Agentic Index
Agentic AI Platform Change Log
Agentic Index tracks material changes across agentic AI platforms, including new agent capabilities, workflow orchestration updates, integrations, MCP/tool-calling changes, memory features, human approval controls, observability, deployment options, pricing, funding, and verification-status updates. This change log is the index that tracks changes to agentic AI platforms, updated weekly. If you are trying to keep up with changes to AI agent platforms without reading every vendor's release notes separately, this is the feed to read.
Public-source research. All entries are sourced from official vendor blogs, changelogs, documentation, or press materials. Support levels marked Needs Review or Publicly Claimed should be independently verified before relying on them for procurement decisions.
22 entries matching filters
Integrations
GitHub Copilot has added general availability support for Anthropic's Claude Fable 5.1 model across multiple IDEs, the Copilot CLI, and the Copilot app. The model is optimized for long-horizon, autonomous coding tasks and complex agentic workflows.
Expands the choice of foundation models for enterprise and pro users, providing an option for deep codebase research and autonomous feature development. This model requires data retention by default for safety classifiers, which may impact compliance requirements unless zero data retention is explicitly negotiated.
Agent capability
GitHub Copilot has added support for xAI's Grok 4.6 reasoning model. The model is specifically optimized for agentic coding and executing complex, multi-step workflows.
Buyers gain access to another top-tier reasoning model, allowing teams to match the most effective AI model to their specific autonomous coding requirements.
Agent capability
GitHub Copilot has integrated Google's Gemini 3.7 Flash model across its IDE extensions, CLI, and cloud agent. The model offers enhanced performance for web development, agentic coding workflows, and codebase research.
Development teams can now select Google's latest Flash model for complex coding tasks, providing more flexibility and vendor neutrality. It is available under usage-based billing for enterprise and pro users.
Integrations
GitHub launched Agent Plugins 1.0 for VS Code, the Copilot CLI, and the Copilot desktop app. The release introduces portable plugins that allow developers to extend Copilot's capabilities consistently across multiple environments.
Organizations can build custom tools and integrations once and deploy them across all developer surfaces, ensuring a unified and efficient developer experience.
Agent capability
Microsoft's MAI-Code-1.1-Flash model is now available in GitHub Copilot, introducing native vision support for image understanding. The model improves coding quality and tool use while operating at a 73% lower list price than its predecessor.
Organizations can significantly reduce costs for high-volume, lightweight agentic workflows without sacrificing capability. The addition of vision support also unlocks new multimodal use cases for developers.
MCP / tool calling / API
GitHub released the Copilot SDK for Java, enabling server-side Java applications to create agent sessions, register tools, and stream structured responses. The framework-agnostic SDK includes Bring Your Own Key (BYOK) support and integrates with Jakarta EE and Spring.
Enterprise Java teams can now natively build and integrate custom Copilot agents using idiomatic Java code. This lowers the barrier to extending Copilot's capabilities within existing enterprise Java environments.
Observability / auditability
The Copilot usage metrics API now reports activity per individual agent via a new optional array. This data is available in 1-day and 28-day reports for enterprises and organizations.
Gives engineering leaders and administrators granular visibility into how specific AI agents are being utilized across their teams, aiding in ROI measurement and adoption tracking.
Human approval / guardrails
Lite and Balanced effort levels for Copilot code review are now generally available, replacing the preview-era Low and Medium names. Administrators can now set an organization-wide default for these review depths.
Provides teams with standardized controls over how deeply the AI reviews code, allowing organizations to balance thoroughness with review speed across all repositories.
Memory / state
GitHub released a batch of updates for Copilot across desktop, CLI, and VS Code. Key additions include an experimental worktree command for isolated worktrees, a rewind command to restore conversations without Git, session management in the CLI, and multilingual on-device dictation in VS Code.
Improves developer productivity by providing more robust context management, easier undo capabilities, and enhanced accessibility features across multiple IDEs and interfaces.
Security / enterprise
GitHub is retiring the Copilot Billing Preview app. Users are now directed to review and manage their Copilot spend directly in their native billing settings, which include AI usage pages, budget caps, and raw usage data exports.
Consolidates billing and spend management into the native GitHub billing interface, simplifying how organizations track and control their AI expenditures.
Workflow orchestration
Users can now create Copilot cloud agent automations that trigger automatically when an issue comment or pull request comment is created. Common use cases include generating documentation or summarizing discussions based on comment activity.
Enhances workflow automation by allowing developers to kick off AI tasks directly from collaborative comments, reducing manual context switching.
Security / enterprise
Enterprise administrators can now customize managed settings by targeting specific enterprise teams with itemized configuration files. This replaces the previous requirement to use a single configuration for the entire enterprise.
Allows large organizations to scale governance and apply tailored security or operational configurations to different teams without relying on a rigid global approach.
Agent capability
Users can now set the reasoning level for supported models when delegating tasks to the GitHub Copilot cloud agent. This setting controls how much computational reasoning the model applies to a given prompt before returning an answer.
Gives developers finer control over the trade-off between response speed and depth of reasoning, allowing them to optimize performance for complex coding tasks.
Integrations
Anthropic's Claude Opus 5 is now available as a selectable model across GitHub Copilot surfaces, including VS Code, Visual Studio, Copilot CLI, the cloud agent, and mobile apps. The new model is designed for complex, long-running agentic coding tasks and multi-step execution. Administrators for Copilot Enterprise and Business plans must enable the model policy in settings.
Providing access to Anthropic's newest model allows engineering teams to leverage stronger reasoning capabilities for intricate regression verification and autonomous workflows. This expansion ensures developers have the flexibility to route demanding tasks to an engine optimized for extensive tool coordination, improving overall reliability.
Observability / auditability
GitHub released a new Copilot usage metrics impact dashboard for enterprise administrators and organization owners. This visual dashboard categorizes users by AI adoption phases, measuring not just seat activation but depth of feature utilization based on a 28-day rolling window.
Buyers and administrators can now track ROI and product maturity beyond basic active-user counts. By surfacing underutilized features and suggesting targeted enablement steps, organizations can drive deeper adoption and realize greater value across their engineering teams.
Security / enterprise
GitHub introduced new enterprise security and administration controls for the Copilot code review agent. The agent now runs behind a firewall by default to restrict network access, features independent runner configurations separate from the Copilot cloud agent, and supports custom setup steps via a YAML file.
Enterprise administrators gain granular control over network security and infrastructure isolation during AI-assisted code reviews. Operating behind a default firewall addresses strict compliance requirements for enterprise environments and limits potential data exfiltration.
Observability / auditability
GitHub opened a public preview of Copilot agent session streaming for Enterprise Cloud with managed users. Prompts, responses, and tool calls across all Copilot clients can stream to SIEM platforms like Microsoft Purview or be pulled via REST API.
As agents act more autonomously, security and compliance teams need granular visibility. Streaming session activity into existing SIEM tooling lets them audit agent behavior and satisfy governance requirements without new tooling.
Agent capability
Copilot Vision reached general availability across Free, Pro, Business, and Enterprise. Users can attach images and PDFs to chat prompts so Copilot reasons about visual context alongside code. It is enabled by default with no admin policy change required.
Developers can feed visual bugs, architecture diagrams, and UI mockups straight to the agent instead of translating them to text. Available on every tier by default, so teams get multimodal help immediately.
Browser/computer use
Browser tools for Copilot agents in VS Code are generally available and on by default. Agents can open pages, navigate, click, type, capture console errors, and screenshot inside the editor, with consent controls for private tabs and admin domain toggles.
Lets Copilot verify UI changes and debug frontend issues without leaving the editor. Domain allowlists and explicit consent keep the agentic access governable in secure environments.
Agent capability
Kimi K2.7 Code, an open-weight model from Moonshot AI, is now generally available in the Copilot model picker, hosted on Azure and billed by usage. It is disabled by default for Business and Enterprise, requiring admin enablement.
A cheaper open-weight option for teams optimizing inference spend. Because it adds a new provider, enterprise admins opt in and govern access under their own compliance policies.
Pricing / packaging
AI credit session limits are now settable in the Copilot CLI (v1.0.66+) and SDK (v1.0.5+), letting developers and admins cap what an agent can spend in a single execution session.
Prevents runaway autonomous sessions from generating surprise overages under usage-based billing. Engineering leaders get hard budget control and cost predictability when shipping agentic features.
Agent capability
Copilot for JetBrains IDEs added Claude as an agent provider (public preview), letting developers run the Claude Code CLI as the agent inside Copilot Chat. The same release made org- and enterprise-defined custom agents generally available, giving admins a governed set of published agents for their teams.
Provider flexibility lets teams route agent work to Claude from inside Copilot. Admin-published org agents give a path to standardize and govern agent behavior, though the interim bypass-permissions default is a caveat for regulated environments.
Current rankings
These changes feed the index. See where each category stands: the best vendors of 2026, ranked by verified capability coverage.
Capability coverage
Every change above is graded against the same 14 capability axes. Each axis has its own page showing how much of the index documents it, which categories concentrate it, and the leading platform in each.
The AI agent directory for business · The full feature taxonomy · Pricing research
Common questions
How do I keep up with changes to AI agent platforms?
Read one maintained change log instead of tracking each vendor's release notes separately. The Agentic Index change log records 981 verified updates across agentic AI platforms, updated weekly, covering new agent capabilities, MCP and tool calling changes, memory, human approval controls, observability, deployment options, pricing and funding. Every entry names the vendor, the date, the change type and an impact level, and links to that vendor's profile.
How often is the agentic AI change log updated?
Weekly. The most recent update was Sep 4, 2026, covering LivePerson, Serval, ServiceNow.
How do I track changes for one specific AI agent vendor?
Filter the change log by vendor, or open that vendor's profile page, where the same verified entries appear alongside its capability coverage across the 14 point Agentic Index taxonomy.
Where can I find a single feed of agentic AI platform updates?
This page is that feed. Agentic Index reviews vendor blogs, changelogs, documentation and press materials every week and files what materially changed, so keeping up with changes to AI agent platforms is one page rather than several hundred. Entries are filterable by vendor, change type, impact level and verification status, and each one carries the evidence basis it was graded from.
Last updated: Sep 4, 2026 · 981 tracked changes across 372 vendors