Platform Overview
Relevant source files
The following files were used as context for generating this wiki page:
The AI Network Operations Platform is a self-hosted assistant designed for enterprise network engineering teams. It acts as an AI Network Engineer capable of handling complex tasks like infrastructure discovery, multi-vendor troubleshooting, configuration management, and DDI (DNS, DHCP, IPAM) operations through natural language interaction.
The platform is built on a "local-first" and "secure-by-default" philosophy. It ensures that every action taken by the AI is visible, audited, and explained. Most importantly, it enforces a human-in-the-loop requirement: every network change must be approved by a human engineer before execution.
Sources: README.md:3-8, CLAUDE.md:3-13, CLAUDE.md:23-31
Key Capabilities and Specialized Teams
The platform functions by routing your requests to specialized AI "agents." You can think of these agents as specialized team members who handle specific parts of the network.
- Master Architect: The supervisor who listens to your request and decides which specialist is best suited to handle it.
- Discovery Specialist: Handles inventory enumeration, scanning subnets, and identifying LLDP/CDP neighbors.
- Troubleshooting Specialist: Diagnoses why a route is missing, why a BGP session is down, or why traffic is being dropped by an ACL.
- DDI Specialist: Manages DNS zones, traces resolution paths, and drafts changes for DHCP scopes or DNS records.
- Security Specialist: Audits firewall policies for shadowed or redundant rules and assesses security posture.
- Documentation Specialist: Generates network inventories, topology diagrams, and operational runbooks.
- Consultant: If your request is unclear, this agent asks a clarifying question instead of guessing.
Sources: CLAUDE.md:19-31, backend/app/llm/prompts/init.py:175-235, backend/app/agents/consultant/agent.py:72-76
Supported Vendors and Protocols
The platform uses a standardized "plugin" system. This means it talks to different hardware using a consistent connector, normalizing data from various vendors into a single format that the AI can understand.
| Vendor Family | Supported Systems/Services |
|---|---|
| Cisco | IOS, IOS-XE, NX-OS |
| Juniper | JunOS |
| Arista | EOS |
| Firewalls | Palo Alto (PAN-OS), Fortinet (FortiOS) |
| Load Balancers | F5 BIG-IP |
| DDI Platforms | BlueCat, Infoblox |
| Cloud/Virtual | AWS, Azure, VMware |
Sources: CLAUDE.md:14-27, README.md:27-31
System Architecture
The platform consists of seven core components that work together to provide a seamless interface for network operations.
- Frontend (Operations Console): The web interface where you chat with the AI, view the topology map, check inventory, and approve changes.
- API (Central Brain): The main entry point that manages user sessions, authenticates engineers, and coordinates with the AI agents.
- Worker (Background Engine): A background process that handles long-running tasks like running discovery scans, taking configuration backups, or analyzing packet captures.
- Postgres (System of Record): A traditional database that stores everything from device inventories and audit logs to AI reasoning traces.
- Neo4j (Living Topology Map): A graph database that tracks how every device, interface, subnet, and application connects to everything else.
- Redis (Task Coordinator): A high-speed memory store used to pass messages between the API and the background workers.
- Ollama (Local AI Engine): An optional component that allows the platform to run AI models locally on your own hardware for maximum privacy.
Sources: README.md:12-25, deploy/docker/README.md:8-16, deploy/docker/docker-compose.yml:25-144
Deployment and Workflow
Deployment Options
The platform is designed to be flexible, supporting both lab environments and large-scale production deployments.
- Docker Compose: Ideal for labs or small teams. It runs all components on a single server using a single configuration file.
- Kubernetes (Helm): The production standard. It provides high availability, security hardening, and scales across multiple nodes. It includes advanced features like automated backups and network segmentation.
Sources: README.md:38-48, deploy/docker/README.md:3-6, deploy/kubernetes/netops/README.md:3-12
Change Approval Workflow
Safety is the platform's primary constraint. No agent can modify the network directly. Instead, they follow a strict lifecycle for changes:
- Draft: An agent (like DDI or Configuration) creates a proposed change based on your request.
- Pending Approval: The change appears in the console for a human engineer to review.
- Pre-change Backup: Before any execution, the platform automatically takes a configuration backup of the affected devices.
- Approved/Executing: Once a human approves, the Automation Agent executes the change.
- Completed or Rolled Back: If the change fails, the platform can use the pre-change backup to restore the original state.
Sources: backend/app/agents/ddi/agent.py:16-25, backend/app/agents/security/agent.py:18-24, README.md:4-8
Platform Diagrams
System Interactions
The following diagram shows how users interact with the platform and how the platform connects to your infrastructure and external services.
This diagram illustrates the external connections the platform maintains for management and operations.
Sources: CLAUDE.md:14-27, README.md:27-31, deploy/kubernetes/netops/README.md:22-26
Agent Routing and Specialist Flow
When you ask a question, the Master Architect determines the intent and routes it to the appropriate specialist team member.
This diagram shows how requests are triaged and the distinction between read-only actions and state-changing requests.
Sources: CLAUDE.md:29-31, backend/app/llm/prompts/init.py:175-235, backend/app/agents/ddi/agent.py:16-25
Topology Data Model
The platform maintains a comprehensive map of your network, tracking relationships between different objects.
This diagram shows what the platform knows about your network and how these components relate to one another.
Sources: CLAUDE.md:43-49, backend/app/agents/documentation/agent.py:20-22
Deployment Comparison
Comparing the lab-ready Docker Compose setup against the production-hardened Kubernetes deployment.
This visual comparison highlights the difference between a simple single-server lab and a resilient production environment.
Sources: deploy/docker/README.md:3-6, deploy/kubernetes/netops/README.md:14-41
Summary
The AI Network Operations Platform provides a unified, AI-driven interface for managing multi-vendor enterprise networks. By centralizing discovery, troubleshooting, and changes into a single platform governed by human approval, it reduces operational complexity while maintaining high security and audit standards. Its modular architecture allows it to scale from a simple lab server to a hardened, multi-node Kubernetes cluster.
Sources: README.md:3-8, CLAUDE.md:3-13