JR358869 · Posted Sep 2 · Buyer-amplified Sep 4 Company 9 / Role 14 → Act Tier: Enterprise platform — read enterprise platform playbook before outreach. The role's AI depth warrants pulling from the AI-native playbook as well.
Portfolio match: "Trust Is the New Interface" + Alibaba.com, co-leads. The Trust essay's five handoffs map cleanly onto Slack's shipped agent surfaces, and they extend into the multiparty territory this role exists to address. Alibaba supplies the enterprise-scale proof Salesforce expects at VP level: trust architecture across a $50B+ GMV platform. Human-Agent System Design artifacts and Brand Pulse support as forward-looking and live agentic evidence respectively.
Why now: Lenahan promoted this role personally today (Sep 4), two days after posting. He describes a "Principal Architect-level design IC" defining how humans and agents collaborate inside Slack. Salesforce's Agentforce is the company's primary strategic bet; Slack is the delivery surface. Ashby's 2023 data shows inbound application volume peaking at 2.5–3× in week one. That pattern may have shifted since, particularly for AI roles, but it is directionally useful. Posting is 2 days old. Full first-week advantage available through Sep 9.
Equity note: Slack is a division of Salesforce, a $250B+ public company. Compensation is Salesforce RSUs, not pre-IPO equity, so stage and equity window scores 1. The role's strength is scope, AI exposure, and craft depth. Company score of 9 (AI centrality 2 + design influence ceiling 3 + trajectory 3 + equity 1) still clears the Act threshold; role score of 14 carries the verdict.
Score note: The grid view scores this role at 15. This brief scores it 14 because total comp potential drops from 3 to 2 — the $310K–$496K base range is strong, but the rubric's top tier requires significant pre-IPO equity upside, which Salesforce RSUs are not. The difference does not change the verdict.
Phase 1 — The Moment
Salesforce is turning Slack into the surface where Agentforce agents work alongside people, and Slack's Chief Design Officer went public today looking for the designer who will define what that interaction layer looks like. From Lenahan's Sep 4 post:
"New ways for people and agents to direct and collaborate with one another" — where "agents are co-workers, not just tools."
Slack has already shipped the single-supervisor version of this. Agent DMs, session states (active, processing, suspended, closed), stop controls, plan cards, and approval patterns for consequential actions are all live in the developer changelog. Each assumes the person who invoked the agent is the person who reviews what it did.
Nothing shipped covers the case where those are different people. An agent acting in a shared channel — modifying a canvas, sending on someone's behalf, surfacing information with access implications — raises an approval question Slack's own design guidance leaves open. The guidance specifies on-behalf-of labeling and preview-before-send. It does not say who approves when the invoker and the affected party diverge. The collaboration article by Small, Piper, and Pickard names shared roles and mutual accountability as conditions for AI collaboration, then stops short of an implemented authority model.
The mandate under the title: multiparty authority over shared agent actions, in a workspace used by millions of teams.
Lenahan's background. A RivalSense alert dated August 23 has him joining Slack as CDO from Google, where he was Senior Director of UX for Commerce–Merchant Shopping. A 2022 DesignMap case study identifies him as Carta's VP of User Experience. Both are secondary sources; the arc is plausible but not confirmed by Lenahan or either employer. Confirmed: he posted this vacancy himself on Sep 4 and said the hire reports to him. A recently arrived CDO hiring for his own vision team is building a function to his specifications, which means he is selecting for alignment. The first message has to show that you see the problem the way he does.
Urgency: Act Now. Posting is 2 days old, 7 days remaining at full first-week value. Buyer amplification today means he is watching inbound now. The window for a differentiated first impression runs through Tuesday.
Phase 2 — Portfolio Mapping
Lead with the Trust essay and Alibaba together. They do different jobs and carry equal weight. The Trust essay speaks to the design problem in the posting: its five handoffs map onto Slack's shipped surfaces, and the multiparty gap is where the framework extends past what anyone has built. Alibaba speaks to the company context, because Salesforce weights enterprise platform scale at VP level and $50B+ GMV with measurable transaction and NPS movement is that evidence.
The handoffs against what Slack has live:
- Intent-setting → agent DM, suggested prompts, composer actions
- In-progress → processing state, thinking display, plan/task cards, stop control
- Output review → source attribution, completion recaps, created content links
- Decision gate → preview, approve, modify, reject
- Loop feedback → review history, on-behalf-of labeling, App Home audit surface
All five hold for one person supervising one agent. What Slack needs is each handoff redesigned for the case where the invoker, the affected parties, and the workspace administrator are three different people with three different levels of authority. Authority can return to the invoker while accountability sits with someone else. That is the hard part of the interaction design, and it is the territory the Human-Agent System Design artifacts cover.
Support layer. Brand Pulse is live proof of designing the oversight surface for an autonomous agent: what the person watching an agent needs to see, when, and why.
TinyFish framing. Most recently as Head of Product at TinyFish, Juno shipped an enterprise web agent platform from 0 to 1 in three months, working daily with agent traces, auditability, attribution, and governance in enterprise deployments. Use it as practitioner credibility — the production grounding that makes the forward-looking artifacts credible. Not as portfolio proof.
Lenahan cannot currently see a recent, high-resolution piece of interaction design that Juno invented herself for a multiparty agent problem. The Trust essay is a framework. Brand Pulse is single-supervisor. The authority-topology artifact — who holds authority, who holds accountability, and what happens to a shared agent action when those diverge — is the thing that closes it. Its publication date determines whether a portfolio review sees that evidence at all. See Phase 4.
Objection — the one Lenahan will raise. "Your most recent title is Head of Product. Where is the recent hands-on craft?" The posting asks for 12 years of senior IC design experience and a portfolio showing both vision and craft. Answer with Agentic Labs: Brand Pulse, Retail Velocity, and Carrier IQ are solo-built, live, and published at junochen.com. That is current IC design work with her hands on it. Then give the bridge narrative plainly — she went to product to build AI-natively from zero and came back to design to put that depth against a specific high-stakes domain — and move on.
Phase 3 — The Outreach Package
First Contact Message
Michael — your Sep 4 post describes the VP Product Design IC as defining "new ways for people and agents to direct and collaborate with one another." That names a harder problem than most agent-design roles do. In a shared workspace, the person who invoked the agent and the people affected by what it does are often not the same, and Slack's shipped agent primitives don't yet have a model for that split.
I've been working on this directly. Earlier this year I published "Trust Is the New Interface," a framework for human-agent handoffs across five decision points, and I've built three live agentic products — brand sentiment, field intelligence, quote automation — that implement those patterns. Most recently as Head of Product at TinyFish, I shipped an enterprise web agent platform from 0 to 1 in three months, working daily with agent traces, auditability, and governance. Before that, as Head of Design and Research at Alibaba.com, I led the platform redesign for a $50B+ GMV B2B marketplace.
Slack's version of the authority problem is the most consequential one I've seen. Would you have 20 minutes to talk about how I'd approach it?
Resume Framing Note
Open the summary on agentic AI: the Trust essay framework, Agentic Labs as three live solo-built products, TinyFish as most recent role context in past tense — "Most recently Head of Product at TinyFish, shipped enterprise web agent platform 0→1 in 3 months." Alibaba metrics go second: $50B+ GMV, +20% transactions, +2.2pt NPS. This is an enterprise platform buyer and platform-scale proof carries weight here. Attribute Thermo Fisher, Red Cross, Allē, and Equinox+ as Product Design Director at BCG Digital Ventures. Lenahan wrote "vision, craft, and hands-on execution" — use those three words in the summary. Avoid "design-led transformation" and anything that reads as a management pitch; the role is an IC seat. Keep the verbs at the level of what she drew, built, and shipped.
Cover Letter Hook
Slack's agent platform has shipped the primitives for one person supervising one agent: session states, approval gates, stop controls. What hasn't shipped is the model for when that agent acts in a shared channel where the invoker, the affected parties, and the workspace admin hold different authority over the same outcome.
Phase 4 — Window Summary
| Action | Deadline | What degrades without it |
|---|---|---|
| Submit application + send first contact message to Lenahan | Sep 6 | First-week volume advantage decays after Sep 9 (Ashby 2023); his attention on this role peaks within 72 hours of his own post |
| Publish authority-topology artifact (multiparty handoff map) to junochen.com | Sep 11 | The Phase 2 proof gap stays open through any portfolio review — framework understanding without a designed solution |
| Follow up referencing the published artifact, if no response | Sep 12 | First-round screens are likely scheduled from the initial pool after Sep 12 |
- Slack's shipped agent sessions: The August 20 developer changelog details the four session states (active, processing, suspended, closed), multi-agent stacking, and the stop control — the exact primitives the authority-topology artifact should build from.
- Salesforce's Agentforce earnings context: The Q2 FY27 results reported $1.5B+ in Agentforce ARR growing 240% year-over-year and 7 billion Agentic Work Units delivered, establishing the business pressure behind this hire.
- Multiplayer AI collaboration framing: Salesforce's design-research article on AI collaboration by Small, Piper, and Pickard defines sharing, coordinating, cooperating, and co-creating as distinct collaboration modes — useful vocabulary for the authority-topology artifact and for interview conversations about multiparty agent actions.
- Lenahan's location discrepancy: His Sep 4 recruiting post says "San Francisco or remote" while the employer vacancy page lists San Francisco only — worth clarifying in the first conversation since it may affect how the role is positioned internally.