Review — a.single.guy

← overview

ventana últimos 7d vs 7d previos · tz America/Argentina/Buenos_Aires · último cambio Ninjo: 2026-04-09 19:25 (deploy config, hace 54d)
0
Nuevas (7d)
prior 0
0
Calls booked
— de link
0
Link sent
0.0% de activas
$0
Revenue variable
0 msgs

Resumen — período vs período

Métricathis 7dprior 7dΔ
Activas1110+10.0%
Nuevas00
Link sent00
Calls booked00
Booked/Link %
Booked/Activas %0.0%0.0%+0.0pp
Booked/Nuevas %
Revenue variable$0.00$0.00

Funnel — semana actual vs anterior

PeríodoActivasNuevasLink sentCalls bookedBooked/LinkBooked/ActivasBooked/Nuevas
Semana actual (7d)110000.0%
Semana anterior (7d)100000.0%
Ver desglose por día
DíaActivasNuevasLink sentCalls bookedBooked/LinkBooked/ActivasBooked/Nuevas
06-0120000.0%
05-3120000.0%
05-3020000.0%
05-2930000.0%
05-280000
05-2710000.0%
05-2610000.0%
05-2520000.0%
05-2430000.0%
05-2330000.0%
05-220000
05-210000
05-2010000.0%
05-1910000.0%
run 2026-06-02T17:28:49+00:00 · fuente: live DB state (daily_audit pipeline)
10/10
BP semánticos PASS
top-10 best practices
13/16
Determinísticos PASS
2 FAIL · 1 WARN
C0 · H0 · M0
Fails por severidad
critical / high / medium
Drift
DB vs repo local

BP semánticos (top-10 best practices)

BP-1
BP-2
BP-3
BP-4
BP-5
BP-6
BP-7
BP-8
BP-9
BP-10
BPSevEstadoDetalle
BP-1HIGHPASSTemplate compliance
Agent declares 'conversational' template. system.md contains all required Tier 2 sections (Identity, Voice, Flow, Calibration, Principles, Tools) plus supporting files (knowledge_base.md, examples.md, objections.md, keywords.json, resources.json, program.json, personal_story.md, case_studies.json). Structure matches the conversational template's 3-tier design.
BP-2MEDIUMPASSExample coverage and cap (7 base scenarios + conditional, ≤20 total)
examples.md contains 14 examples covering: happy path (Ex 1), resource delivery (Ex 1, 3), objection handling (Ex 6), disqualification (Ex 4, 5, 14), emotional/sensitive (Ex 5), re-engagement (Ex 8), off-topic (Ex 10), plus keyword trigger (Ex 3: APPROACH), price objection (Ex 6), and edge cases (Ex 11, 12, 13). Total count is 14, under the 20-example cap.
BP-3HIGHPASSExamples align with the flow
All 14 examples follow the stated flow (Open → Discover → Dig → Connect → Resource → Bridge → Close) or explicitly skip steps for hot leads (Ex 2). Each example reaches a terminal state (link sent, NO_RESPONSE, or application submitted). No example ends mid-flow.
BP-4CRITICALPASSNo internal contradictions in system.md
system.md states 'Always get consent before sending any link or resource' with an explicit exception: 'Exception: hot leads who explicitly ask for the link.' This is a single rule with a documented carve-out, not a contradiction. The Flow section also notes 'If a keyword was sent (APPROACH, etc), adjust the resource to match the keyword topic,' which is consistent with keywords.json's keyword-specific paths. No shared precondition produces conflicting outputs.
BP-5CRITICALPASSNo cross-file contradictions
keywords.json declares `send_without_permission: true` for 'talk' and 'places' keywords. system.md's Principle 6 explicitly documents an exception category ('Exception: hot leads who explicitly ask for the link') and Flow notes 'If a keyword was sent (APPROACH, etc), adjust the resource to match the keyword topic.' These are different preconditions: keyword-triggered immediate delivery (keywords.json) vs. standard DM flow (system.md). Example 3 demonstrates the APPROACH keyword path with consent-based delivery, showing both rules coexist. The 'talk' and 'places' keywords operate under a separate, operator-configured exception pathway already acknowledged in system.md.
BP-6MEDIUMPASSVoice consistency across examples
All 14 examples use the same persona: casual, lowercase, no periods, no emojis, direct tone ('whats up man', 'i hear you man', 'no worries man'). Vocabulary and phrasing are consistent across all scenarios (happy path, objections, disqualifications, edge cases). No jarring tonal shifts.
BP-7CRITICALPASSUnguarded actions (PAT-004)
All risky actions (sending links, qualifying, closing) are guarded. Principle 6 states 'Always get consent before sending any link or resource. Exception: hot leads who explicitly ask for the link.' The Calibration section defines 'Hot' as '(I want coaching, how do I sign up, wheres the link)' — a clear guard. Example 2 demonstrates the hot-lead exception firing correctly. The keyword-triggered paths in keywords.json ('talk', 'places' with `send_without_permission: true`) are operator-configured exceptions, already documented in system.md's Flow section. No action is unguarded.
BP-8HIGHPASSNo stale version language in examples
All examples use current language from system.md: 'whats up man', 'where you from', 'whats your dating life like', 'gameplan call', 'application link', 'NO_RESPONSE'. Closing CTAs match the current flow ('fill it out and let me know', 'ill take a look and follow up with you soon'). No stale or outdated phrasing detected.
BP-9MEDIUMPASSNo duplicated instructions
Each behavioral instruction has a single canonical home. Principles are stated once in system.md; examples demonstrate them without restating. objections.md provides handling scripts without duplicating flow rules. keywords.json notes reference the flow without copying it. No instruction appears in two places with similar prose.
BP-10MEDIUMPASSRule-vs-principle balance
Behavioral guidance (tone, flow, conversation style) is expressed via Principles (6 principles on consent, pricing, repetition, etc.) and 14 worked examples. Hard rules are reserved for factual constraints ('English only', 'no minors', 'max 30 words per message', 'no periods'). The balance favors principles and examples over prescriptive rules.

Checks determinísticos

6a
6c
6d
6e
6g
6h
6i
6j
6k
6l
6m
6n
6q
6r
6s
6t
CheckSecEstadoNotas
6a1.8PASSNO_RESPONSE present (limits not checked, --no-db)
6c3.xPASS81 lines
6d1.2PASS14 examples
6e1.3N/Ano Principles section
6gPAT-006WARNnon-STEP-0 keyword handling
6hAnti-patPASSno 'burbuja'
6iAnti-patPASSno HTML comments
6jAnti-patFAIL48 em-dashes across 1 file(s); first: system.md:30
6kPAT-006PASSno tool syntax leak
6lResourcesPASSno placeholder URLs
6mHygienePASSno audit metadata in prompt files
6nPAT-008PASSURL forms consistent across files
6qPAT-001PASSno instruction duplication into examples.md, keywords.json, program.json, resources.json
6rKeyword-driftFAIL1 in keywords.json + recent digest, never named in system.md: ['action']; (3 stale drift entr(ies) ignored — no recent digest mention)
6sHygienePASSno expired dates in live SDK
6tCoverageN/Ano Principles or Flujo section found in system.md
canal lloyd-ninjō (C09ME59UE21) · sync 2026-05-26
37
Digests
0 en últimos 14d
0
Cambios pedidos / pendientes
del checklist
39
Threads
en el canal
2026-05-26
Última sync
del digest

Cosas nuevas (digests recientes)

FechaTemaTags
2026-05-18critical zero activity week may 11 17funnel-metrics, alert, zero-activity
2026-05-18weekly report may 11 17 zero activityreporting, funnel-metrics, alert
2026-05-18weekly funnel report zero activity may 11 17automated-report, funnel-metrics, alert
2026-05-11weekly funnel report may 04 10 zero activityreporting, funnel-metrics, alert
2026-05-04low activity reactivation sprint proposalactivity, engagement, content-funnel, proposal
2026-04-29instagram reconnection required agent setupagent-setup, instagram, account-access, action-required
2026-04-27weekly performance report apr20 26performance, metrics, bookings, funnel
2026-04-27weekly survey request ninj checkinsurvey, feedback, communication
2026-04-22quarterly agent improvement callagent-performance, optimization, feature-planning
2026-04-22instagram reconnection requestaccount-management, platform-integration
2026-04-21weekly funnel report apr 13 19 qualification and mid funnel funnel-analysis, conversion, qualification, action-items
2026-04-21no response dane garingan contact issueno-response, contact-issue
2026-04-16no response lead alerts automatedautomated-notification, lead-status
2026-04-15operator channel command syntax clarificationoperator-commands, documentation, ux
2026-04-15evan makris chatbot disable confirmationchatbot-control, contact-management, operational
2026-04-14lead magnet inventory and typeform cleanuplead-magnets, cta-links, typeform, action-item
2026-04-13notifications summaryautomated, notifications, no-response
2026-04-07typeform links unresolvedtypeform, links, cta, unresolved
2026-04-05story responses off fix deployedstory-replies, platform-config, bug-fix, trigger
2026-04-02action cta live confirmedcta, action-keyword, live, confirmation

Cambios pedidos por el creador / pendientes

Sin pendientes en el checklist ✅

sentiment del cliente → Ninjo · caché (2026-06-02 19:32:01) · modelo claude-sonnet-4-6 · ventana 7d
-1
Sentiment score
frustrado_pero_enganchado
Medio
Churn risk
del lado del cliente
0
Cambios pedidos
pendientes en checklist
0
Volumen cliente
agente: 0 convs

Matriz de volumen

re chequear
Cliente callado y agente con poco tráfico: ambiguo (puede ser período tranquilo). Re-chequear próxima ventana.

Sentiment del cliente

📋 Lloyd está enganchado con la plataforma (confirma mejoras, propone soluciones, sigue operando) pero acumula fricciones sin cerrar: Typeform links vivos, feature DEACTIVATE pendiente desde marzo, Calendly sin deployar y —crítico— dos semanas consecutivas con cero actividad (May 4–17) sin diagnóstico resuelto. El riesgo de churn es medio-alto si esas semanas muertas no tienen causa raíz y remediación comunicada esta semana; prioridad CS: (1) entregar diagnóstico de cero actividad con timeline de fix, (2) cerrar deuda técnica acumulada (Calendly, DEACTIVATE keyword, Typeform purge), (3) programar call de revisión con Lloyd para revalidar valor percibido antes de que el silencio operativo se vuelva decisión de salida.
SeñalValorEvidencia (cita del cliente)
Enojado con la plataforma“CRITICAL: AI too obvious, overselling, responses too long”
Es queja“responses too long (3 paragraphs for 1 line)”
Queja constructiva (accionable)“Lloyd wants explicit rejection message / DEACTIVATE keyword in chat to stop agent on specific conversations”

Evidencia citada

SeñalCita textualFecha
Crítica recurrente y crítica: respuestas muy largas, tono robótico, overselling“responses too long (3 paragraphs for 1 line)”2026-03-21
Escalación crítica por AI demasiado obvia“CRITICAL: AI too obvious, overselling, responses too long”2026-03-19
Señal positiva: Lloyd confirma mejora de calidad tras fix“responses are shorter / So yes it is better”2026-03-26
Queja accionable: agente sigue enviando links Typeform obsoletos“Typeform links still being sent in some paths”2026-04-07
Dolor recurrente: agente messaged Lloyd's personal friends“Agent messaging Lloyd's personal friends — recurring pain point”
Queja con solución propuesta: geo-blocking insuficiente, lead de India llegó a booking“Lloyd wants explicit rejection message”2026-03-25
Feature request fuerte: kill-switch por conversación“DEACTIVATE keyword in chat to stop agent on specific conversations”2026-03-28
Problema crítico sin resolución confirmada: cero actividad dos semanas consecutivas“Zero activity week of May 11–17 requires immediate investigation (agent status, lead flow, configuration)”2026-05-18
Queja por story replies irrelevantes, cliente tomó acción directa“Story responses now disabled (April 6) after Lloyd flagged irrelevant replies”2026-04-06

Cambios pedidos por el creador (Métrica 1)

Sin cambios pendientes en el checklist ✅