Microsoft introduces Scope platform

Microsoft has introduced Scope, an open-source platform built on Kubernetes. It aims to help product teams transform interactions with AI coding agents into usable data for improving their software.

Scope focuses on three improvement areas

Scope helps teams identify improvements in three areas: documentation that guides agents effectively, correct discovery and adoption of software development kits (SDKs), and tools that help agents complete tasks successfully.

AI-generated code in Microsoft's repositories

In 2025, Microsoft CEO Satya Nadella mentioned that 20 to 30 percent of the code in the company's repositories was generated by AI. This figure was shared during a discussion with Meta CEO Mark Zuckerberg, who asked about the proportion of AI-generated code.

Kubernetes as a container orchestration system

Kubernetes, also known as K8s, is an open-source system that automates the deployment, scaling, and management of software. Originally developed by Google, it is now maintained by a global community and owned by the Cloud Native Computing Foundation.

Scope supports continuous DevAx learning

Available under the MIT license, Scope supports a continuous DevAx learning loop. It allows teams to use reproducible experiences to evaluate and improve their products based on agent interactions.

AI coding agents increase code production

A recent report showed that coding agents have significantly increased raw code production. However, this does not translate into increased software production due to delays in the human review stage.

Human review stage becomes a bottleneck

Review delays have increased by 49 percent, and review requests have nearly doubled. Despite the use of AI review tools, human oversight remains essential, and these tools have not replaced the need for human input.

DevAx addresses agent interaction challenges

For product teams, the challenge is ensuring that an experience that works well for humans also works for AI agents. The Developer Agent Experience (DevAx) extends traditional developer experience (DevEx) practices to include how agents interact with software.

Azure Cosmos DB team uses Scope

The Azure Cosmos DB team used Scope to compare basic models with its Agent Kit in end-to-end scenarios. This helped identify and resolve 20 problems with the Agent Kit and 27 with documentation, improving agent behavior significantly.

Scope helps improve documentation and tools

Scope helps teams identify issues with documentation, SDK adoption, and tools that assist agents. Teams can update instructions, evaluate revised examples, and test whether changes lead to better agent performance.

Open source foundation for agent evaluation

Scope is built on an open-source foundation to allow broader participation in evaluating AI agents. This approach helps integrate diverse scenarios, failure cases, and perspectives into the evaluation process.

MIT license and self-hosted options

The MIT license allows teams to use and extend Scope freely. Self-hosted deployments enable teams to evaluate private workloads without sharing their code or execution data.

Scope supports multiple coding agents

Scope pilots existing coding agents like GitHub Copilot, Claude Code, and Visual Studio Code with Copilot. Teams can define tasks, provide starting code, and choose agent configurations to simulate different user perspectives.

Customizable and reusable profiles

Customizable profiles in Scope allow teams to vary the simulated user's perspective and communication style. These profiles can be reused and versioned to record the configuration of each execution, ensuring consistency and traceability.