Skip to main content

February

2/10/2026

Global Datastores Now Available
Pre-built, read-only datastores are now available to all users so you can test agents, run demos, or explore the platform without uploading your own documents.

Contextual AI now provides global datastores—curated knowledge bases that are available to every account by default. Use them to try out the platform, demo agents to stakeholders, or prototype workflows before connecting your own data. Available collections include Materials Science Research Papers, Star Wars Wookiepedia, Rocket Science, RFC Library, 3GPP Spec, and Raspberry Pi Docs.

How to use
In the platform, go to Datastores in the left navigation. Global datastores are marked with a Demo badge. When creating an agent, select any global datastore from the dropdown. In Agent Composer YAML, reference a global datastore by its ID.

Global Datastores documentation

2/2/2026

Web Search Attributions in Responses
Agents can now surface attributions in the final response using information from web search, and those attributions can be rendered directly in your application.

When an agent uses web search, the platform can now generate attributions in the response—clear references to the sources that informed the answer. These attributions are available in the API and can be rendered inline in your application so users can see where information came from and follow links to the original content.

How to enable
No action required. This feature is enabled by default and is Generally Available for all new production tenants.

Note
Some source pages do not allow their content to be displayed inline. In those cases, users can use the Source documents button to view the content.


January

1/27/2026

Agent Composer Now Available in Public Preview
Contextual AI’s Agent Composer is now available in public preview, enabling users to build customized agent workflows with visual and programmatic tools.

We’re excited to announce that Agent Composer is now available in public preview. Agent Composer is Contextual AI’s framework for building customized query workflows by assembling pre-made components such as search tools, third-party API/MCP integrations, and various LLMs.

Self-serve users can start with pre-built templates (Basic Search and Agentic Search) that cover most Q&A and multi-step retrieval use cases. Enterprise users can access full Agent Composer capabilities, including:

Key Features

  • Pre-built templates — Start with Basic Search or Agentic Search (self-serve) or use specialized templates for Root Cause Analysis, Deep Research, Task Execution, and Structured Extraction (enterprise)
  • Visual workflow builder — Drag-and-drop interface for customizing agent behavior, adding tools, and defining multi-step workflows
  • Custom YAML workflows — Build fully customized agent workflows programmatically
  • Prompt-based generator — Describe your use case in natural language and let the system generate an agent configuration

Documentation