June 2026 Summaries

Merge offers an AI gateway solution, known as Merge Gateway, that acts as a control layer between applications and AI models, facilitating reliable AI operations in production by routing requests and centralizing governance. This solution allows organizations to manage, add, or swap AI models without altering downstream integrations, thanks to a single API endpoint. Key components include model access from multiple providers, flexible routing logic based on weighted benchmarks, cost governance with budget alerts, and detailed observability through comprehensive logging of API requests. The Merge Gateway provides features such as automatic failover during provider outages, customizable routing using weighted benchmarks, and robust cost tracking, allowing organizations to maintain control over AI usage and expenses. Users can avoid building their own AI gateways by utilizing popular third-party solutions like Portkey and OpenRouter, each offering different levels of comprehensive observability, routing capabilities, and cost management.

Merge's Cookie Policy outlines the use of cookies to enhance user experience on their site, requiring acceptance of data collection as per their Privacy Policy. The document also delves into the importance of AI audit logs, which provide detailed records of tool calls made across MCP connectors, offering visibility crucial for detecting security breaches, compliance violations, and potential data access issues. Implementing AI audit logs aids in compliance with security frameworks like SOC 2 Type 2 and GDPR, speeds up incident response, and helps detect data loss prevention violations. Best practices include making logs searchable, defining retention and access policies, and establishing escalation paths to ensure logs drive actionable responses. The Merge Agent Handler for Employees (AHFE) simplifies the management of AI audit logs, offering features like automated provisioning, central policy enforcement, and least-privilege access controls, allowing organizations to secure their internal AI usage effectively.

Merge utilizes its product, Agent Handler for Employees (AHFE), to securely manage and monitor how its employees, referred to as "Mergies," access and use approved AI tools and connectors within the company’s systems. By employing AHFE, Merge ensures that employees can efficiently use AI while maintaining security and visibility over tool usage, with access granted through an Okta group to a pre-configured bundle called the "Jarvis Tool Pack." The implementation involves detailed authentication processes and allows the security team to track and manage tool access, modifying permissions as necessary to align with approved workflows. The company emphasizes best practices like starting with a minimal scope of connectors, avoiding tools with destructive capabilities, and using shared scopes for critical systems to reduce security risks and enhance AI reliability. Merge plans to further refine this system by creating tailored tool packs for specific teams to minimize data exposure and improve functionality.

Merge's platform facilitates seamless integration with Box through its Merge Agent Handler, simplifying the process of connecting Claude Code to Box's API for developers seeking to automate workflows. By using the Merge CLI, developers can bypass the complexities of OAuth 2.0 authorization and token management, allowing them to focus on tasks such as searching files, managing folders, and retrieving shared content directly from the terminal. The Merge Agent Handler provides an abstracted authentication layer, tool-level scoping, and comprehensive audit logging, which enhances security and efficiency for teams working with Box. This integration is particularly useful for developers who need to access Box content for downstream processes, ensuring that tasks like retrieving contracts or organizing project folders can be automated without leaving the terminal. Additionally, Merge offers tools to manage employee access to AI systems, using identity providers for provisioning and enforcing data loss prevention policies, thereby maintaining centralized control over document access.

Merge, an integration platform, emphasizes the importance of AI model governance as governments, like the US, begin to regulate access to specific models such as Anthropic's Fable 5 and Mythos 5 for foreign nationals. To address the growing need for model governance in enterprises, Merge has developed the Agent Handler, which allows IT departments to define model access rules based on employee attributes synced from identity providers and HR systems. These attributes include citizenship, group membership, role, and work location, enabling organizations to enforce role-based access to AI models securely. As model-specific regulations are expected to become more common, Merge provides the infrastructure needed for scalable AI model governance, allowing employees to see which models they have access to and request additional access if needed.

Merge's platform facilitates the integration of Cloudflare with Codex through the Merge Agent Handler, enabling seamless automation and management of edge-infrastructure tasks. By connecting Codex directly to Cloudflare, Merge allows for accurate access to real-time zone configurations, DNS records, and KV namespaces, ensuring that the generated code matches the actual infrastructure rather than relying on potentially inaccurate descriptions. The platform simplifies the process by handling API token storage, rotation, and access control, providing a secure and auditable environment. This integration is particularly beneficial for engineering teams as it supports provisioning, access control, policy enforcement, and detailed audit logging, all of which help manage and secure how employees interact with AI tools like Codex. Merge's approach eliminates the complexity of self-hosted solutions and centralizes control, making it easier for organizations to manage integrations and maintain operational integrity.

Merge employs a unique design and engineering process that eliminates traditional phases like prototyping by leveraging AI tools such as Claude Code to swiftly transition from concept to production code. This approach enables rapid product development by allowing engineers to build backends and use AI to create initial front-end drafts, which designers then refine for user-friendliness. This integrated workflow facilitates faster iteration and avoids the typical bottlenecks associated with handoffs between design and engineering teams. The process encourages designers to work directly in the code environment, enhancing both the design quality and engineering velocity by eliminating the gap between design concepts and the final shipped product. This method, which empowers designers with commit access and bypasses traditional review gates like Figma, relies on trust in the designer's judgment and can significantly improve the overall efficiency and quality of product development.

Agent Handler for Employees is a governance tool designed to manage and secure AI interactions within organizations, allowing employees to connect to AI tools like Claude, ChatGPT, and Cursor across various systems while maintaining central control. It addresses the challenge of balancing employee productivity with IT security by providing authenticated, scoped, and auditable access to AI tools. The solution integrates with identity providers like Okta and Azure AD to provision users and assign tool access based on roles, enabling consistent permissions across AI interfaces. Every AI interaction is logged, detailing the employee's identity, the action taken, and the outcome, ensuring transparency and compliance. Additionally, Agent Handler allows IT departments to define access policies, configure data loss prevention (DLP) rules, and manage permissions from a central dashboard, thereby ensuring that AI tool calls are scanned for sensitive data before reaching third-party systems.