Thirty-Five Scoreable, Four Actionable, Zero Hiring for Provenance

Read this first. The format calls for 100 entries across ten clusters. I scored 35.

That is what was left after running the full standing set of ATS and VC boards, then cutting anything without an identifiable senior product-design mandate and a design mechanism I could name out of the posting's own language. Twenty-four of the 35 carry a posting that is live, date-certified, and open to the Bay or US-remote. Those get verdicts you can act on this week. Eleven do not. Real companies with real mechanism evidence, and nothing you can move on right now. They still get full profiles and a named trigger, because a monitoring grid earns its keep by stopping you from rediscovering the same eleven companies on your own scans next month.

The sixty-five-entry gap is the finding. I could have closed it with sixty-day-old reqs, New York-only listings, and public companies whose equity dimension scores a 1 regardless of how sharp the brief reads. Those are the Ignore verdicts below, and each one shows its work.

Nine clear both Act thresholds on the arithmetic. Four survive the gates: Giga, Horizon3.ai, Mintlify, Render. Maven, Fieldguide and Ambience clear the numbers and get capped at Watch by posting age. Altana and Cognita clear the numbers and are dead on date. Nine on merit, four you can act on.

Company	Company score	Role score	Posting age	Where it lands
Giga	12/12	15/15	Day 17	Act
Horizon3.ai	10/12	13/15	Day 9	Act
Mintlify	10/12	12/15	Day 15	Act
Render	10/12	12/15	Day 19	Act
Fieldguide	11/12	14/15	Day 22	Watch — decay gate
Ambience	11/12	14/15	Day 38	Watch — decay gate
Maven Clinic	10/12	15/15	Day 40	Watch — decay gate
Altana	10/12	14/15	Day 61	Ignore — dead on date
Cognita	10/12	14/15	Day 58	Ignore — dead on date

Ordering is the Priority Action Queue's job and I am leaving it there. One deliberate breach of that boundary: Giga is not close to second.

Two corrections that outlast this cycle

Take these before you read a single entry.

Kill attribution and provenance as a scan category. Zero qualifying roles this cycle. No live, dated, Bay-eligible posting made "where did this output come from and can I trace it back" the governing mandate. Provenance gets more design conversation than any other problem in agentic AI and produces the fewest reqs, because companies are folding it into professional sign-off roles where traceability belongs to the reviewer and never gets a surface of its own. High confidence. The pattern held across every regulated-vertical posting I pulled.

Method change

Stop scanning for provenance as a category. Scan for sign-off, then read the provenance language buried inside the requirements.

Pick your channel off the posting's age, not off how much you like the job. The hot postings and the well-matched postings are two different sets this cycle. Everything freshest on this board, day 2 through day 19, sits in enterprise governance workflow, where the AI mandate is thin and the actual design problem is configuration. The roles that score highest on portfolio value and AI exposure have already aged into the 22-to-42-day band. Past day 21 the field is assembled and a recruiter is triaging. The form goes into a queue at that point and the design leader never sees it.

How to read the entries

Everything is filed by mechanism: the design problem the posting actually describes. Who the human is. What surface they are staring at. What breaks when the design is wrong. Industry tells you close to nothing about whether your work maps. Mechanism does.

Mechanism is a different axis from tier archetype (AI-native, growth-stage platform, enterprise platform, healthcare/regulated), which governs how the company evaluates you once you are in the room: what they ask, what they discount, what they assume is table stakes. The two axes do not line up. Collapsing them is how strong candidates walk in leading with the wrong pillar. Mechanism picks your case study. Tier picks your register.

Two gates govern every verdict below:

Posting age. Hot for 21 days. Watch through day 42 with written justification. Dead after.
Link integrity. A posting exists if the direct application URL loads. No aggregator screenshots. No "it was live last month."

Cluster 1: Professional sign-off on machine output

Credentialed user, machine drafts, human's name on the result. Accountability sits with someone who holds a license. This is the densest accountability mechanism on the board and the tightest match to your Thermo Fisher and Carrier IQ work, and every live entry in it has aged past hot. These searches opened three to six weeks ago. Go around the form.

1. Maven Clinic. Virtual women's and family health delivered through employer benefits, with clinicians working inside an AI-supported care platform. Senior Staff Product Designer, Care Delivery.

Stage / Size / Capital: $125M Series F, Oct 2024, reportedly at $1.7B.
Offices: US remote within approved hub areas. Bay-eligible.
Value chain: Downstream, at the point of care. Value lands in clinical outcome and employer retention, and both run through the care-delivery surface. Design scope is high.
Best at: Making specialized clinical care feel like a consumer product without stripping the clinical rigor out of it.
Signals: Posted June 22. Day 40, late Watch band, confirmed live against primary ATS in this pass. Watch justification: specialized regulated domain, thin senior pool, and clinician-facing AI roles run long fill cycles. A separate VP of Design requisition ran earlier and has expired. This Senior Staff mandate is distinct and does not inherit that clock. Note that both existed. The org wants a senior IC and it wants function leadership.
Tier: Healthcare/Regulated. Tier Playbook: Read Healthcare/Regulated playbook before outreach.
Rubric scores: Company 10/12 (AI centrality 2, Stage/equity 3, Design influence 3, Trajectory 2). Role 15/15 (Total comp 3 at $220-265K plus equity, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: They wrote uncertainty communication, confidence indicators, deferral to human care, and human-to-AI handoffs into the requirements. What that describes is a clinician who either over-trusts a suggestion or quietly stops opening the tool. Lead with Thermo Fisher mySupply and 100% adoption across six pharma partners. Adoption is the number that moves this evaluator, because a tool a professional is permitted to refuse will be refused unless someone designed the refusal path first. Your differentiator: you have designed for a user whose license is exposed, not just their calendar.
Verdict + action: Watch, and know exactly why. The rubric clears Act at company 10/12 and role 15/15. The decay gate caps it at Watch, day 40. Find the design leader on Care Delivery and open on the deferral problem. Clock expires August 3.

2. Fieldguide. AI platform for audit and advisory firms: planning, executing, reviewing and closing regulated engagements with agentic assistance. Staff Product Designer.

Stage / Size / Capital: $75M Series C led by Goldman Sachs Alternatives, Feb 2026, $700M valuation. $125M total. 51-200 employees.
Offices: US remote. Bay-eligible.
Value chain: Downstream vertical application riding general models. Value concentrates in engagement throughput per auditor. Design owns the throughput.
Best at: Compressing the evidence-to-conclusion loop in audit without breaking the paper trail regulators expect to find at the end of it.
Signals: Posted July 10. Day 22, barely over, confirmed live against primary ATS this pass. Watch justification: audit-domain AI design is narrow, and the pool that can speak fluently about regulated engagement workflows is small. Company language describes a growing embedded design team. No design executive named publicly.
Tier: Healthcare/Regulated by evaluation pattern, since the workflow is a regulated professional one with high-consequence failure. Tier Playbook: Read Healthcare/Regulated playbook before outreach.
Rubric scores: Company 11/12 (AI 3, Stage/equity 3, Design influence 2, Trajectory 3). Role 14/15 (Total comp 2 at $180-210K plus equity, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Their user is an auditor deciding whether the evidence carries a conclusion they will personally sign. Lead with Carrier IQ, and specifically the reviewer surface where machine-extracted quote data gets accepted or corrected. The decision worth narrating is what you routed to human correction versus what you let auto-apply. Back it with Thermo Fisher's $20M+ margin under pharma compliance constraints. Your differentiator: you were building governance-visible review workflows before anyone attached the word AI to them, so you can talk about the audit trail as a design object instead of a compliance tax.
Verdict + action: Watch on paper. Treat it as Act in practice. Last cycle this publication called Fieldguide a clear Act. It has crossed day 21 and I am honoring the rule rather than quietly bending it, because the rule governs the channel, not the opportunity. Apply and reach the hiring manager the same day.

3. Ambience Healthcare. Ambient clinical documentation and workflow AI that clinicians review before it enters the medical record. Staff Product Designer.

Stage / Size / Capital: Well-funded private AI-native healthcare company. Current round terms did not resolve in this pass and I am not attaching a number I cannot source. Stage/equity scored 2 on inference from company profile and the Bay-Area AI-native band. Unverified. Get the round, the last 409A date and the option-pool position in the first call.
Offices: San Francisco.
Value chain: Downstream, embedded in clinician workflow. Value shows up as documentation time saved and coding accuracy, both of which design can see.
Best at: Getting clinicians to stop editing the output, which is the only honest measure of ambient documentation quality.
Signals: Posted June 24. Day 38, confirmed live against primary ATS this pass. Watch justification: clinical AI is specialized and the senior product-design pool is genuinely thin. The mandate spans clinical AI experiences, product strategy, design-system patterns, and postmortem learning around failures. That last item is rare and it is revealing. They have had failures, and they have institutionalized studying them.
Tier: Healthcare/Regulated. Tier Playbook: Read Healthcare/Regulated playbook before outreach.
Rubric scores: Company 11/12 (AI 3, Stage/equity 2 on inference, Design influence 3, Trajectory 3). Role 14/15 (Total comp 2, no range posted, inferred from market; Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Their failure mode is a documentation error that becomes permanent record. The postmortem language is your way in. Lead with the failure-attribution section of "Trust Is the New Interface," the part asking who owns a mistake once a handoff has already happened. Ground it in American Red Cross: six systems into one, national deployment, $847K disbursed, where a data error carried downstream consequences you could not recall. Your differentiator: you published a framework for the exact postmortem question they wrote into the job description, and they can read it before they meet you.
Verdict + action: Watch. Expires August 5. Name the essay in the first message and let it do the credentialing. Resolve the capital question yourself in screening. A 14/15 role score sitting on an unverified equity picture is not a decision public sources can make for you.

4. Socure. Identity verification and fraud risk decisioning. RiskOS is where a fraud operator approves, rejects or escalates an identity or a transaction. Staff Product Designer, RiskOS Platform.

Stage / Size / Capital: Late-stage private. No fresh round surfaced this pass. Stage/equity scored 2 on inference from late-stage position. Unverified.
Offices: San Francisco hybrid.
Value chain: Midstream decisioning layer between data providers and the businesses making the approve/decline call. Design owns the operator's evidence view, which is where the value actually gets realized.
Best at: Assembling scattered identity signals into one defensible verdict.
Signals: Posted June 26. Day 36, confirmed live against primary ATS this pass. Watch justification: platform-level fraud-decisioning design is a narrow field, and the posting exposed neither team size nor a design leader, which typically stretches the search.
Tier: Enterprise platform. Tier Playbook: Read Enterprise platform playbook before outreach.
Rubric scores: Company 9/12 (AI 3, Stage/equity 2, Design influence 2, Trajectory 2). Role 13/15 (Total comp 2 at $210-245K plus bonus and equity, Scope 2, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: The failure is asymmetric and they know it. False acceptance costs money, false rejection costs a customer, and the operator has seconds to choose. That is an evidence-hierarchy problem, and your nearest analog is the Alibaba.com search and PDP work: ranking evidence so a buyer either commits or leaves. Lead with +20% daily transactions and +2.2pt NPS on a $50B+ GMV platform. Your differentiator: you have designed evidence hierarchy where being wrong in either direction showed up in revenue, which is Socure's economics exactly.
Verdict + action: Watch. The weakness is the company, not the role. Influence ceiling 2 and trajectory 2 sitting under a strong mechanism. Expires August 7.

5. Cognita. Clinical imaging AI where a specialist interprets machine-surfaced evidence before rendering a read. Staff Product Designer, Imaging. No stable direct requisition URL resolved this pass. The role appeared only through the company careers index, which fails the direct-application-page rule.

Stage / Size / Capital: Private, early growth. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established in the retrieved record.
Value chain: Downstream clinical application. Value concentrates at the read, and design owns the read end to end.
Best at: Presenting imaging evidence in a form a radiologist will adjudicate rather than override by reflex.
Signals: First published June 4. Day 58. Dead under the decay gate, and the link failed the URL rule on its own.
Tier: Healthcare/Regulated.
Rubric scores: Company 10/12 (AI 3, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 14/15 (Total comp 2 inferred, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Specialist interpretation of machine-generated evidence is the closest thing on this entire list to Carrier IQ's extraction-review surface, moved into a clinical read. If it reposts, that is the opener.
Verdict + action: Ignore-until-trigger. Two independent gate failures. Trigger: a repost with a resolvable date and a direct application URL. Check the careers index monthly.

6. Vitalize. Hospital recommendation and automation software with audit trails and field-deployed workflows. Staff Product Designer.

Stage / Size / Capital: Private, early growth. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established in the retrieved record. Verify before investing.
Value chain: Downstream health-system application with an in-person implementation motion. Design influence is high because adoption is won on site.
Best at: Building design infrastructure for hospital operations that requires actual fieldwork instead of remote discovery.
Signals: First published May 20. Day 73. Dead. The mandate named design infrastructure, audit trails and in-person hospital fieldwork, which is an unusually specific brief, and a good one.
Tier: Healthcare/Regulated.
Rubric scores: Company 9/12 (AI 2, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 13/15 (Total comp 2 inferred, Scope 3, Craft depth 3, AI exposure 2, Portfolio value 3).
Portfolio match: Audit trails plus hospital fieldwork is Thermo Fisher mySupply with the pharma label removed. Same on-site adoption fight, same governance-visible record. The 100% adoption number is the proof point if this reopens.
Verdict + action: Ignore-until-trigger. Trigger: a repost, or any Vitalize design requisition with a resolvable date inside 21 days.

Cluster 2: Agent supervision and reversal

Someone watches an autonomous system work and decides when to intervene. Your published essay was written for this cluster, and the cluster is thinner than the discourse implies: three roles, all past hot, one with a posting date I could not resolve at all. Klarity has the sharpest brief anywhere on this board.

7. Klarity. AI that reads contracts and documents, then hands extracted terms to a human who accepts or corrects them. Staff Product Designer.

Stage / Size / Capital: Private, growth stage. Round terms unresolved. Stage/equity scored 2 on inference from the posted band and hiring level. Unverified.
Offices: San Francisco, on-site.
Value chain: Downstream application on document models. Value sits entirely at the accept-or-correct moment, and design owns that moment outright.
Best at: Making the confidence level of a machine extraction legible enough that a reviewer triages instead of re-reading everything.
Signals: Posted July 2. Day 30, confirmed live against primary ATS this pass. Watch justification: the brief is narrow enough that the qualified field is small. It names human-in-the-loop workflows, transparency, trust calibration, evaluation criteria, design systems and code-based prototypes. That is the most complete written statement of this mechanism on the board.
Tier: AI-native. Tier Playbook: Read AI-native playbook before outreach.
Rubric scores: Company 9/12 (AI 3, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 14/15 (Total comp 3 at $250-313K plus equity, Scope 2, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: They put "trust calibration" and "evaluation criteria" into the requirements, which means somebody there already reads in this space. Lead with the Verify rung of the Watch-Verify-Delegate ladder from your essay, the argument about what has to be true before a user earns the right to skip verification. Prove it with Carrier IQ's quote automation, where the same accept-or-correct decision sat between a model and a binding number. Your differentiator: you run enterprise agents in production at TinyFish right now, working traces, attribution and reversibility daily, so when they ask how you would design the confidence signal you answer from what breaks rather than from theory.
Verdict + action: Watch. Best-written brief in the cluster. Expires August 13. Apply, and put the essay link inside the application, not in a follow-up.

8. Guild.ai. Agent control plane: deployment, observability and governance surfaces through which enterprise buyers and developers understand and constrain agent behavior. Principal Product Designer.

Stage / Size / Capital: 2-10 employees per LinkedIn. A company funding claim exists and I could not corroborate it independently this pass. Treat capital as unverified.
Offices: San Francisco hybrid.
Value chain: Midstream tooling between model providers and enterprise deployments. Design influence is theoretically unlimited and practically contingent on whether the company is still here in eighteen months.
Best at: Too early to answer honestly. The mandate is right. There is not much company behind it yet.
Signals: Posted July 9. Day 23, confirmed live against primary ATS this pass. Watch justification: a Principal-level agent-governance brief at a sub-10-person company draws a small, self-selecting field. Flat environment, no established design function.
Tier: AI-native. Tier Playbook: Read AI-native playbook before outreach.
Rubric scores: Company 9/12 (AI 3, Stage/equity 1 because headcount signals execution risk rather than equity upside, Design influence 3, Trajectory 2). Role 15/15 (Total comp 3 at $240-290K plus bonus and equity, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Their user is an operator inspecting a live agent deployment and deciding whether to let it keep going. Lead with Retail Velocity, the field-intelligence surface where an operator acts on agent output under conditions they cannot fully verify, and pair it with three agentic products shipped at Agentic Labs, all published and inspectable. Your differentiator: most candidates for an agent control plane have used agents. You ship them into enterprise deployments at TinyFish and work the trace, attribution and reversibility questions when the behavior goes wrong.
Verdict + action: Watch, with an equity caveat I want you to sit with. A 2-to-10 person company hiring at Principal means you are the design function and the outcome is binary. Role score 15, stage dimension 1. That spread is the whole decision. Expires August 20.

9. HackerOne. Security platform extending into agent-to-agent security workflows with explicit human oversight requirements. Director, AI Product Design.

Stage / Size / Capital: Late-stage private. Round terms unresolved. Stage/equity scored 2 on inference from late-stage position and the posted SF band. Unverified.
Offices: San Francisco and other US locations.
Value chain: Midstream platform connecting a researcher community to enterprise security teams. The Director reports to the CPO, and that is the single most important structural fact in this entry.
Best at: Turning a distributed, adversarial researcher community into a governable enterprise input.
Signals: Original posting date remains unresolved. This publication flagged the same unresolved date last cycle and withheld action. Two cycles without a resolvable date is itself a signal. The req is probably older than the page implies. Verdict capped at Watch on that basis alone.
Tier: Enterprise platform. Tier Playbook: Read Enterprise platform playbook before outreach.
Rubric scores: Company 9/12 (AI 2, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 14/15 (Total comp 3 at $270-300K in SF plus equity, Scope 3, Craft depth 2, AI exposure 3, Portfolio value 3).
Portfolio match: The posting asks for transparency, accountability and scaling the design function, which is a design problem stapled to an org problem. Most candidates can speak to one of the two. Lead with the Alibaba structural-gap story told as an org story: you named the gap, argued the mandate into existence, then ran three cross-functional sprints across homepage, search and PDP. With a CPO, the part that lands is that the remit did not exist until you argued it into being. Your differentiator: agent-security fluency plus documented evidence that you have created a design mandate inside a company that was not offering one.
Verdict + action: Watch, date-capped. Do not apply cold. Ask a recruiter or a HackerOne design contact when the req opened before you spend the hour. If it opened inside three weeks, the rubric puts it at the front of this cluster immediately.

Cluster 3: Human handoff when automation quits

One real entry. Companies are hiring for this mechanism, almost always as a component of a sign-off role rather than as its own mandate. Giga is the exception and the strongest role on this list by the numbers.

10. Giga. Voice, chat and email AI agents deployed in enterprise and regulated customer operations. Staff Product Designer.

Stage / Size / Capital: $61M Series A led by Redpoint, Nov 2025. 51-200 employees. Stage/equity scored 3 on round size, velocity and the SF band rather than on a C–E label. Verify the option pool and the last 409A before you weight the equity.
Offices: San Francisco, on-site.
Value chain: Downstream application sitting directly on the customer relationship. When the agent fails, what breaks is the customer's opinion of the client's brand. That asymmetry hands design unusual internal leverage.
Best at: Running agents in customer operations where compliance constraints do not permit an experimental failure rate.
Signals: Posted July 15. Day 17, hot. The mandate includes mentorship and a minority of time supporting other designers, which confirms an existing team and an IC-plus-manager shape.
Tier: AI-native. Tier Playbook: Read AI-native playbook before outreach.
Rubric scores: Company 12/12 (AI 3, Stage/equity 3, Design influence 3, Trajectory 3). Role 15/15 (Total comp 3 at $240-260K, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Their failure mode is precise: an agent that keeps talking when it should have escalated, inside a regulated customer request. That is the Delegate rung of your trust ladder and its reversal condition, the section of "Trust Is the New Interface" arguing that delegation is only safe when the exit is designed before the entry. Ground it in Carrier IQ, where automated quoting had to yield to a human before a number became binding. Your differentiator: you are shipping enterprise web agents into production at TinyFish and working escalation behavior in live deployments, so you can describe escalation from inside a running system. No portfolio-only candidate can do that.
Verdict + action: Act. Move this week. A clean 12/12 on company and 15/15 on role is the top score anywhere on this board. Apply, then find the design lead and open on the escalation-exit argument. Hot band closes August 5.

Also carrying this mechanism: Maven Clinic (entry 1), where the handoff is clinical deferral instead of customer escalation.

Cluster 4: Time-critical alerting under load

A machine produces more findings than a human can act on, so the design problem becomes ordering, and getting the order wrong is an attributable failure with a name attached. Two entries, one of them geographically dead. The mechanism is real. The only other credible alerting mandate I found this cycle was New York-scoped.

11. Horizon3.ai. NodeZero autonomously attacks a customer's environment and hands exploitable attack paths to security and IT operations teams. Staff Product Designer.

Stage / Size / Capital: $100M Series D, $178.5M total, NEA-led.
Offices: San Francisco, New York or Washington DC hybrid.
Value chain: Downstream security application with its own autonomous engine. Value lands when remediation actually happens, which is a design outcome and not an engineering one.
Best at: Continuous autonomous penetration testing that produces proof of exploitability instead of a list of theoretical vulnerabilities.
Signals: Posted July 23. Day 9, hot. Partnership with other designers referenced. No team size, no named design leader.
Tier: AI-native. Tier Playbook: Read AI-native playbook before outreach.
Rubric scores: Company 10/12 (AI 3, Stage/equity 2, Design influence 2, Trajectory 3). Role 13/15 (Total comp 2 at $200-220K plus equity, Scope 2, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: The failure mode they name is remediation effort aimed at the wrong thing: noisy prioritization, false positives, missed paths. Signal ordering for an operator with limited hours. Your match is Retail Velocity, the CPG field-intelligence surface where the design question was which signal reaches the rep first and what gets held back. Pair it with Brand Pulse's real-time sentiment work for the volume-management pattern. Your differentiator: you have designed prioritization where a wrong ordering decision carried a measurable commercial cost, so you can talk about suppression as a design decision with an owner attached instead of a filtering preference.
Verdict + action: Act. Apply this week. Ask early who owns findings prioritization today. If the answer is the detection team, design's real ceiling here is lower than the rubric suggests.

12. Datadog. Infrastructure observability and developer surfaces. The Senior Staff Design Lab and Staff Modern Compute mandates are the two strongest alerting briefs outside Horizon3. Staff Product Designer, Modern Compute.

Stage / Size / Capital: Public. No pre-IPO equity.
Offices: New York only. Hard geographic fail.
Value chain: Midstream observability layer. Design owns how operators triage under volume, which is genuine scope, but a public-company equity picture caps the opportunity thesis regardless.
Best at: Making high-cardinality telemetry navigable by a human holding a pager.
Signals: Roles live, New York-scoped. No US-remote or Bay variant surfaced this pass.
Tier: Enterprise platform.
Rubric scores: Company 7/12 (AI 2, Stage/equity 1 as a public company, Design influence 2, Trajectory 2). Role 11/15 (Total comp 2, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Same ordering-under-load problem as Horizon3, so Retail Velocity anchors again. Not worth building further until geography moves.
Verdict + action: Ignore-until-trigger. Two gates fail at once. Trigger: a Bay Area or US-remote posting on a Datadog observability surface. This scan has now discarded three New York-only mandates on location alone.

Cluster 5: Attribution and provenance

The empty cluster. One entry, and it is dead. Read this one for what it corrects in the method.

13. Altana. Agent-first global trade systems covering provenance, confidence, consent and customs decisions under human-in-the-loop operation. Head of Product Design.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established as Bay-eligible in the retrieved record. Verify if it reposts.
Value chain: Downstream decisioning application over supply-chain data. Value concentrates at the customs and compliance decision, which the head of design would own outright.
Best at: Making the provenance of a trade claim inspectable at the moment someone has to act on it.
Signals: First published June 1. Day 61. Dead under our own rules, and I am not bending them for a good job description. This is the strongest provenance and human-in-the-loop language anywhere on the board and the sole reason this cluster came back empty.
Tier: AI-native.
Rubric scores: Company 10/12 (AI 3, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 14/15 (Total comp 2 inferred, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Provenance plus consent plus a consequential customs call is the Thermo Fisher governance-visible workflow moved into trade, with "Trust Is the New Interface" supplying the confidence-and-consent vocabulary they already wrote into the posting themselves.
Verdict + action: Ignore-until-trigger. The rubric clears Act. The decay gate kills it. Carry that pattern forward. Trigger: a repost, or any Altana design leadership requisition with a fresh first-published date. Set a monthly check on the Greenhouse board.

What the emptiness means for your scanning: provenance now lives inside professional sign-off roles, attached to the reviewer rather than to a surface of its own. That is where your Thermo Fisher and Carrier IQ work lands hardest anyway. Reallocate the scan time.

Cluster 6: Multi-product AI architecture

One entry, tied for the highest role score on this board at 15/15. Once a single AI pattern has to hold across four separately branded surfaces, the design problem stops being interaction and becomes architecture. The cluster is thin because most companies do not have four surfaces yet.

14. Babylist. Baby registry that expanded into commerce, media and health, now building AI across all of it. Director, Product Design, AI Builder.

Stage / Size / Capital: $750M+ in 2025 revenue and an eighth consecutive profitable year. 201-500 employees. Profitable and private, which is a materially different equity thesis than a venture-priced round. Ask about it directly and ask early.
Offices: Requisition lists US with no city named. Bay-eligible on its face. Verify the hub expectation in screening, because a US-wide Director listing frequently carries an unstated office preference that only surfaces at offer stage.
Value chain: Downstream consumer commerce with owned media and an expanding health surface. Value concentrates in the registry relationship, and every AI decision either strengthens or fragments it.
Best at: Owning a customer at a life-stage moment and extending that relationship in four directions without losing the thread.
Signals: Posted July 13. Day 19, hot. Reports to the VP/Head of Design and Research, leads three to four designers, stays hands-on. An IC-plus-manager mandate written out explicitly, which is rarer than it should be.
Tier: Growth-stage platform. Tier Playbook: Read Growth-stage platform playbook before outreach.
Rubric scores: Company 9/12 (AI 2, Stage/equity 2, Design influence 2, Trajectory 3). Role 15/15 (Total comp 3 at $244,600-305,700 plus 22.5% target bonus and equity, Scope 3, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: Their failure mode is an AI pattern that tests beautifully inside registry and then makes commerce, media and health feel like three unrelated companies. Lead with Allē: 30M members, dual-surface redesign spanning consumer and provider, 3.2× redemption, $42 CAC. One coherent pattern holding across audiences with opposing needs, and the redemption number proves the coherence paid for itself. Your differentiator: they want a leader who still designs, and Allē plus Equinox+ shows you shipping architecture-level consistency while running the team. That pool is narrower than it sounds.
Verdict + action: Watch. Role 15/15 against company 9/12 keeps it off Act, and the AI centrality of 2 is what holds it there. Hot band closes August 3. In the first conversation ask whether "AI Builder" means internal tooling or customer-facing surfaces. The answer rewrites the entire role.

Cluster 7: Governance-visible enterprise workflow

The densest cluster of fresh postings and the weakest on AI substance. Money, approvals and audit trails move through software, and someone senior decided the AI story needs a design owner. Read them closely: three of the five name no AI failure mode at all. Take these for the equity window and the craft. Not for the case study.

15. BuildOps. Operations and financial software for commercial contractors: inventory, procurement, accounts payable, payments and a "Financial OS" layer used in the office and in the field. Staff Product Designer.

Stage / Size / Capital: $127M Series C led by Meritech, Mar 2025, $1B post-money. $250M+ total. 201-500 employees.
Offices: San Francisco or Toronto.
Value chain: Vertically integrated operations platform. Value concentrates in the contractor's cash cycle, which runs straight through the surfaces this role owns.
Best at: Serving an office user and a field user inside one workflow without making either one carry the other's interface.
Signals: Posted July 30. Day 2, hot. Reports to the Head of Design, with leadership responsibility across domains and other designers.
Tier: Growth-stage platform. Tier Playbook: Read Growth-stage platform playbook before outreach.
Rubric scores: Company 10/12 (AI 2, Stage/equity 3, Design influence 2, Trajectory 3). Role 11/15 (Total comp 2 at $170-210K SF plus bonus, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: The posting gestures at trustworthy AI behavior across consequential financial workflows, but the concrete problem is a split-context user: a technician on a truck approving something a controller reconciles at a desk hours later, with different information and a different tolerance for friction. Thermo Fisher mySupply matches directly, because procurement there moved between lab floor and finance under exactly that split. $20M+ margin, 100% adoption in 12 months. Your differentiator: you designed one approval flow that had to work one-handed in the field and stand up to audit at the desk, and the adoption number proves neither side got sacrificed. That is a much narrower claim than "I have done procurement," and it is the one the Head of Design will recognize.
Verdict + action: Watch. Role score lands one point under Act, held down by the comp band and AI exposure. Freshest governance posting on the board. Apply now, before the field assembles.

16. Airwallex. Global business finance platform. The Spend product covers cards, expenses and approvals for administrators and employees. Staff Product Designer, Spend.

Stage / Size / Capital: Large private fintech, 1,001-5,000 employees. Public sources conflict on the latest valuation, roughly $11B to $12B. Treat the number as unresolved and the equity window as late.
Offices: San Francisco hybrid.
Value chain: Midstream financial infrastructure with a downstream admin product on top. Design owns the admin layer. Not the rails.
Best at: Multi-currency money movement for companies that outgrew their domestic bank.
Signals: Posted July 30. Day 2, hot. Establishes domain ownership. Names no AI-specific failure or reversal surface anywhere in the requisition.
Tier: Enterprise platform. Tier Playbook: Read Enterprise platform playbook before outreach.
Rubric scores: Company 8/12 (AI 1, Stage/equity 2, Design influence 2, Trajectory 3). Role 11/15 (Total comp 3 at $200-270K plus bonus and equity, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: One user is a spend administrator constructing an approval chain. The other is an employee trying to get reimbursed without reading a policy PDF. Lead with American Red Cross: six systems into one, $847K disbursed, national deployment. Money moving through an approval chain under audit, with non-expert claimants at the far end. Your differentiator: disbursement under scrutiny, where the approval trail and the claimant experience had to be one design instead of two.
Verdict + action: Watch. AI centrality of 1 means this role advances your positioning by zero degrees. Take it if the comp band and the SF office outweigh the case study you will not get.

17. Valon. Mortgage servicing platform. This role builds reusable product infrastructure used by servicing operators, deployment strategists and technical teams. Staff Product Designer, Product Infrastructure.

Stage / Size / Capital: $100M Series C led by WestCap, Oct 2024. $230M raised since 2019.
Offices: San Francisco, New York or US remote.
Value chain: Vertically integrated servicer plus platform. Value concentrates in regulatory compliance and cost per loan. Design supports rather than shapes.
Best at: Replacing decades-old servicing systems without breaking the regulatory reporting stacked on top of them.
Signals: Posted July 28. Day 4, hot. Mentorship responsibility. Function-joining, not function-building.
Tier: Enterprise platform.
Rubric scores: Company 7/12 (AI 1, Stage/equity 2, Design influence 2, Trajectory 2). Role 10/15 (Total comp 2 at $197-232K plus equity, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: Reusable primitives across regulated money-movement workflows is a design-system problem wearing a fintech costume. Equinox+ is the honest match: 0 to MVP in three months across five brands from one system.
Verdict + action: Scores Watch on the role threshold at 10/15. Overriding to Ignore-until-trigger, because AI exposure is a 1, the company sits at 7, and a design-system infrastructure role in mortgage servicing produces nothing that advances your next search. Trigger: an AI or decisioning mandate appears in a Valon design posting, or a design leadership role opens above this one.

18. Zscaler. Cloud security platform. The role covers data-dense, configuration-heavy product surfaces and asks for AI-assisted prototyping. Staff Product Designer.

Stage / Size / Capital: Public. No pre-IPO equity.
Offices: San Jose hybrid. $137,200-$196,000.
Value chain: Midstream security infrastructure. Design owns configuration surfaces, which is real craft under a low strategic ceiling.
Best at: Making a sprawling zero-trust policy surface configurable without a services engagement attached.
Signals: Posted July 29. Day 3, hot. Reports to a Senior Manager of Product Design alongside content design and research, which confirms an established specialist organization. The AI language refers to prototyping method, not product behavior. No agent approval, reversal or escalation right named anywhere.
Tier: Enterprise platform.
Rubric scores: Company 6/12 (AI 1, Stage/equity 1, Design influence 2, Trajectory 2). Role 8/15 (Total comp 1 below your floor, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 1).
Portfolio match: None worth building. Configuration density is the least differentiated thing you own.
Verdict + action: Ignore-until-trigger. Below both thresholds, and the top of the band sits under your floor. Trigger: a Director-plus Zscaler design role with an AI decisioning mandate and a band above $250K.

19. ServiceNow. Enterprise workflow platform. This role covers identity-security and AI-ready platform patterns. Senior Staff Product Designer, Identity Security Platform.

Stage / Size / Capital: Public. No pre-IPO equity.
Offices: Santa Clara hybrid. $190,900-$334,100 plus equity.
Value chain: Midstream enterprise workflow platform. Established UX organization spanning product design, research, content and systems. Influence exists and is distributed thinly across a very large function.
Best at: Turning enterprise process into configurable workflow at a scale nobody else operates at.
Signals: Original posting date unresolved, which caps any verdict independently. The posting names AI-ready platform patterns and no specific agent approval, escalation or reversal surface.
Tier: Enterprise platform.
Rubric scores: Company 6/12 (AI 1, Stage/equity 1, Design influence 2, Trajectory 2). Role 9/15 (Total comp 2 as public equity, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 1).
Portfolio match: Identity-security workflow patterns overlap your Alibaba platform work at the systems level, but a public-company platform role at this scope yields an incremental case study.
Verdict + action: Ignore-until-trigger. Company below threshold, date unresolved, no pre-IPO equity. Trigger: an AI-decisioning design leadership role at ServiceNow with a resolvable posting date.

Cluster 8: Builder-facing automation

The user constructs the automation instead of consuming it. Two Acts here, both earned on company strength rather than AI depth. Be clear-eyed about what you are trading: these pay well inside accelerating companies, and the case study they produce is weaker than what Giga or Fieldguide would hand you.

20. Render. Cloud platform where developers assemble, deploy and scale multi-service applications. Staff Product Designer.

Stage / Size / Capital: $100M Series C extension at $1.5B, Feb 2026. 51-200 employees.
Offices: US remote.
Value chain: Midstream infrastructure abstraction, historically a graveyard for design influence. Render is a partial exception, because the whole competitive claim is the quality of the abstraction, which pulls design upstream of where infrastructure companies usually file it.
Best at: Hiding the right amount of infrastructure. Not all of it.
Signals: Posted July 13. Day 19, hot. Concurrent senior-level variants share overlapping language, so read the multiple listings as level calibration, not multiple funded seats.
Tier: Growth-stage platform. Tier Playbook: Read Growth-stage platform playbook before outreach.
Rubric scores: Company 10/12 (AI 2, Stage/equity 3, Design influence 2, Trajectory 3). Role 12/15 (Total comp 3 at $218-300K plus equity, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: The failure mode the product itself names is an abstraction that conceals deployment state, dependency or operational consequence until it is too late to act. A visibility-versus-simplicity trade, and your strongest evidence is Equinox+: 0 to MVP in three months, five brands, 4.8 stars. A system that stayed simple at the surface and inspectable underneath. Your differentiator: your team ships on infrastructure of exactly this shape at TinyFish, so you can critique Render's current abstraction from use rather than from a heuristic walkthrough.
Verdict + action: Act. Apply this week. Hot band closes August 3.

21. Mintlify. Documentation platform where technical teams build knowledge surfaces consumed by humans and by AI agents. Senior/Staff Product Designer.

Stage / Size / Capital: $45M Series B led by a16z with Salesforce Ventures, Apr 2026, $500M valuation. 51-200 employees.
Offices: San Francisco or New York, on-site.
Value chain: Midstream tooling sitting on a genuinely interesting seam, since documentation is becoming the interface between one company and another company's agents.
Best at: Treating documentation as a product surface instead of a publishing chore.
Signals: Posted July 17. Day 15, hot. Emphasis on end-to-end product work and design systems. No provenance or agent-control mandate named, despite the AI-readable market positioning.
Tier: AI-native by market position, growth-stage platform by evaluation pattern. Tier Playbook: Read Growth-stage platform playbook before outreach. They will assess you on shipping speed and range, not on AI theory.
Rubric scores: Company 10/12 (AI 2, Stage/equity 3, Design influence 2, Trajectory 3). Role 12/15 (Total comp 2 at $170-265K, an unusually wide band; Scope 3, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: The interesting part of this job is unwritten. One page has to be legible to a developer skimming and to an agent parsing, and those two readers want opposite structures. Bring the five-handoffs framework from "Trust Is the New Interface" to the machine-reader half. Your differentiator is unusually specific: at TinyFish your agents are consumers of documentation like Mintlify's, so you can describe what makes a docs page fail an agent from the receiving end. Nobody else in their pipeline sits at that vantage point.
Verdict + action: Act. The wide band means level is negotiable, so anchor to Staff in the first conversation. Apply this week. Hot band closes August 7.

22. Zip. Procurement orchestration where administrators construct request, approval, sourcing, payment and reporting workflows. Senior/Staff Product Designer.

Stage / Size / Capital: $190M Series D led by BOND, Oct 2024, $2.2B valuation. No fresh round since.
Offices: San Francisco hybrid.
Value chain: Midstream orchestration across finance, legal and IT approvals. Design owns the configuration surface.
Best at: Getting the first procurement request routed correctly without a six-month implementation.
Signals: Posted July 6. Day 26. No compensation in the retrieved primary record.
Tier: Enterprise platform.
Rubric scores: Company 7/12 (AI 1, Stage/equity 2, Design influence 2, Trajectory 2). Role 9/15 (Total comp 1 with no range posted, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: Distinct from BuildOps despite the shared procurement vocabulary. Zip's user is a configurator, not an operator: someone writing the routing logic a stranger will later be bound by. The failure is a misconfigured workflow sending a consequential purchase to the wrong approver, and the victim is downstream and unaware it happened. Thermo Fisher mySupply's approval architecture across six partner organizations is the match, and the point to press is that you designed the routing logic and the requester's experience at the same time, which is precisely where most procurement tools break: the configurator gets a powerful editor and the requester gets whatever falls out of it.
Verdict + action: Ignore-until-trigger. Below both thresholds. Trigger: a fresh funding round, or a Zip design posting naming an AI decisioning mandate.

23. Bubble. Visual application building with AI assistance. Director of Product Design.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established as Bay-eligible in the retrieved record.
Value chain: Midstream builder tooling. A Director mandate here would carry real scope over how non-engineers construct software.
Best at: Letting a non-engineer ship a real application without a rewrite waiting at the end of it.
Signals: First published January 14. Six months old. Dead by a wide margin. This is the clearest case on the list of newly discovered not meaning newly posted, a distinction this publication has had to draw before.
Tier: Growth-stage platform.
Rubric scores: Company 9/12 (AI 2, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 12/15 (Total comp 2 inferred, Scope 3, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Builder-facing abstraction under a Director mandate maps to the Alibaba structural-gap story, told as function-building rather than as a redesign.
Verdict + action: Ignore-until-trigger. Trigger: a repost with a fresh first-published date and a confirmed Bay or US-remote location.

24. PermitFlow. Construction-permitting workflow software for builders and developers. Staff Product Designer, plus concurrent Principal and design-systems listings.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: New York hybrid. Hard geographic fail.
Value chain: Downstream vertical operations software. Value concentrates in permit cycle time, which design directly affects.
Best at: Turning a jurisdiction-by-jurisdiction bureaucratic maze into one predictable pipeline.
Signals: Three concurrent listings, which reads as level calibration or a systems hire rather than three independent product mandates. All New York.
Tier: Growth-stage platform.
Rubric scores: Company 8/12 (AI 1, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 11/15 (Total comp 2 inferred, Scope 3, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: Fragmented-system consolidation for a non-desk workforce, which is Red Cross territory: six systems into one, national deployment.
Verdict + action: Ignore-until-trigger. Trigger: a San Francisco or US-remote PermitFlow design posting. The trigger condition here is geographic availability, not role availability.

Cluster 9: Trust calibration for non-experts

The user has no domain expertise and no way to check the machine's work. Structurally this is the richest mechanism for your Red Cross and Equinox+ pillar, and the most disappointing cluster on the board. The two freshest roles carry thin AI mandates, one posts a band under your floor, and three fail on ownership structure before the mandate even matters.

25. Handshake. Career platform connecting students and early-career candidates to employers, now AI-mediated. Staff Product Designer.

Stage / Size / Capital: Late-stage private. LinkedIn shows a 501-1,000 band. Stage/equity scored 2 on inference from late-stage position. Unverified.
Offices: San Francisco, on-site.
Value chain: Downstream two-sided marketplace. Design owns both sides of the match.
Best at: Owning the university relationship, which is a distribution moat competitors cannot buy.
Signals: Posted July 30. Day 2, hot. Older indexed "AI Enterprise" and "Staff AI" role URLs no longer appear as distinct mandates on the current feed. Do not count them as separate seats.
Tier: Growth-stage platform. Tier Playbook: Read Growth-stage platform playbook before outreach.
Rubric scores: Company 8/12 (AI 2, Stage/equity 2, Design influence 2, Trajectory 2). Role 11/15 (Total comp 2 at $190-240K plus equity, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Their user is twenty-two, has no way to judge whether an AI-generated career recommendation is good advice, and is transacting on a platform whose other user is a recruiter with opposite incentives. Allē is the structural match: 30M members, dual-surface consumer and provider, 3.2× redemption at $42 CAC. You had to build trust for a consumer who could not evaluate the recommendation while keeping the professional side honest. Your differentiator: two-sided consumer trust at scale, with a redemption number proving the consumer side actually believed it.
Verdict + action: Watch. Freshest posting on the board. Apply, then ask in screening whether the AI mandate is a funded product direction or a feature request with a good title attached.

26. Affirm. Buy-now-pay-later platform building agentic shopping and discovery experiences. Staff Product Designer, Agentic Experiences.

Stage / Size / Capital: Public. No pre-IPO equity. That is the structural disqualifier, and no amount of mandate quality repairs it.
Offices: US remote. $195,000-$255,000 in the highest listed zones.
Value chain: Downstream consumer credit embedded at checkout. Design owns the moment a consumer accepts a financing decision they do not fully understand.
Best at: Underwriting at the point of purchase without a revolving-credit business model sitting behind it.
Signals: Posted July 13. Day 19, hot. The mandate names hybrid search and discovery, movement between conventional UI and conversational assistance, and asks explicitly for trust, clarity and predictable outcomes.
Tier: Enterprise platform. Tier Playbook: Read Enterprise platform playbook before outreach.
Rubric scores: Company 8/12 (AI 2, Stage/equity 1 as a public company, Design influence 3, Trajectory 2). Role 13/15 (Total comp 2, Scope 2, Craft depth 3, AI exposure 3, Portfolio value 3).
Portfolio match: The hard problem is the seam between browsing and asking: a user who starts by scrolling, switches to a question, and gets an answer that commits them to debt. That is the Alibaba homepage, search and PDP redesign almost line for line. Same hybrid browse-versus-query behavior, at $50B+ GMV, delivering +20% daily transactions and +2.2pt NPS. Your differentiator: you ran the search-and-discovery redesign itself, not an AI layer bolted onto someone else's. Affirm's team is solving the underlying discovery problem, and most agentic-AI candidates have never touched it.
Verdict + action: Watch, with the equity window scored honestly at 1. One of the three sharpest mandates on this board attached to an absent equity thesis. If you take a conversation, take it for the problem, and be clear with yourself about what you are giving up.

27. Coinbase. Regulated consumer crypto platform. This role sits in consumer experience and engagement. Staff Product Designer, Experience and Engagement.

Stage / Size / Capital: Public. No pre-IPO equity. Compensation not resolved this pass.
Offices: US remote.
Value chain: Downstream consumer application in a regulated category. Design owns the consumer's confidence in an irreversible transaction, which is genuinely interesting scope.
Best at: Making a category with catastrophic user-error potential feel operable by a first-timer.
Signals: Posted June 26. Day 36. The retrieved mandate established no AI approval, provenance or reversal mechanism anywhere.
Tier: Enterprise platform.
Rubric scores: Company 6/12 (AI 1, Stage/equity 1, Design influence 2, Trajectory 2). Role 10/15 (Total comp 2 unresolved, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: Non-expert trust in irreversible action is Red Cross and Allē territory, but with no AI mechanism the case study is a consumer-fintech redesign you could already write today.
Verdict + action: Scores Watch on the role threshold at 10/15. Overriding to Ignore-until-trigger, because the company sits at 6, there is no pre-IPO equity, the AI mandate is absent, and the posting is at day 36. Trigger: a Coinbase design role naming an AI decisioning or agent-supervision surface.

28. Flex. Consumer product letting renters split their largest recurring expense. Principal Product Designer.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference from the posted band and hiring level. Unverified.
Offices: San Francisco or New York hybrid. $280,000-$350,000 plus equity, the best band on the entire board.
Value chain: Downstream consumer fintech. Design owns the portfolio and the design system, partnering with existing design leadership.
Best at: Restructuring the timing of a payment people cannot skip.
Signals: Posted June 24. Day 38. The AI language in this role concerns design-process tooling, not AI behavior inside the product. Mature existing function, with mentorship and standards-raising across the portfolio.
Tier: Growth-stage platform.
Rubric scores: Company 7/12 (AI 1, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 10/15 (Total comp 3, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 1).
Portfolio match: Rent-splitting for non-expert consumers is a real trust problem and the role is not scoped to it. Pursue it anyway and Allē carries the consumer-scale craft.
Verdict + action: Scores Watch on the role threshold at 10/15. Overriding to Ignore-until-trigger, because AI exposure is a 1 and portfolio value is a 1. This role produces no case study at any comp level. If walking past a $350K ceiling stings, that is the right feeling to have about a job that would leave you positioning your next search on the same three pillars you already hold. Trigger: a Flex design mandate naming AI behavior inside the product rather than inside the design process.

29. GoodParty.org. AI-supported workflows for independent candidates and the citizens evaluating them. Staff Product Designer.

Stage / Size / Capital: Organized as a nonprofit. No conventional startup equity exists. Compensation not posted.
Offices: US remote.
Value chain: Downstream civic application. Design influence is high in a small mission-driven org.
Best at: Making campaign infrastructure available to people who cannot buy it.
Signals: Posted July 14. Day 18, inside the hot band. The mandate covers AI-supported political workflows used by people with no way to evaluate the guidance they are given.
Tier: Growth-stage platform.
Rubric scores: Company 8/12 (AI 2, Stage/equity 1, Design influence 3, Trajectory 2). Role 10/15 (Total comp 1 unposted, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Non-expert trust calibration with civic stakes runs genuinely close to Red Cross: national deployment, vulnerable users, $847K disbursed under scrutiny.
Verdict + action: Scores Watch on both thresholds. Overriding to Ignore-until-trigger, and I want the override visible. The equity dimension is not low here, it is structurally absent, and this search evaluates opportunities on long-term upside. A nonprofit fails that test as a category rather than as a score. Trigger: none that would change this. Revisit only if your criteria change.

30. Linktree. Creator link-in-bio platform with AI product features. Staff Product Designer, AI.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established as Bay-eligible in the retrieved record.
Value chain: Downstream consumer creator tooling. Design influence is moderate. Distribution is the moat, not the interface.
Best at: Owning creator identity at the single link everyone actually clicks.
Signals: The official page exposes a September 9, 2025 posting date. Nearly eleven months. The AI language is still live on the page, which is the trap exactly: a page that looks current because someone edited it, not because anyone reposted it.
Tier: Growth-stage platform.
Rubric scores: Company 8/12 (AI 2, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 11/15 (Total comp 2 inferred, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Consumer AI features for non-technical creators is Equinox+ territory, and thin: 0 to MVP in three months, 4.8 stars.
Verdict + action: Ignore-until-trigger. Trigger: a repost with a 2026 date and a confirmed US location.

31. Headspace. Consumer mental health and behavioral wellbeing app building AI-mediated support. Staff AI Product Designer.

Stage / Size / Capital: 501-1,000 employees. Ownership and employee equity structure unresolved this pass. Stage/equity scored 1 on that unresolved structure plus company maturity. Treat it as a flag, not a clearance.
Offices: US remote or San Francisco hybrid.
Value chain: Downstream consumer app with an enterprise benefits channel. Value has migrated toward the employer contract, and that migration usually pulls design authority away from the consumer surface.
Best at: Consumer-grade craft in a category that mostly ships clinical-grade ugliness.
Signals: Posted July 20. Day 12, hot. The posting asks for AI that feels natural, trustworthy and human-centered. It names no clinical escalation surface, no crisis path, and no commitment about what the AI will refuse to do. Read the absence.
Tier: Healthcare/Regulated.
Rubric scores: Company 7/12 (AI 2, Stage/equity 1, Design influence 2, Trajectory 2). Role 10/15 (Total comp 1 at $140,400-190,000, below your floor; Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Not worth building out. Pursue it anyway and Equinox+ carries the consumer-wellbeing craft while Red Cross carries the vulnerable-population credibility.
Verdict + action: Scores Watch on the role threshold at 10/15. Overriding to Ignore-until-trigger, because the band tops out below your floor and an AI mental-health mandate that never names an escalation path is a design problem somebody has not finished thinking through. Trigger: a Director-plus posting with a clinical-safety mandate and a band above $220K.

Cluster 10: 0 to 1 inside a regulated domain

Function-building mandates in constrained industries. Your third pillar, and the cluster is real but soft: thin AI content across all four, one with no compensation posted, one dead on date. Vetcove is the cleanest mechanical match to your work on this entire list, which makes the missing band genuinely irritating.

32. Vetcove. Purchasing and inventory platform where veterinary hospitals research, compare and buy clinical supplies. Staff Product Designer.

Stage / Size / Capital: 51-200 per LinkedIn's company band, though the same profile's discovery count runs materially higher. Treat the size estimate as platform-reported and internally inconsistent. Stage/equity scored 2 on inference. Unverified.
Offices: Remote. No compensation posted. Read that as a leveling-maturity signal rather than an oversight.
Value chain: Downstream marketplace between distributors and clinics. Value concentrates in the purchase decision, and design owns it end to end.
Best at: Making a fragmented multi-distributor supply market comparable on one screen.
Signals: Posted July 21. Day 11, hot. No design leader, no team count, no published design writing, no public design system located anywhere. This is a function-building mandate whether or not the title admits it.
Tier: Healthcare/Regulated. Tier Playbook: Read Healthcare/Regulated playbook before outreach.
Rubric scores: Company 8/12 (AI 1, Stage/equity 2 inferred, Design influence 3, Trajectory 2). Role 10/15 (Total comp 1, unposted; Scope 3, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: The failure mode is a procurement or inventory decision that leaves a clinical team without a product on the day they need it. That is Thermo Fisher mySupply. Same buyer, same consequence: six pharma partners, 100% adoption in 12 months, $20M+ margin. Your differentiator: nobody else in their pipeline has shipped clinical-supply procurement and can name what actually drives adoption in a clinic, which is that the ordering screen has to survive being used badly by someone in a hurry.
Verdict + action: Watch. Tightest domain match on this list, weakest AI exposure, unknown comp. Take one exploratory conversation with the hiring manager to establish band and design-function scope before you invest further. If they are building the function, negotiate the title.

33. Grow Therapy. Mental health marketplace connecting patients, therapists and practice operators, with conversational AI in the mix. Senior/Staff Product Designer.

Stage / Size / Capital: Growth-stage private. Terms unresolved. Stage/equity scored 2 on inference from the posted band and hybrid-hub footprint. Unverified.
Offices: San Francisco, New York or Seattle hybrid.
Value chain: Downstream three-sided marketplace with insurance intermediation. Design owns matching and the care experience.
Best at: Making insurance-covered therapy findable, which is a distribution problem dressed as a product.
Signals: Original posting date unresolved. Verdict capped at Watch on that basis. The posting describes a growing design team, gives no count, and states no reporting line for design.
Tier: Healthcare/Regulated. Tier Playbook: Read Healthcare/Regulated playbook before outreach.
Rubric scores: Company 8/12 (AI 2, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 11/15 (Total comp 2 at $182-230K plus equity, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: The structural fact the posting keeps quiet is that this is three-sided, and the third party, the insurer, pays without ever touching the product. Two-sided marketplace experience does not cover that. Thermo Fisher mySupply does: six pharma partners inside a procurement structure where the entity approving the spend was not the entity using the tool, and adoption still landed at 100% in 12 months. Back it with Red Cross's disbursement chain, where money moved through parties with different incentives and only one of them saw an interface. Your differentiator: you have shipped systems where the payer and the user were different people with conflicting definitions of success, which is the exact structure Grow Therapy's AI has to sit inside.
Verdict + action: Watch, date-capped. Verify the req opening date before you invest. In the first screen, ask which of the three sides owns the consequential AI decision. The posting never says.

34. Miter. Payroll and workforce management for construction and field-services employers. Staff Product Designer.

Stage / Size / Capital: Growth-stage private. Terms unresolved. Stage/equity scored 2 on inference from the posted band. Unverified.
Offices: San Francisco or New York hybrid.
Value chain: Vertical HR and payroll platform. Value concentrates in compliance accuracy, and design supports it.
Best at: Handling the certified-payroll and prevailing-wage rules that generic payroll products punt on.
Signals: Posted June 25. Day 37. Senior, hands-on, no AI decision surface named anywhere in the posting.
Tier: Growth-stage platform.
Rubric scores: Company 7/12 (AI 1, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 10/15 (Total comp 2 at $200-250K plus equity, Scope 2, Craft depth 3, AI exposure 1, Portfolio value 2).
Portfolio match: Pursue it and Red Cross is the match for consolidating fragmented workflows for a non-desk workforce: six systems into one, national deployment.
Verdict + action: Scores Watch on the role threshold at 10/15. Overriding to Ignore-until-trigger, because AI exposure is a 1, the company sits at 7, and the posting is at day 37 with no compensating specialization. Trigger: a Miter design posting naming an AI or automation decision surface, or a Director-plus opening.

35. Nourish. Nutrition care connecting patients and providers. Staff Product Designer.

Stage / Size / Capital: Private, growth stage. Terms unresolved. Stage/equity scored 2 on inference. Unverified.
Offices: Not established as Bay-eligible in the retrieved record.
Value chain: Downstream clinical-adjacent consumer application with an insurance channel. Design owns both patient and provider surfaces.
Best at: Getting nutrition care reimbursed and delivered inside the same flow.
Signals: The official record dates to February 12, 2025. Roughly eighteen months. Fully dead.
Tier: Healthcare/Regulated.
Rubric scores: Company 8/12 (AI 2, Stage/equity 2 inferred, Design influence 2, Trajectory 2). Role 11/15 (Total comp 2 inferred, Scope 2, Craft depth 3, AI exposure 2, Portfolio value 2).
Portfolio match: Patient-and-provider dual surface is Allē, and the regulated-reimbursement structure sits closer to Red Cross.
Verdict + action: Ignore-until-trigger. Trigger: a repost with a 2026 date.

What the four filters cost you this cycle

Seventeen entries above end in Ignore-until-trigger. They died on four filters, in this order of frequency.

Decay. Altana at day 61, Cognita at 58, Vitalize at 73, Bubble since January, Linktree since September 2025, Nourish since February 2025. Altana by itself is why the provenance cluster is empty. Internalize this: a page that has been edited is not a page that has been reposted, and job boards will not draw that line for you.

Geography. Datadog and PermitFlow are New York only, and this cycle also turned up matches in London, Stockholm, Bangalore, Raleigh and Alexandria. At Staff and above, geography is a gate rather than a preference. The trigger for these companies is a US opening, not a role opening.

Ownership structure. Affirm, Coinbase, Zscaler and ServiceNow are public, the stage-and-equity dimension scores 1, and mandate quality does not repair it. GoodParty.org is a nonprofit, where the dimension is absent rather than low. Two reference cases keep this filter ahead of the job description in the reading order: Smartsheet's $8.4B take-private and WW International's Chapter 11 filing. Ownership structure rewrites an equity thesis faster than any product news does.

Mandate. Valon, Zip, Miter and Flex all offer real craft and no AI decision surface. Flex pays the best band on this board and would produce nothing publishable.

What I verified, what I couldn't, what carries forward

Every posting cited under a live verdict was verified against its primary applicant-tracking URL in this pass. Funding and valuation figures come from named press or company sources except where flagged unresolved, and every stage-and-equity score resting on inference rather than a sourced round says so on its face.

Three entries carry unresolvable posting dates and are capped at Watch or Ignore on that alone, not on merit: HackerOne, Grow Therapy, ServiceNow. Seven have unverified capital terms. Ambience is the one I most want closed; a 14/15 role score deserves a known equity picture behind it. Put the round, the last 409A and the option-pool position in your first call.

One correction to carry past this cycle. The freshest postings and the best-matched postings are not the same set, and they need different channels. Past day 21 the field is assembled and a recruiter is triaging, which puts Fieldguide, Klarity, Maven and Ambience on the direct-contact path. Inside day 21 speed on the standard channel still pays, and three of the four Acts cross that line within a week.

Next cycle: whether the CPO-hire-to-design-posting lag holds at companies under 200 people, where the evidence has been thin in both directions.

Things to follow up on...

Early queue position pays: A field experiment in Management Science found hiring probability is highest when an applicant faces fewer existing competitors, and in observed cases the first interview came a median of six days after the first application — the evidence behind why day 2 beats day 22 on the standard channel.
Referrals beat inbound by 13x: Ashby's analysis of 38 million applications across 93,000 jobs put application-to-interview rates at 40% for referrals, 25% for sourced candidates, and 3% for inbound, which is the arithmetic argument for treating channel as a separate variable from posting age.
Day 43 is not abnormal: LinkedIn reports an average of 66 days to make a hire, and its own product organization once averaged 83 days merely to reach approved-for-hire, so an aged requisition is ordinary rather than dead — the decay rule governs your channel, not the opportunity.
Two new CPOs, no design posting yet: CrowdStrike named AJ Shipley chief product officer on July 15 with a mandate covering security for agents, models, and agentic workflows, and DocuSign's Graham Sheldon started July 6 with product, design, and research under him — both sit inside the 30-to-90 day window where design leadership postings historically follow.