Accessibility Adjustments

Use these optional tools to adjust reading and display preferences. These tools cannot resolve every accessibility barrier. Please contact the website owner if you need assistance.

  • Text adjustments
  • Content scaling 100%
  • Font size 100%
  • Line height 100%
  • Letter spacing 100%
  • Colour adjustments
  • Orientation adjustments

DigitalOcean opens Signals preview and adds a private agent data service

Listen to this article

DigitalOcean opened Signals to public preview on October 12, broadening access to a service that flags troublesome AI agent conversations. The company also announced Agent Data Services in private preview, a separate system for preparing and querying the information agents use.

The announcements connect two practical problems for teams running agents. They need usable context before a task starts and evidence of what happened afterward. The previews address both, but their access requirements and limits differ.

Signals finds conversations worth reviewing

According to the current documentation, Signals can identify failed tool calls, repeated attempts and signs that users are dissatisfied. Teams must enable collection for each environment. Only subsequent conversations are analyzed, and accounts on a free trial cannot use the feature.

There is also a data handling decision. DigitalOcean’s privacy documentation says Signals processes raw interaction traces, potentially including prompts, tool arguments and intermediate reasoning. Automated redaction happens before content appears in the console and may miss sensitive material. Traces are processed and stored in a United States region. The company says this content is excluded from training generalized AI models.

Testing has its own bill

DigitalOcean is also promoting Simulations alongside Signals. These tests play out conversations against an agent in a sandbox, with a model acting as the user and another judging the outcome. The documentation calls evaluations advisory and recommends human review before production deployment.

The pricing page makes an important distinction. There is no separate charge per simulation run, but model tokens, scenario generation and the agent’s runtime are billed. Generation can cost money even when the resulting scenarios never enter a test.

The data preview is more restricted

Agent Data Services aims to ingest documents, media and execution records into a managed catalog that supports SQL and API queries. Its announcement describes automated processing and indexing, with source lineage linking records to their origins.

For now, access is limited to invited design partners, with workspaces hosted in the Memphis data center. The service is free during private preview, with future pricing subject to change. The private preview has no service level agreement or support commitment and is outside DigitalOcean’s standard compliance certifications.

The documented restrictions are substantial. Queries cannot change data. Dataset specific permissions and row or column access controls are absent. Customer written processing functions and connectors to systems such as CRM platforms are unsupported. Database records must be exported and ingested before queries can reach them.

Teams can use these previews to explore a connected workflow for context, monitoring and testing. A production decision still depends on the actual data permissions, billable resources and failures each team needs to check. We reviewed public documentation and did not test either service.

Server racks bearing the imgix logo. Illustrative photograph by imgix on Unsplash. Used under the Unsplash License.

Jordan Reid
Jordan Reid

Jordan Reid is focused on AI tools, agents, developer products, and the way technology changes everyday work. Jordan approaches a launch from the user’s side of the screen. What can it actually help someone finish? The voice is practical, conversational, and skeptical of products that turn a simple job into five new settings. Coverage follows coding assistants, creative software, browser agents, and the workflows around them, with attention to pricing, permissions, setup, and the human work that remains.

Leave a Reply

Your email address will not be published. Required fields are marked *

Gravatar profile