Airbyte adds action-taking AI agents to OpenAI marketplace
Thu, 9th Jul 2026 (Today)
Airbyte has updated its Agents platform so AI agents can take actions as well as retrieve data. The release also makes the platform available through the OpenAI App Marketplace.
It adds a command-line interface for developers and introduces write functions for Salesforce, allowing agents to update records in the customer relationship management system.
The latest release is intended to expand how AI agents work with business data across internal systems and software applications. Airbyte framed the changes as part of a broader shift in which AI tools are expected not only to answer questions, but also to carry out tasks.
"AI agents are quickly evolving from passive assistants into active operators," said Michel Tricot, Chief Executive Officer and Co-Founder of Airbyte.
"For agents to create real business value, they need reliable access to data and the ability to take action. With these updates, Airbyte is making it easier to build agents that don't just answer questions, but actually get work done," Tricot said.
Marketplace access
By joining the OpenAI App Marketplace, Airbyte is making its service available within a broader ecosystem of AI tools used to build and run agent-based workflows. Users can connect agents to hundreds of enterprise data sources through Airbyte's connector catalogue.
The marketplace listing is meant to reduce the need for companies to build custom data pipelines before connecting agents to customer records, internal systems and business applications. It includes OAuth support and credential management, while Airbyte's context store indexes data for search.
Command line tools
The new Airbyte Agent CLI gives developers a single-binary interface for using Airbyte Agents in terminals, scripts, and continuous integration or continuous delivery pipelines. The tool can list connectors, inspect schemas, run reads and searches, and perform create and update operations across connected systems.
The CLI also lets developers discover command structures dynamically through schema discovery. That provides another way to work with enterprise data alongside Airbyte's software development kit for Python integrations and its support for the Model Context Protocol used by large language model clients.
Salesforce writes
Adding write operations to the Salesforce Agent Connector marks a notable shift in how agents can interact with one of the most widely used sales and customer management systems. Until now, many AI agents working with enterprise data have been limited to pulling information or generating summaries.
With the new feature, agents can update customer records, change status fields, and handle workflow steps directly in Salesforce. One use case, Airbyte said, would let sales teams make deal status changes through ChatGPT and have those updates reflected in Salesforce automatically.
Wider strategy
The release is part of Airbyte's broader effort to position its platform as infrastructure for agent-based business workflows. Businesses are adopting AI agents across sales, support, operations, and engineering, increasing demand for controlled two-way access to enterprise systems.
Airbyte's platform is built on its open-source data replication software and is used to move structured and unstructured data across cloud and hybrid environments. The company says it is trusted by 7,000 enterprises and provides a shared foundation for both analytics pipelines and AI agents.