The context
Eve's Essentials is the proof-of-concept for the AI Marketing Operator playbook I deploy for client brands and agency partners. Every pattern in the stack was battle-tested in live commercial conditions on this brand before being adapted for partner organizations.
I started it in January 2025. The premise was simple: fine-tuned zodiac apparel that treats the astrology consumer as a taste consumer, not a novelty consumer.
What made the premise unusual was the operating model. From day one, I was clear that this wouldn't be a founder-plus-team operation. It would be a founder-plus-AI operation — and everything that got built had to work under that constraint.
What I built
An agentic marketing stack where each layer does one thing well and Claude sits at the coordination layer.
- ReasoningClaude Cowork — brief writing, creative direction, campaign strategy, copy for every touchpoint, ad-hoc analysis.
- ExecutionMake.com — Meta CAPI events, Shopify inventory triggers, Klaviyo enrollments, Pinterest sourcing pipelines.
- LifecycleKlaviyo — welcome, browse, cart, post-purchase, winback, with dynamic content that adjusts to Shopify state.
- MediaMeta APIs — ad account management, campaign build, creative swaps, CAPI verification, audience refreshes.
- CommerceShopify — source of truth for product, order, customer.
- MeasurementGA4 + Meta Pixel — cross-verified against server-side events.
- CreativeCanva + Figma — templates AI populates with brand-consistent variants.
Nothing here is exotic. What's different is the seam between them. Every piece is invoked, cross-referenced, or updated through Claude, so the operator's time goes into decisions, not execution.
The creative slate
Every asset that ran through the campaigns was generated, directed, and produced through the same stack.
- MotionShort-form brand videos generated in Veo + Google Flow — product-first, character-consistent, platform-native at 4:5 and 9:16.
- StillEditorial imagery generated and upscaled in Nano Banana — Mr Porter / Fear of God / Acne Studios register.
- Ad creativeFTC-compliant social-proof ads produced in Canva with a 12-variant copy library maintaining brand voice across formats.
- CopyProduct descriptions, ad copy variants, lifecycle emails, and landing page copy all authored through Claude with codified brand voice references.
Selected reels and stills are featured on the Studio page.
Brand & infrastructure
Beyond the marketing stack — the identity, systems, and vendor operations that make the brand possible.
- Brand kitFull identity system — logo, color, typography, brand voice guide. View the live kit →
- Campaign designFull Meta campaign brief buildout via Claude — strategic architecture, audience segmentation, creative direction, copy variants, and UTM structure authored end-to-end before a dollar of spend went live. Same brief pattern applied to both Traffic and Sales campaign flights.
- Skill filesStructured brand voice, editorial standards, working preferences, and operational history codified as skill files — treating context as a system rather than per-conversation re-explanation.
- POD migrationFull vendor migration Gooten → Printify on SwiftPOD. SKU mapping, shipping zone reconfiguration, test order validation, Klaviyo Placed Order event verification.
- AttributionPixel + CAPI server-side attribution designed around Shopify Customer Accounts redirect constraints. Multi-source attribution architecture across Meta and Klaviyo truth-source revenue.
- Multi-model orchestrationClaude, Gemini, and OpenAI applied as strategic thought partners — drafting growth strategies, quarterly plans, GA4 dashboards, multi-channel campaign architectures, and stress-testing performance hypotheses before budget commitment.
How it operates in practice
A typical day: I check yesterday's performance in GA4 and Meta Ads Manager. Claude drafts the day's creative variants and pushes them to Meta via API. Klaviyo flows run themselves against Shopify events; if a threshold trips — an open-rate drop, a conversion spike on a segment — I get a summary and decide whether to intervene.
Pinterest sourcing runs on a Make scenario I check weekly. New products get spun up in a documented workflow that Claude handles end-to-end — description, variants, pricing tiers, Klaviyo enrollment, Meta ad creative — from a single brief.
The result isn't "faster marketing." It's marketing with an entirely different unit economics — because the constant-cost human layer is now one person, and the variable-cost execution layer is AI.
Outcomes
The agentic playbook was validated across two consecutive Meta campaigns.
- Traffic flight44 days, $540 spend, 92,974 impressions across three audiences. 4.4% blended CTR (4x fashion DTC average; 2x "excellent" tier). $0.14 CPC. 3,854 link clicks, 2,037 landing page views.
- Top creativeBaby Shower Onesie held 5.30% CTR over 14,000 impressions with 407 landing page views.
- Sales campaignDay-0 hit 6.46% CTR (3x "excellent" tier) at $0.57 CPC.
- Peak creativePisces Onesie video sustained 8.47% CTR at scale.
- Infrastructure10+ integrated MCPs across Meta Marketing API, Shopify GraphQL, Klaviyo, Make.com, Pinterest, GA4, Microsoft Clarity, Loox. 7+ scheduled tasks running daily and weekly diagnostics.
- Morning standupAutomated cross-reference of Klaviyo + Shopify truth-source revenue against Meta performance, auto-flagging stop-loss triggers against industry benchmarks — before 9:06 AM daily.
What this proves
That "AI-augmented marketing" isn't a productivity story — it's a structural one. The old marketing org chart assumed a division of labor between strategy, creative, media, lifecycle, analytics, and ops because no single person could hold all of those functions at scale.
The AI-augmented org chart doesn't need the division. What it needs is an operator who can hold the full stack in her head and direct AI to execute the layers that used to require separate specialists.
For founders building their next thing, this is the operating model. For CMOs restructuring their teams, this is what "AI-native" actually looks like in production.