◆ Module 02 · Maestra Copywriting · Marketing · Branding · Psicología · Neurología

Maya does the CMO work
of a 20-person team.

6 production-grade capabilities. Brand-voice from your Foundation Pack. Zero buzzwords enforced at type-level. AI Act art 50 compliant. The 12M-impression playbook ran by a 20-person team · Maya runs it solo.

6

Capabilities live in production

~€0.005

Avg cost per generation · Sonnet 4.6

480k

Equivalent paid-ads spend saved · 12M impressions @ €40 CPM

0

Buzzwords allowed · hard-blocked at validator level

◆ Stage 2 · solution-aware · ¿cuál ha ganado?

Conoces el problema · sabes que existen soluciones · ¿pero qué arquitectura encaja en una venta regulada EU?

Maestra Copywriting · Marketing · Branding · Psicología · Neurología · Maya cubre las disciplinas que la primera generación de herramientas de AI sales dejó a medias. Aquí cómo vemos las cuatro arquitecturas del mercado · y por qué apostamos por el sistema operativo de ingresos.

ArquetipoEjemplosEstado 2026Por qué
AI SDR fully-autonomousfirst-gen autonomous SDRsretention pressuresin memoria persistente · sin handoff humano · sin compliance ni observability nativos
Copilots de textotext copilotssobreviven · upsell cappedañaden velocidad al admin · NO lo absorben
Voice infrastructureVapi · Bland · RetellPaaS commoditycapa primitiva · NO sistema operativo · no orquesta cross-channel
Revenue Operating SystemROSS (Maya + hermanas)nuestra apuestaabsorbe el admin repetitivo · audit trail SHA-256 · AI Act nativo · FinOps por acción

◆ What Maya does · 6 production capabilities

Each one wired to your Foundation Pack.
Zero generic output.

01

POST /api/maya/generate · /api/maya/spark-email

SPARK email + sequence

Cold email + 4-message sequence drafts · Foundation Pack voice

  • Spanish or English · auto-detected from ICP
  • Hook + body + CTA enforced separation
  • Zero buzzwords (synergy · leverage · journey · ecosystem) hard-blocked
  • 4-message sequence: full-framework · new-proof · new-angle · breakup (days 0/3/7/14)
Workflows.io SPARK8 hook patterns canonOPPA (DEC-V11-43)4-line rule

Output: ≤90 words · subject ≤5 words · anti-pattern reject

02

POST /api/maya/repurpose-waterfall

Repurposing waterfall

1 newsletter → 12 derivatives · LinkedIn + X + YouTube + carousel + blog

  • Vary hook patterns across the 5 LinkedIn pieces (≥3 different)
  • Each derivative has a different angle · enforced by validator
  • Numbers from the newsletter only · zero fabrication
  • ~€0.05 per waterfall · single Sonnet 4.6 call
Koen Stam waterfallWorkflows.io 1→12 multiplicationWelsh newsletter-firstGary Vee 64-piece pyramid (B2B-adapted)

Output: 5 LinkedIn + 3 X threads + 2 YT Shorts + 1 carousel + 1 SEO blog

03

POST /api/maya/viral-radar

Niche Viral Radar

Weekly niche post classifier · top 2 by comments · 5 ranked recommendations

  • Paste 3-5 creators × last 7 days of posts · classify each
  • Top-2 sorted by comments (sentence-grade signal)
  • Recommendations: dominant LM · hook intersection · hot topic · gate mechanic · anti-pattern
  • Trigify auto-fetch pending (PENDING_EXTERNALS D-bis)
GTM Architects 2026 skill4-dimension classifier (LM type · hook · topics · gate)STEP 6 heuristic

Output: Markdown brief · 4 aggregate tables · top 2 posts · 5 actions

04

POST /api/maya/post-to-ad

Post-to-Ad pipeline

Top organic post → 3 Sponsored Content variants · different hook patterns

  • Engagement gate: ≥10 combined reactions+comments enforced
  • Each variant uses a different hook pattern · same body intent
  • Persisted as draft campaign_specs · push goes live with 43A approval
  • Cerbos tier-mapped: Autopilot+ can publish · Startup gets view + first-touch only
DEC-V11-43 action 42K + 43D step 1Omar evidence-rule validator

Output: 3 ad variants → ross_growth.campaign_spec (state=pending_review)

05

POST /api/maya/content-calendar?mode=sunday_plan

Sunday-plan calendar

Weekly backlog · 50/25/25 TOF/MOF/BOF split · varied hooks

  • Topics typed at enum level (12 canonical) · TOF: giveaway · build-in-public · personal-story · thought-leadership
  • Funnel-split drift detector recommends correction when >15% off target
  • Each post lands in ross_content.content_post · state=backlog (review before schedule)
  • Format auto-recommended per topic (carousel for frameworks · short-video for stories)
12M-impression playbook (20-person team)12 canonical topics × 6 formatsCalendar discipline canon

Output: 5-7 posts per week · scheduled_for slot · funnel-split health

06

POST /api/maya/profile-audit

Profile audit

6-element LinkedIn scorecard · 0-100 with weights

  • Cover slide (15%) · Picture (10%) · Headline (25%) · CTA buttons (15%) · About (20%) · Featured (15%)
  • Paste-mode (LinkedIn blocks unauth scraping · Unipile auto-pull pending)
  • Anti-pattern detection: buzzwords cap headline/about at 50
  • Persisted to ross_content.profile_audit · history view per team member
12M-impression playbook · profile optimization section

Output: Total score + per-element subscore + ranked recommendations

◆ Why Maya · 4 non-negotiables

What Maya refuses to do.

Zero buzzwords · enforced at validator

synergy · leverage · journey · ecosystem · hard-blocked. Refuses output containing them.

Drafts-not-sends · audit-trail mandatory

Every output lands in pending_approval state. Human reviews before LinkedIn / Smartlead / Cal.com fires.

Foundation Pack voice-cast · no generic output

7-file tenant Foundation Pack (market_positioning · audience_delight · creator_style · founder_voice · customer_journey · deal_motion · claude_md) injected on every call.

AI Act art 50 disclosure when output is voice-bound

When Maya feeds Sara (voice), watermark + disclosure applied. EU pre-deadline 2026-08-02.

◆ Maya · 3 superpowers

Maya escribe. Maya distribuye. Maya califica.

01 · Pillar pyramid

1 → 64 micro-pieces

Pillar show distribution canon Gary Vee · 1 long-form fuente · 64 micro-pieces cross-channel · LinkedIn · X · IG · YouTube Shorts.

02 · Schwartz 5 levels

Awareness stratified

Unaware · problem-aware · solution-aware · product-aware · most-aware · cada copy targeting un nivel exacto · zero spray-and-pray.

03 · QA scorer

8.1+ threshold

Voice calibration per-tenant corpus · auto-retry 3× hasta score ≥8.1 · zero output bajo umbral publicado.

Pillar → micro-pieces

Una pieza pillar se reparte en derivados cross-channel, cada uno con un ángulo distinto · validador anti-fabricación.

Share of AI voice

Mide tu presencia en respuestas generativas (ChatGPT · Perplexity · Claude) sobre tu categoría · seguimiento por tenant.

QA antes de publicar

Cada salida pasa el QA scorer contra el corpus de voz del tenant · auto-retry hasta superar el umbral · nada sale por debajo.

Ver Maya en cockpit →

Maya is one of 4 ROSS agents

See what Maya does for your brand.