Get Started with DIAL

DIAL is an open-source, enterprise-grade platform that gives you a single gateway to language models, a place to build and run AI applications, and the access control, observability, and collaboration features needed to operate them in production.

New to the platform? Read What is DIAL for the big picture, then pick a path below. Use the documentation directory to browse every section.

Just want to see DIAL?

Open the hosted edition at chat.dialx.ai — no install, real models, works on any machine.

Want DIAL on your own machine?

ai-dial-samples to run the DIAL platform locally with Docker.

Need the documentation source for your agent?

The original source of this documentation lives in ai-dial. If your agent needs the source files, pull them from there.

Need the API reference?

The DIAL API specification describes every DIAL Core endpoint: chat completions, embeddings, files, conversations, prompts, sharing, publications, Toolsets, and administration.

Choose your path

I want to…	Start here	Continue with
Understand what DIAL is, what it can do, and how it is built	What is DIAL	Capabilities and architecture
Connect a coding agent or harness (Claude Code, Codex, GitHub Copilot, and others)	Agentic tools	Connect Claude Code, connect Codex, or see how DIAL governs agents
Understand security and governance in DIAL	Security and governance	Authentication and access control and usage limits and cost control
Work with the DIAL Chat interface	Chat User Guide	Start a conversation and find models and agents
Administer the platform with DIAL Admin	Admin Panel overview	Manage the AI gateway and users and access
Deploy DIAL for my organization	Cloud deployment	Configure authentication and prepare for production
Run DIAL locally	ai-dial-samples	Local setup to build your own Docker Compose setup
Explore the services that make up DIAL	Services	Architecture highlights and DIAL Core
Build an application or agent	Choose an app type	Build a Quick App or build with the DIAL SDK
Connect enterprise tools and documents	Tools and enterprise data	Connect an MCP server or build a document Q&A app

Documentation directory

Understand DIAL

Understand DIAL explains the platform before you choose what to build or deploy.

• Foundations — concept map, glossary, and platform history
• Capabilities — model access, agents, tools, guardrails, evaluations, and collaboration
• Architecture — services, request flow, deployment profiles, data lifecycle, and limitations
• Security and governance — identity, permissions, usage limits, and data boundaries

Building with DIAL

Building with DIAL collects developer tutorials, integration guides, and SDK references.

• Apps — choose among Quick Apps, Code Apps, and Custom Apps
• Toolsets — connect MCP servers and expose REST APIs as tools
• Interceptors — build and assign request and response middleware
• Adapters — connect model providers or implement an adapter
• Developer tools: SDK reference, Chat customization, and UI libraries
• Integrations — chatbots, productivity add-ins, workflow automation, and agentic tools
• Working with DIAL resources — files, sharing, publications, notifications, and delegated access
• Advanced topics — prompt caching and performance optimization

Operating DIAL

Operating DIAL covers how to install DIAL, connect models, and keep the platform running.

• Local setup — run Docker Compose with an application, Ollama, vLLM, Azure OpenAI, or a remote DIAL Core
• Cloud deployment — install on AWS, Azure, GCP, or another Kubernetes cluster
• Model deployment: Amazon Bedrock, Azure OpenAI, Vertex AI, and Databricks
• Configuration — configuration files, environment variables, and Docker Compose settings
• Auth and access control — API keys, JWT, roles, rate limits, and SSO providers
• Observability — tracing, metrics, logging, alerting, and analytics setup
• Production readiness — availability, scaling, secrets, backup, upgrades, cost, and hardening
• Troubleshooting — diagnose deployment and runtime problems

Administering DIAL

Administering DIAL covers day-to-day work in DIAL Admin.

• AI Gateway — models, applications, Toolsets, routes, translators, and application runners
• Artifacts — published skills, files, conversations, and prompts
• Deployments — container images, workloads, and model servings
• Evaluation — datasets, test suites, and runs
• Analytics — dashboards, sessions, queries, tables, and pipelines
• Governance: users and access, roles, API keys, folder permissions, interceptors, approvals, usage limits, and compliance FAQ
• Configuration backup and global properties — import, export, and global interceptors

Services

Services documents each service's role, dependencies, deployment, and configuration.

• Gateway and interfaces: DIAL Core, DIAL Chat, Chat Overlay, DIAL Chat Themes, and DIAL Admin
• Model adapters: DIAL Model Provider, which runs the OpenAI, Bedrock, and Vertex AI adapters in one service
• Applications and tools: Quick Apps, OpenAPI to MCP, DIAL RAG, Generic RAG, and Deep Research
• Code build and execution: App Controller, App Builder Python, and Code Interpreter
• Deployment and quality: Deployment Manager, DIAL Helm, and DIAL Evaluations
• Data and automation: DIAL Analytics, Analytics Enrichment Runner, DIAL Log Parser, and DIAL Scheduler

Chat User Guide

The Chat User Guide explains the interface and everyday workflows.

• Conversations and scheduled tasks — chat with models and agents, attach documents, and schedule recurring work
• Catalog — find models and agents, and create agents, Toolsets, skills, and prompts
• File Manager — organize and upload files
• Sharing and publishing — collaborate with colleagues and request publication
• Usage and settings — review consumption and personalize Chat

Use cases and demos

• Use cases — scenarios covering assistants, knowledge bases, no-code tools, model access, governance, production agents, integrations, quality, and compliance
• Demos — videos grouped by platform overview, Chat, apps and agents, deployment and administration, developer features, and integrations