Governing Coding Agent Sprawl with Unity AI Gateway

Software development has entered a new era. The best engineering teams are now shifting development from human-driven to agent-driven. All organizations should be aggressively looking to deploy these new 10x engineers as broadly as possible but they are concerned about governing and monitoring adoption. For coding against to be successful, they need access to sensitive company data such as engineering tickets, design documents, and customer issues, and organizations fear the security risks and runaway costs of ungoverned and unmonitored adoption.

Today, we’re introducing the Coding Agent Support in Unity AI Gateway. Our goal is to deliver coding tool freedom for developers and unified governance for admins.

AI Coding Agent Sprawl

Opus 4.6, Composer 2, GPT-5.4, Kimi-2.5, Gemini 3 Pro -- new models are released every week that reshape the frontier of cost and quality. Coding tools themselves are also constantly evolving, and software developers want choice, For example, within Databricks, our software developers flexibly mix usage between Cursor, Codex, Claude Code, and others -- often using multiple tools at the same time! Adopting multiple coding tools is a business necessity, which then introduces key challenges for administrators trying to move fast to support developer productivity with multiple tools.

Security, data privacy, and cost reviews for new tools can slow teams down. In addition, these AI coding tools require much stronger due diligence for several reasons:

• Security Risk: MCPs can give agents access to sensitive data
  MCP tools are most useful when they have access to critical data within your organization, so it’s easy to accidentally make them the most privileged developer in your organization. How do organizations audit and govern agent’s access to data?
• Cost Explosion: Agent costs are exploding
  With AI usage growing, agent costs are becoming a top R&D cost driver. Access to AI and flexibility of tools needs to be balanced with reasonable cost guardrails. How can admins ensure effective cost controls across multiple tools?
• Visibility Gap: Executives lack visibility into tool adoption
  With the expanding capabilities of these tools, organizations are racing to adopt AI. To scale AI across an organization, measurement is critical to identify the key blockers. How can executives easily see who is using AI if everyone’s using a different tool?

So, how can organizations move fast to enable AI productivity while ensuring data privacy and cost visibility?

Introducing Coding Agent Support in Unity AI Gateway

To simplify this, we are introducing the coding agent support in Unity AI Gateway, a unified governance hub for popular coding tools like Codex, Cursor, and Gemini CLI. Our gateway unifies access controls, usage statistics, operational observability, cost management, guardrails, and inference capacity into a single platform, giving you centralized control over your AI agents.

• Pillar 1: Centralized Security and Audit: All agent data access can be centrally governed with all audit logs in Unity Catalog with MCP servers managed in Databricks and centralized tracing with MLflow.
• Pillar 2: Single Bill and Cost Limits: Admins can set cost limits that apply across whichever tools developers want to use. With capacity from Foundation Model API, which offers first-party inference for all popular models, admins get one all-in bill from Databricks.
• Pillar 3: Full Observability in the Data Lakehouse: Critical data like lines of code written per user, cost per month per user, and more are all automatically ingested into your Data Lakehouse, alongside the rest of your critical data.

Security and Compliance for Coding Agent, MCP and LLM Interactions

AI Gateway unifies security governance across coding agents, LLM interactions and MCP integrations. Your development workflows run on the same trusted platform as your analytics and AI, with centralized controls

• Data privacy: We ensure your private data stays within the Databricks security perimeter.
• Audit-ready logging: Automatically capture traces in Unity Catalog for compliance and security reviews.
• Single identity across all services: Developers authenticate once with Databricks credentials for all tools—GitHub, Atlassian, and others—with no separate logins per service. This securely connects agents to critical data sources while enforcing consistent governance.

By unifying all integrations, including coding agents and MCP tools, organizations can centralize logging, enforce policies, and monitor usage across the entire ecosystem, ensuring consistent security and compliance.

Simplify cost management and billing

Databricks’ Foundation Model API provides inference for OpenAI, Anthropic, and Gemini models, and the best open source coding models like Qwen in a single platform. The Gateway also lets you bring external capacity in, expanding governance to all your tokens, regardless of where they flow.

This means your coding tools can connect to the same capacity as your other agents, and costs are centralized into a single bill and observability platform! Foundation Model API offers day one launches for every frontier LLM model, so developers can use the newest, best models that are coming out immediately.

With our centralized Gateway, admins can stop switching tabs between admin consoles to control rate limits and budgets for every single coding tool. Instead, organizations can give developers a single budget across all coding tools to burn down on their agent of choice!

Unified Observability for AI Coding Tools

By treating AI coding tool usage data as a first-class citizen in Unity Catalog alongside your enterprise datasets, you gain a unified, governed framework for deep operational intelligence. This ensures all coding activity is auditable, secure, and ready for automated workflows.

With our OpenTelemetry ingestion, coding tool metrics and traces are automatically centralized to Unity Catalog-managed Delta tables.

With all the data landing in the Lakehouse, enterprises are finding creative ways to combine usage data with their business’ ontology.

• Track adoption per org: Join AI Gateway metrics with Workday to map GenAI adoption by department, region, or seniority, helping identify where to target enablement.
• Quantify Developer Velocity: Measure the tangible impact of AI assistance on output.

Example: "A 20% increase in token usage per developer drove a 15% reduction in pull request cycle time, directly linking AI tool usage to increased developer velocity."

• Proactive Capacity Planning: Monitor users hitting rate limits to data-justify securing additional capacity or dedicated throughput before productivity is throttled.

What Our Customers Are Saying

We have been working to get visibility into our coding tool usage across teams. We need a centralized way to monitor spending, manage token budgets, and catch anomalies before they become costly problems. We look forward to leveraging AI Gateway’s monitoring capabilities to give us the control and transparency we need to scale AI responsibly. — George Torres, Senior Director of AI Engineering, First American

As we broaden adoption of AI coding tools to support hundreds of developers across regions, we are counting on AI Gateway to provide native support for experimental features and advanced tooling including web search and large-context models. We need a unified platform that seamlessly enables beta capabilities and provides real-time usage dashboards, so we can scale AI development with confidence while maintaining rigorous governance and compliance across our healthcare analytics organization. — Iyibo Jack , Chief Product Officer, Milliman MedInsight

Get Started

Starting today, the AI Gateway for coding tools is available for all Databricks customers. Cursor, Gemini CLI and Codex CLI support are ready for immediate use.

Check out the documentation to get started.