AI Automation Templates — Enterprise Portfolio

12 production-ready AI agents built with LangGraph and Claude API. Standalone, documented, and deployable in under an hour.

---

Available for Missions

AI Automation freelance consultant specializing in the design and deployment of custom AI agents for enterprise clients (PME & ETI).

Available on Malt and Upwork for AI agent development missions, business process automation and Claude API integrations.

🌐 Website → stephane-bomin.fr

📅 Book a free 30-min discovery call → Book here

Stack: LangGraph · CrewAI · Claude API · Supabase · Streamlit · MCP Protocol

Demo Videos

| Agent | Demo | |-------|------| | 🧬 Enterprise RAG Agent | ▶ Watch demo | | 🎯 Prospection Commerciale B2B | ▶ Watch demo | | 📊 Agent Finance Close | ▶ Watch demo | | 🔔 Agent Monitoring & Alertes | ▶ Watch demo |

Why This Collection

These templates target the 3 levels that define enterprise-grade AI automation:

• Vertical AI Agents — encoded business logic, regulated domains, high technical barrier
• Multi-Agent Architectures — LangGraph orchestration, shared state, conditional routing
• Enterprise Governance — timestamped audit trails, permissions, regulatory compliance

What SaaS and no-code tools don't do: deep SI integration, audit trails for compliance, custom business logic, sensitive data governance.

---

Templates

💰 Sales & Revenue

| # | Name | Use Case | |---|------|----------| | 04 | SDR / Revenue Agent | Enrichment → ICP Scoring → 3-email sequence → Gmail send | | 01 | B2B Workflow Agent | Email → CRM lookup → Classification → Ticket → Reply → Gmail |

📚 Knowledge & Documents

| # | Name | Use Case | |---|------|----------| | 06 | Enterprise RAG Agent | Private knowledge base with governance, permissions, anti-hallucination, audit trail | | 07 | Advanced Document Analysis Agent | Structured extraction → Risk verification → Synthesis → Risk matrix → PDF | | 10 | Contractual Pipeline Agent | Clause extraction → Risk analysis → Synthesis → Improved contract generation |

📊 Finance & Risk

| # | Name | Use Case | |---|------|----------| | 03 | Finance Close Agent | Multi-agent financial close — reconciliation, variances, journal entries, disclosure | | 08 | M&A Due Diligence Agent | Multi-axis analysis → Scoring → Risk matrix → Verdict → Valuation range | | 05 | Autonomous Data Analyst Agent | Natural language → SQL → Execution → Insights → Executive summary |

⚙️ Governance & Ops

| # | Name | Use Case | |---|------|----------| | 02 | EU AI Act Compliance Agent | AI compliance audit — classification, gaps, remediation plan, audit trail | | 11 | LLM Quality Evaluation Agent | Tests → 7-dimension scoring → Regression detection → Deployability badge → Report | | 09 | Monitoring & Alerts Agent | Violation detection → Causal analysis → Graduated alerts → Gmail → Supabase | | 12 | SI Integration & Webhook Agent | Reception → Validation → Mapping → Retry → Dead Letter → Multi-destination |

---

Packs

| Pack | Agents | Business Value | |------|--------|----------------| | 💰 Revenue Engine | 04 + 01 + 09 | Full automated commercial pipeline | | 📚 Knowledge Hub | 06 + 07 + 10 | Documents turned into intelligent assets | | 📊 Finance & Risk | 03 + 08 + 05 | Automated critical financial processes | | ⚙️ AI Governance & Ops | 02 + 11 + 09 + 12 | Compliant, connected, monitored AI agents |

---

Typical Client Pipelines

Full B2B commercial pipeline

04 — SDR Revenue → 01 — B2B Workflow → 10 — Contractual Pipeline

Prospect enrichment → lead qualification → email handling → contract generation

---

AI compliance pipeline (August 2026 deadline)

02 — EU AI Act Compliance → 11 — LLM Quality Evaluation

Audit AI systems → validate quality before deployment

---

M&A / investment pipeline

07 — Document Analysis → 08 — Due Diligence → 10 — Contractual Pipeline

Analyze documents → due diligence → generate agreements

---

Enterprise finance pipeline

05 — Data Analyst → 03 — Finance Close → 09 — Monitoring

Analyze data → close books → monitor continuously

---

SI & data pipeline

12 — SI Integration → 06 — RAG Enterprise → 05 — Data Analyst

Ingest events → enrich knowledge base → analyze

---

Tech Stack

• Orchestration : LangGraph (multi-agents, conditional routers, loops)
• LLM : Anthropic Claude Haiku + Sonnet depending on the node
• Interface : Streamlit
• Database : Supabase (PostgreSQL)
• Email : Gmail API (OAuth2)
• Web search : Serper
• Embeddings : Sentence Transformers (multilingual)
• Data : Pandas, Plotly
• PDF : FPDF2, PyMuPDF
• HTTP : Requests (webhooks, third-party APIs)

---

LLM Architecture

Each template uses two models depending on the node type:

• claude-haiku-4-5-20251001 — extraction, classification, validation, JSON, structured analysis
• claude-sonnet-4-6 — long-form generation, reports, recommendations, synthesis

---

Switching LLM Provider

Default: Anthropic Claude (GDPR compliant, EU servers).

| Provider | Model | Cost vs Claude | Notes | |----------|-------|----------------|-------| | Anthropic (default) | claude-haiku-4-5 | reference | GDPR, EU servers | | DeepSeek | deepseek-chat | ~10x cheaper | China servers | | Mistral | mistral-small | ~3x cheaper | EU servers, GDPR | | OpenAI | gpt-4o-mini | ~2x cheaper | US servers |

Migration in 3 steps:

pip install openai

In config.py:

LLM_BASE_URL = "https://api.deepseek.com"
LLM_MODEL = "deepseek-chat"
LLM_API_KEY = "sk-..."

In graph.py:

from openai import OpenAI
client = OpenAI(api_key=LLM_API_KEY, base_url=LLM_BASE_URL)

---

API Cost Estimates

Monthly estimates with Anthropic Claude Haiku + Sonnet (default provider):

| Template | Light usage | Medium usage | Heavy usage | |----------|-------------|--------------|-------------| | 01 — B2B Workflow | Low | Medium | High | | 02 — EU AI Act | Low | Medium | High | | 03 — Finance Close | Low | Medium | High | | 04 — SDR Revenue | Low | Medium | High | | 05 — Data Analyst | Very low | Low | Medium | | 06 — RAG Enterprise | Low | Medium | High | | 07 — Document Analysis | Low | Medium | High | | 08 — Due Diligence | Low | Medium | High | | 09 — Monitoring | Very low | Low | Medium | | 10 — Contractual Pipeline | Low | Medium | High | | 11 — LLM Evaluation | Very low | Low | Medium | | 12 — SI Integration | Very low | Low | Medium |

Light: a few uses/day. Medium: regular team usage. Heavy: high-volume production.

With DeepSeek: divide by 8 to 10. With Mistral Small: divide by 3.

---

Cloud Deployment

Streamlit Cloud (free)

# Connect on share.streamlit.io
# Secrets: ANTHROPIC_API_KEY, SUPABASE_URL, SUPABASE_KEY

Railway

# Procfile: web: streamlit run app.py --server.port $PORT --server.address 0.0.0.0
# Remove pywin32 from requirements.txt before Linux deployment

Docker

FROM python:3.11-slim
WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt
COPY . .
EXPOSE 8501
CMD ["streamlit", "run", "app.py", "--server.address", "0.0.0.0"]

---

Installation

cd XX-template-name
pip install -r requirements.txt
streamlit run app.py

Each template includes a .env file to configure and a README.md with complete instructions and test data.

---

Requirements

• Python 3.10+
• Anthropic API key
• Supabase account (templates with persistence)
• Gmail account + OAuth2 credentials (email templates)
• Serper API key (SDR template with web enrichment)

---

Actively maintained collection.