DIAL Gateway

DIAL functions as a unified gateway to language and embedding models from all major (Amazon Bedrock, Google Vertex AI, and Azure OpenAI) and alternative vendors, as well fine-tuned micro models, self-hosted models and models listed on HuggingFace or DeepSeek.

Compatibility with OpenAI enables swift creation and integration of new model adapters.

DIAL Gateway speaks all industry-standard protocols such as OpenAI Chat Completions and Responses, Anthropic Messages, MCP and A2A. Connect to coding harness your teams already use such as Claude Code, GitHub Copilot, or Codex. You can seamlessly use DIAL agents and tools in process automation systems like n8n, Power Automate or Apache Airflow.

The gateway reaches systems your data and work live in such as MS SharePoint, Google Drive, Confluence, GitHub, Snowflake or Salesforce.

DIAL speaks the Model Context Protocol in both directions: agents built on DIAL can talk to other MCP tools on the market and DIAL can be reached by other systems via MCP just as easily.

Deploy and use your MCP tools on DIAL or turn a plain REST API into a fully supported MCP server. Reaching the wider ecosystem takes configuration, not development work.

DIAL enables granular, attribute and role-based access control reaching down to individual apps, models, tools or files. Connect to Okta, Entra ID, Auth0, Google, Cognito or Keycloak and leverage the access control policy you already have. Issue, rotate and revoke API keys from one UI console.

All requests from users or agents are checked first to ensure nothing and no one gets more access than the task in front of them requires.

DIAL enables guardrails to control every request and response before they reach a model or a user. The control logic is owned by Interceptors and can cover PII obfuscation, data-leak prevention, prompt-injection defense, regulatory compliance, content filtering or any other rules you specify.

Call your own models, create custom guardrails with a dedicated SDK or connect to providers such as Google Model Armor and Microsoft Presidio. Responsible AI enforced once, at the platform or agent level, instead of every team building it themselves.

DIAL runs inside your cloud or on-premises. Data stays within your boundary, encrypted in transit and at rest and is never shared with third parties. Only approved connections get through and each environment has its own allowlist. Agent code runs in a dedicated sandbox, unable to reach other agents, your network, or data it was not granted access to.

Logs can be anonymized before storage, and you decide what is kept and for how long. Every DIAL release is gated on continuous vulnerability scans.

DIAL's load balancer allows to spread traffic across models, regions and cloud subscriptions, drawing on provisioned throughput first and pay-as-you-go only once that is spent, so you get full value from what you have already committed to.

When a model, provider or region goes down due to an outage, capacity limit or regulatory restriction, traffic fails over to tested and verified destinations. Disruptions stop turning into downtime.

DIAL serves as a unified governed layer for every dollar spent on AI. Set spending limits per user, team, project, business unit or API key, enforced in real time. Pay for actual use, not per-seat licenses sitting idle. Route each task to the model that fits it, not the priciest by default.

Attribute costs, charge back to the right team, prove ROI, reconcile against vendor bills.

DIAL captures and logs every request down to a single call. It all lands in storage you own, and ships via OpenTelemetry to the monitoring stack you already run.

DIAL Analytics turns logs into a clear picture of how AI is used, what it costs, who uses it or for what purpose — attributed to a user, team, project or department. You can run custom metrics on top of this data to get insight into any topic such as AI champions and usage patterns, performance of specific tools and models, cost attribution, or user sentiment.

The Gateway enforces the rules; Admin is where you set them. Manage and audit your entire AI platform (users, files, apps, models, agents, tools, MCP servers, access, cost and more) from a single user-friendly UI.

Evaluate models and agents against quality benchmarks before, during and after they go live. Access analytics and audit dashboards.