EnterpriseHub

Executive Summary

EnterpriseHub is an AI-powered real estate platform that transforms lead management and business intelligence for real estate professionals and agencies. By automating lead qualification, follow-up scheduling, and CRM synchronization, EnterpriseHub eliminates the 40% lead loss caused by slow response times.

Key Benefits:

• Instant Lead Qualification: Three specialized AI bots (Lead, Buyer, Seller) qualify prospects in real-time using a proven Q0-Q4 framework, enforcing the critical 5-minute response SLA
• Unified Operations: Consolidate qualification results, CRM updates, and analytics into one platform—replacing fragmented spreadsheets and disconnected dashboards
• Actionable Insights: Streamlit BI dashboards provide real-time visibility into lead flow, conversion rates, commission tracking, and bot performance metrics

Target Audience: Real estate teams, brokerages, and agencies seeking to scale operations while maintaining personalized client engagement.

Business Impact: Production-ready with 89% token cost reduction, 87% cache hit rate, and P95 latency under 2 seconds. The platform integrates seamlessly with GoHighLevel CRM and supports multi-LLM orchestration (Claude, Gemini, Perplexity).

Quick Start: Launch the demo in seconds with make demo—no API keys or database required. For full deployment, complete setup in under 10 minutes using Docker Compose.

---

Real estate teams lose 40% of leads because response time exceeds the 5-minute SLA. This platform automates lead qualification, follow-up scheduling, and CRM sync so no lead goes cold.

Demo Snapshot

What This Solves

• Slow lead response -- Three AI bots (Lead, Buyer, Seller) qualify prospects in real time using a Q0-Q4 framework, enforcing the 5-minute response rule
• Disconnected tools -- Qualification results, CRM updates, and analytics live in one platform instead of spreadsheets + separate dashboards
• No visibility into pipeline health -- Streamlit BI dashboard surfaces lead flow, conversion rates, commission tracking, and bot performance metrics

Service Mapping

• Service 4: Multi-Agent Workflows (Agentic AI Systems)
• Service 6: AI-Powered Personal and Business Automation
• Service 8: Interactive Business Intelligence Dashboards
• Service 10: Predictive Analytics and Lead Scoring

Certification Mapping

• IBM Generative AI Engineering with PyTorch, LangChain & Hugging Face
• IBM RAG and Agentic AI Professional Certificate
• Duke University LLMOps Specialization
• Google Data Analytics Certificate
• IBM Business Intelligence Analyst Professional Certificate

Screenshots

Architecture

graph TB
    subgraph Clients["Client Layer"]
        LB["Lead Bot :8001"]
        SB["Seller Bot :8002"]
        BB["Buyer Bot :8003"]
        BI["Streamlit BI Dashboard :8501"]
    end

    subgraph Core["FastAPI Core — Orchestration Layer"]
        CO["Claude Orchestrator<br/><small>Multi-strategy parsing, L1/L2/L3 cache</small>"]
        AMC["Agent Mesh Coordinator<br/><small>22 agents, capability routing, audit trails</small>"]
        HO["Handoff Service<br/><small>0.7 confidence, circular prevention</small>"]
    end

    subgraph CRM["CRM Integration"]
        GHL["GoHighLevel<br/><small>Webhooks, Contact Sync, Workflows</small>"]
        HS["HubSpot Adapter"]
        SF["Salesforce Adapter"]
    end

    subgraph AI["AI Services"]
        CL["Claude<br/><small>Primary LLM</small>"]
        GM["Gemini<br/><small>Analysis</small>"]
        PP["Perplexity<br/><small>Research</small>"]
        OR["OpenRouter<br/><small>Fallback</small>"]
    end

    subgraph RAG["Advanced RAG System"]
        BM25["BM25 Sparse Search"]
        DE["Dense Embeddings"]
        RRF["Reciprocal Rank Fusion"]
        VS["ChromaDB Vector Store"]
    end

    subgraph Data["Data Layer"]
        PG[("PostgreSQL<br/><small>Leads, Properties, Analytics</small>")]
        RD[("Redis<br/><small>L2 Cache, Sessions, Rate Limiting</small>")]
    end

    LB & SB & BB -->|"Qualification<br/>Requests"| Core
    BI -->|"Analytics<br/>Queries"| Core
    Core -->|"CRM Sync"| CRM
    CO -->|"LLM Calls"| AI
    CO -->|"Retrieval"| RAG
    Core -->|"Read/Write"| Data
    RAG --> VS
    HO -->|"Bot Transfer"| Clients

Loading

Key Metrics

Metric	Value
Test Suite	4,500+ automated tests
LLM Cost Reduction	89% via 3-tier Redis caching
Orchestration Overhead	<200ms per request
API P95 Latency	<300ms under 10 req/sec
Cache Hit Rate	>85% for repeated queries
CRM Integrations	3 (GoHighLevel, HubSpot, Salesforce)
Bot Handoff Accuracy	0.7 confidence threshold

Quick Start

git clone https://github.com/ChunkyTortoise/EnterpriseHub.git
cd EnterpriseHub
pip install -r requirements.txt

# Demo mode — no API keys, no database, pre-populated dashboards
make demo

Full Setup (with external services)

cp .env.example .env
# Edit .env with your API keys

docker-compose up -d postgres redis
uvicorn app:app --reload --port 8000

# BI Dashboard (separate terminal)
streamlit run admin_dashboard.py --server.port 8501

Tech Stack

Layer	Technology
API	FastAPI (async), Pydantic validation
UI	Streamlit, Plotly
Database	PostgreSQL, Alembic migrations
Cache	Redis (L1), Application memory (L2), Database (L3)
AI/ML	Claude (primary), Gemini (analysis), OpenRouter (fallback)
CRM	GoHighLevel (webhooks, contacts, workflows)
Search	ChromaDB vector store, BM25, hybrid retrieval
Payments	Stripe (subscriptions, webhooks)
Infrastructure	Docker Compose

Project Structure

EnterpriseHub/
├── ghl_real_estate_ai/           # Main application
│   ├── agents/                   # Bot implementations (Lead, Buyer, Seller)
│   ├── api/routes/               # FastAPI endpoints
│   ├── services/                 # Business logic layer
│   │   ├── claude_orchestrator.py    # Multi-LLM coordination + caching
│   │   ├── agent_mesh_coordinator.py # Agent fleet management
│   │   ├── llm_observability.py      # LLM cost tracking + tracing
│   │   ├── enhanced_ghl_client.py    # CRM integration (rate-limited)
│   │   └── jorge/                    # Bot services (handoff, A/B, metrics)
│   ├── models/                   # SQLAlchemy models, Pydantic schemas
│   └── streamlit_demo/           # Dashboard UI components
├── advanced_rag_system/          # RAG pipeline (BM25, dense search, ChromaDB)
├── benchmarks/                   # Synthetic performance benchmarks
├── docs/                         # Documentation
│   ├── adr/                      # Architecture Decision Records
│   └── templates/                # Reusable templates for other repos
├── tests/                        # 4,500+ automated tests
├── app.py                        # FastAPI entry point
├── admin_dashboard.py            # Streamlit BI dashboard
└── docker-compose.yml            # Container orchestration

Jorge Bot Audit (February 2026)

Production-ready bot services with enhanced monitoring and A/B testing:

Service	Status	Features
JorgeHandoffService	✅ Production	Circular prevention, rate limiting, pattern learning
ABTestingService	✅ Production	Deterministic assignment, z-test significance
PerformanceTracker	✅ Production	P50/P95/P99 latency, SLA compliance
AlertingService	✅ Production	7 default rules, email/Slack/webhook
BotMetricsCollector	✅ Production	Per-bot stats, cache hits, alerting

Quick Links

• Deployment Checklist
• Integration Guide
• Alert Channels

Deployment & Monitoring

Production-ready infrastructure with observability built in:

┌──────────────────────────────────────────────────────────┐
│  Docker Compose Profiles                                  │
│  ├── postgres (primary DB + Alembic migrations)           │
│  ├── redis (L2 cache, sessions, rate limiting)            │
│  ├── api (FastAPI, 91+ routes)                            │
│  ├── bots (Lead :8001, Seller :8002, Buyer :8003)         │
│  └── dashboard (Streamlit BI :8501)                       │
└──────────────────────────────────────────────────────────┘

Capability	Implementation	Key Metric
Token Cost Optimization	3-tier cache (L1 memory, L2 Redis, L3 PostgreSQL) + model routing	93K → 7.8K tokens/workflow (89% reduction)
Latency Monitoring	PerformanceTracker — P50/P95/P99 percentiles, SLA compliance	Lead Bot P95 < 2,000ms
Alerting	AlertingService — 7 default rules, configurable cooldowns	Error rate, latency, cache, handoff, tokens
Per-Bot Metrics	BotMetricsCollector — throughput, cache hits, error categorization	87% cache hit rate
Health Checks	/health/aggregate endpoint checks all services	Bot + DB + Redis + CRM status

Architecture Decisions

ADR	Title	Status
ADR-0001	Three-Tier Redis Caching Strategy	Accepted
ADR-0002	Multi-CRM Protocol Pattern	Accepted
ADR-0003	Jorge Handoff Architecture	Accepted
ADR-0004	Agent Mesh Coordinator	Accepted
ADR-0005	Pydantic V2 Migration	Accepted

Benchmarks

Synthetic benchmarks measuring platform overhead (no external API keys required).

python -m benchmarks.run_all

See BENCHMARKS.md for full methodology and results.

Observability

Full LLM observability stack: cost tracking, latency histograms, conversation analytics, and alerting.

See docs/OBSERVABILITY.md for details.

Testing

python -m pytest tests/ -v
python -m pytest --cov=ghl_real_estate_ai --cov-report=term-missing

Changelog

See CHANGELOG.md for release history.

Related Projects

• jorge_real_estate_bots -- Standalone 3-bot lead qualification system extracted from this platform
• ai-orchestrator -- AgentForge: unified async LLM interface (Claude, Gemini, OpenAI, Perplexity)
• Revenue-Sprint -- AI-powered freelance pipeline: job scanning, proposal generation, prompt injection testing
• insight-engine -- Upload CSV/Excel, get instant dashboards, predictive models, and reports
• docqa-engine -- RAG document Q&A with hybrid retrieval and prompt engineering lab
• scrape-and-serve -- Web scraping, price monitoring, Excel-to-web apps, and SEO tools
• Portfolio -- Project showcase and services

License

MIT -- see LICENSE for details.