recently as Head of Product at TinyFish, I shipped an enterprise web agent platform from zero to production in three months — working daily with agent traces, auditability, attribution, and governance in enterprise deployments." Past tense. Practitioner credibility. Never cited as portfolio proof.
The hardest objection: the posting asks for content models, API affordances, data architecture, and localization governance. Your published work demonstrates governed human-agent handoffs and platform-scale coherence. What it does not contain is a content-governance artifact from a CMS context — a content model, a localization workflow, a publishing-state design.
That is a portfolio gap, not a capability gap, and the distinction changes how you answer it. Don't manage it as a weakness. Walk through how the five-handoff framework decomposes when the artifact under review is a localized content variant rather than a sentiment score: what Verify has to show, what the approval gate has to expose, what a rollback looks like when the thing published is live on a customer's site. Then say directly that content architecture is the dimension you want to go deeper on. CPO Ben Haefele co-authored a piece with Kevin Wong arguing that teams should define what good looks like before automating the decision; your Trust essay makes the same argument from the design side. That overlap is concrete enough to hold the conversation while you address the gap. (Confirm the co-authorship at the URL before you cite it — the research supports Haefele and Wong, but check it yourself.)
Phase 3 — The Outreach Package
Evaluator hypothesis
The posting says the role reports to a Senior Director of Product Design. The best hypothesis for that seat is Erica Deahl, who identified herself publicly as Senior Director of Product Design at Webflow about five months ago and advertised product-design openings on her team. Her stated priorities — helping teams express intent faster with AI, reconciling AI-speed creation with production-grade development, bridging design and code — line up with this posting's mandate. Confidence: moderate. No public source connects her to this specific requisition.
The executive routes are firmer. Kevin Wong, VP of Design, presented Source at WebflowConf and lists it among his selected 2026 work. Ben Haefele, CPO, wrote the Source announcement. Start with Wong. He owns the design vision for Source and will have direct input on who fills this seat.
First contact message (to Kevin Wong)
Kevin — your WebflowConf presentation on Source described the problem I've spent the last two years working on from the other side: what does a human need to see, and when, to trust that an agent's work is ready to ship?
Most recently as Head of Product at TinyFish, I shipped an enterprise agent platform from zero to production, then published a framework — "Trust Is the New Interface" — that breaks the agentic trust problem into five handoff types, each needing different design surfaces. I've also built three live agentic applications (Brand Pulse, Carrier IQ, Retail Velocity) that put those patterns into production.
The Staff Product Designer role for Agentic CMS is the version of that problem I want to work on next: structured content with agent authorship, human governance, and enterprise publishing, plus a content-model layer I'd like to build deeper in.
Would you have 20 minutes in the next week?
(165 words)
Resume framing note
Open the summary with the Trust essay framework and Agentic Labs, because those establish the exact vocabulary the posting uses — controls, transparency, trust signals for agents. Alibaba goes second: platform-scale coherence across interlocking surfaces, with the transaction and NPS numbers attached. TinyFish appears as recent role context only: "Most recently as Head of Product at TinyFish (enterprise web agent platform), shipped the agentic platform from 0 to 1 within 3 months." Compress the BCG DV 0→1 builds into one line; they establish range but they do not lead for this buyer. Keep "leadership" out of the summary entirely — this posting screens for craft and strategic influence, not org-building. Use the posting's own words where they fit: content governance, structured content, agent trust surfaces.
As of September 12 (7 days ago; status may have changed), the Brand Pulse and Retail Velocity case-study pages returned 404 while the live applications stayed reachable. Verify every case route before you send anything that links to junochen.com. If the pages are still down, the running applications carry the evidence alone — and a hiring manager who hits a 404 starts reading the rest of your material as claims to check rather than work to assess.
Questions for the conversation
"How does the Agentic CMS role divide responsibility between content-model design and agent-authority design?" The posting assigns both. They are different problems evaluated on different criteria. If the role mostly owns content models and API affordances, Alibaba leads for the rest of the conversation. If it mostly owns trust and governance for agent-authored content, the Trust essay and Brand Pulse lead. Ask this in the first ten minutes so you know which version of your background to open up.
"What's the timeline for Source moving from limited preview to broader availability, and how does the CMS roadmap sit against that?" Public language on Source's expansion is only "over the coming months." Whether the CMS ships inside Source's GA window or on a separate track determines the constraint set and how much design latitude the role actually carries.
Cover letter hook
"Source shipped the governance infrastructure. The Agentic CMS role owns what sits underneath it — what a human needs to see to trust agent-authored content before it publishes — and that is the question my Trust essay and three live agentic applications were built to answer."
Phase 4 — Window Summary
| Action | Deadline | What degrades without it |
|---|---|---|
| Verify Brand Pulse and Retail Velocity case-study pages are live at junochen.com | September 22 | Any outreach linking to the portfolio risks a 404 on the two most relevant proof objects |
| Send first contact to Kevin Wong | September 25 | Posting crosses day 7; anchor candidates are already in the pipeline |
| If no response from Wong, escalate to Ben Haefele referencing the "Made to measure" essay (confirm co-authorship at the URL first) | October 2 |
- Wong's "Made to measure" essay: Haefele and Wong argue that teams should define what good looks like before automating decisions, which mirrors the Trust essay's core claim and gives you a shared-vocabulary opening if the conversation reaches design philosophy.
- Source's governed agent controls: Webflow's product page describes risk-based autonomy, an approval inbox, and an audit view — study the specific control vocabulary before any interview so you can reference their implementation choices, not just the category.
- Webflow's design org practices: A January company post describes code-backed prototyping in Cursor, working with real behavior and dynamic content, and a 30+ person distributed design team — useful for calibrating what "code-based prototyping required" actually means inside this org.
- Haefele on agent authority: The CPO's Source launch article explicitly addresses when teams should require human approval vs. grant agents greater autonomy, establishing the company's public position on the exact trust-ladder question the role will design for.