convez.ai
AgenteAutomaçãoAvançado
j-manchesterllc

SETH — Strategic Executive Technology Hub

j-manchesterllc/SETH

SO de IA para executivos com memória persistente e auditoria de marca

Ver no GitHub
InicialPluginScore · de 0 a 100

Sistema de auditoria de marca e inteligência estratégica baseado em orquestração determinística de contexto e ablação estatística. SETH funciona como infraestrutura soberana de IA que mantém contexto operacional persistente (memórias, padrões comportamentais, mapas relacionais) composto a cada interação, eliminando a fragmentação entre chatbots, assistentes e ferramentas de produtividade.

Instalação

$npx convez-ai j-manchesterllc/SETH

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
3
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 4 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Auditoria estratégica de marca com scored de confiança
  • Automação de pesquisa web e extração de dados via browser headless
  • Detecção de padrões comportamentais e mapeamento de relacionamentos
  • Inteligência competitiva e análise de contradições estratégicas
  • Gerenciamento de voz de marca com testes de ablação

Veredito

SETH é uma arquitetura ambiciosa que funde agentic AI, automação web e inteligência comportamental em uma stack executiva. Excelente para quem precisa de memória contextual composta e auditoria de marca determinística, mas demanda infraestrutura robusta (roteamento multi-modelo, armazenamento soberano) e curva de operacionalização significativa.

README

A tradução para o português deste README ainda está na fila.

SETH — Strategic Executive Technology Hub


What is SETH?

SETH replaces the fragmented stack of AI chatbots, human assistants, and productivity tools with a single sovereign intelligence infrastructure that sits between an executive and the world.

Unlike consumer AI assistants that reset every session, SETH maintains a persistent operational context — memories, behavioral patterns, relationship maps, brand voice profiles, and strategic contradictions — all private, all compounding.

After 90 days of use, SETH knows your decision patterns, strategic priorities, communication style, and operational rhythm better than any human assistant could. That's not a feature — it's a structural moat.


Architecture

┌─────────────────────────────────────────────────────────┐
│                    SETH PWA Shell                        │
│  Voice Control · Offline Support · Wearable-Ready       │
├─────────────────────────────────────────────────────────┤
│                                                         │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐              │
│  │Executive │  │  Agent   │  │  Cortex  │              │
│  │  Chat    │  │  Swarm   │  │  Intel   │              │
│  │          │  │ 7 agents │  │ Patterns │              │
│  └──────────┘  └──────────┘  └──────────┘              │
│                                                         │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐              │
│  │  Brand   │  │ Browser  │  │  Watch   │              │
│  │ Manager  │  │Automate  │  │  System  │              │
│  │ Ablation │  │ Headless │  │ Proactive│              │
│  └──────────┘  └──────────┘  └──────────┘              │
│                                                         │
│  ┌──────────┐  ┌──────────┐                            │
│  │Environ-  │  │ Unified  │    Multi-Model Routing     │
│  │ ments    │  │Integration│   Venice → OpenRouter      │
│  │          │  │Cal+Gmail │   Privacy-first fallback    │
│  └──────────┘  └──────────┘                            │
│                                                         │
├─────────────────────────────────────────────────────────┤
│  Sovereign Data Store · Evidence-Derived Confidence     │
│  Memory Decay Curves · Relationship Graphs              │
└─────────────────────────────────────────────────────────┘

Core Modules

ModuleDescription
Executive ChatContext-aware AI with strategic memory, voice control, and persistent operational context
Agent Swarm7 specialized agents — Architect, Sentinel, Quartermaster, Navigator, Diplomat, Chronicler, Vanguard
Cortex IntelligenceBehavioral pattern detection, contradiction alerting, relationship mapping, strategic insights
Brand ManagerVoice auditing, competitive intel, content strategy with ablation-tested confidence scoring
Browser AutomationHeadless browser engine for data extraction, form submission, web research
Watch SystemContinuous monitoring with autonomous alerting — proactive, not reactive
EnvironmentsImmersive command interfaces tailored to work context
Integration LayerNative Google Calendar, Gmail, and productivity tool integration

Key Technical Features

  • Multi-Model Routing — Privacy-first Venice.ai + OpenRouter fallback chain with free/paid tiers
  • Dual-Regime Ablation Testing — Deterministic (temp=0) observations + stochastic (production temp) statistical verdicts
  • Evidence-Derived Confidence — Pre-LLM computation: coverage, freshness, anchor strength, conflict penalty
  • Memory Consolidation — Decay curves with recency × importance, semantic enrichment, entity extraction
  • Structural Conflict Detection — 4 rule-based heuristics (goals-vs-tasks drift, calendar overload, avoidance patterns, email volume)
  • Context Weighting Profiles — Per-audit-type weight multipliers with SHA-256 versioned profiles
  • Resilient DB Operations — Auto-reconnect with retry logic surviving idle-session timeouts
  • Progressive Web App — Offline-capable service worker, installable, wearable-ready voice pipeline

Getting Started

Prerequisites

  • Node.js 18+
  • PostgreSQL database
  • API keys for model providers (see .env.example)

Setup

# Clone the repository
git clone https://github.com/YOUR_USERNAME/seth.git
cd seth

# Install dependencies
yarn install

# Configure environment
cp .env.example .env
# Edit .env with your credentials

# Set up the database
yarn prisma generate
yarn prisma db push

# Seed initial data
yarn ts-node scripts/seed.ts

# Start development server
yarn dev

Environment Variables

See .env.example for all required and optional configuration.

Required:

  • DATABASE_URL — PostgreSQL connection string
  • NEXTAUTH_SECRET — Session encryption key
  • VENICE_API_KEY or OPENROUTER_API_KEY — At least one model provider

Optional:

  • GOOGLE_CLIENT_ID / GOOGLE_CLIENT_SECRET — Google SSO
  • ELEVENLABS_API_KEY — Voice responses
  • BROWSERLESS_API_TOKEN — Browser automation
  • SKYBOX_API_KEY — Environment generation

Deployment

SETH is currently deployed across three surfaces:

DomainPurpose
sethassistant.digitalPrimary product domain
zero-daydynamics.comCorporate / brand domain
jarvisaiassistant.abacusai.appPlatform deployment

Project Structure

├── app/                    # Next.js App Router pages & API routes
│   ├── api/               # Backend API endpoints
│   │   ├── brand/         # Brand manager + ablation testing
│   │   ├── chat/          # AI chat with tool execution
│   │   ├── cortex/        # Cognitive intelligence layer
│   │   ├── tasks/         # Task management + auto-execute
│   │   └── ...            # Auth, memories, watches, agents, etc.
│   ├── chat/              # Chat interface
│   ├── brand/             # Brand manager dashboard
│   ├── cortex/            # Cortex intelligence dashboard
│   ├── investors/         # Investor executive summary
│   └── ...                # Tasks, memories, agents, etc.
├── components/            # React components
│   ├── brand/             # Brand manager UI
│   ├── chat/              # Chat interface components
│   ├── dashboard/         # Dashboard shell & navigation
│   └── ui/                # Shared UI primitives
├── lib/                   # Core business logic
│   ├── brand-manager.ts   # Context intelligence engine
│   ├── browser-automate-core.ts  # Headless browser automation
│   ├── cortex.ts          # Cognitive intelligence layer
│   ├── model-router.ts    # Multi-model routing & fallbacks
│   ├── tools.ts           # AI tool definitions & executors
│   └── venice.ts          # System prompt & message building
├── data/                  # Knowledge base documents
├── prisma/                # Database schema
├── scripts/               # Seed scripts
└── public/                # Static assets & PWA manifest

Revenue Model

TierPriceTarget
Operator$297/moIndividual executives
Principal$997/moHigh-volume executives with custom agents
EnterpriseCustomMulti-seat teams with SSO/SAML

License

Proprietary. All rights reserved.


Ver o original no GitHub

Bom para

Relacionados