What's new?

UAT release	Production release
April 27, 2026	May 11, 2026

This month we are releasing a long-requested feature to edit existing Spark service versions. We also have a number of enhancements that improve the performance and usability on the platform.

Edit existing Spark service versions

Make changes to a Spark service even after it has been published!

Spark services have historically been "locked-in" after a version has been created. Any changes to an existing version would require a new version to be created. In this update, we include the ability to edit a service version without having to create another version.

This makes it easy to add information that was missed at the time of upload or test different configuration values. This supplements the ability to delete service version from the prior release.

Notable Enhancements

• Improved Xcall performance if a specific service version number is specified.

• Even faster publish time for Spark services.

• API Call History quicker user interface and query response for complex queries.

• API Tester OpenAPI code snippets are much more comprehensive.

• Testing Center Testbed Comparison is moved to a background activity to minimize impacts to Spark session.

• Batch improved stability for high chunk counts.

• API Call History and Testing Center, Excel downloads that include expanded tables now have support for long input and output names.

• Coherent Assistant service upload includes additional fields for release notes and description.

• Homepage and Folder overview UI adjustments for consistent pagination.

• MCP authentication flow is greatly simplified and MCP tool updates. See Model Context Protocol (MCP) for more details!

• Fixed an issue with API keys including tenant-admin group. Note this is not a recommended pattern but partially supported for compatibility.

• Security patch updates to addresses vulnerabilities, ensuring enhanced protection and stability of our systems.

Why We're Different at AI for Excel

AI is everywhere, but most of it wasn't built for the way Excel actually works. We've spent years turning spreadsheets into code through Spark, and that's the same foundation behind how our AI reads, reasons over, and governs your workbooks.

In a new demo video, our Product team walks through what that looks like in practice — from a single workbook in Coherent Studio (open and free for you to try) to scaling AI analysis across hundreds with Projects.

Feature Deprecation Notice

We will be deprecating these features in the June release:

Farewell to Spark Assistant

Our older Excel add-in Spark Assistant will be sunset. The new Coherent Assistant has the same features plus many more capabilities. If you are using older Spark Assistant functions such as Call Spark service APIs (Xcall), those can be migrated or updated to the Coherent Assistant version through the add-in debugger or a simple find and replace.

Service Chain Response Object

From the Spark Execute API (v3), the response_data currently includes an object called service_chain. We will be removing this object from future API responses. If your integrations to Spark contain explicit references to the service_chain object, we would suggest removing this reference in your application updates.

Model Templates

Within the Options menu of Spark is the ability to store template documents for Spark services. If you have important documents within Model Templates, they can be downloaded and uploaded into the Documents section within a Folder instead.

Modeling Center

Our beta Modeling Center Excel add-in will also be sunset. Modeling Center was used to facilitate large scale execution within Excel. If you are interested in using modeling Spark services, we have alternatives with the Testing Center and Batch APIs with the Python SDK.

Any questions?

Whether it's about an upcoming change, a new feature, or just getting the most out of Coherent, our Customer Success team is here to help. Reach out to our team anytime at [email protected].

Version number

v8.52.0