Cisco Outshift features DataWell as the system intelligence layer for agentic AI

Cisco’s Outshift division featured DataWell in its AGNTCY Builders Blog, highlighting the Versai Labs product as a way to map how connected AI agents influence each other after deployment. The pitch lands as enterprises face rising AI downtime, unpredictable agent behavior and new EU transparency rules.

Why it matters: - Enterprises are deploying agents faster than they can understand how those systems behave once connected. - Splunk research cited in the release says every technology leader surveyed reported at least one form of AI-related downtime, and 68% worry their agents will behave unpredictably. - The release argues that visibility into dependency structures is becoming a governance requirement, not just an operations benefit.

What happened: - Cisco’s Outshift division featured DataWell this week in its AGNTCY Builders Blog. - Versai Labs positions DataWell as the system intelligence layer beneath AGNTCY-connected agents. - DataWell is live at getdatawell.com.

The details: - AGNTCY, created by Cisco and donated to the Linux Foundation, provides identity, discovery, secure messaging and interoperability for the Internet of Agents. - The Linux Foundation project is backed by Google, Red Hat, Dell, Oracle and more than 80 partners. - Benjamin Torres, founder and CEO of Versai Labs, said AGNTCY solved the connection layer and DataWell maps what happens after trust is established. - DataWell describes its core capability as causal topology, which maps the relationships beneath agentic cooperation. - The system is meant to show how components move each other, not just what each component is doing. - On a real infrastructure dataset, DataWell identified 211 direct relationships across one 8-hour window. - The engine also modeled a 70.5% improvement in disk I/O wait time before any system changes were made. - Noelle Flint, author of Cisco’s feature post, wrote: “This is not metrics. It’s system intelligence.” - The same approach is aimed at LLM infrastructure and security infrastructure, where behavioral shifts and dependency changes can surface before they show up in cost or incident reports. - DataWell’s demo path for AGNTCY-connected environments centers on Agent Zero, an ambassador agent that enters through AGNTCY Identity, communicates over SLIM and registers in the directory. - Agent Zero invites other agents to consent to share telemetry, then DataWell returns a Behavioral Confidence Profile showing where topology is stable, where behavior is shifting, where pressure is building and how confident each finding is. - The company says organizations can request a demo or early access through the DataWell site. - Related links in the release include DataWell’s full write-up and Cisco’s feature post.

Between the lines: - Cisco’s feature suggests AGNTCY is becoming the standard connective layer, while DataWell is trying to own the post-connection intelligence layer. - That positioning targets a core enterprise pain point: connecting autonomous agents is easier than proving they remain stable, explainable and governable after launch. - The EU AI Act pressure cited in the release raises the stakes for traceability, especially for systems that cannot show how decisions and behaviors emerge.

What’s next: - Versai Labs is pushing DataWell into infrastructure, agentic AI architectures and AGNTCY environments through demos and early access. - Agent Zero is the main on-ramp for customers that want to test the system intelligence layer inside connected agent environments. - The company is also using the Cisco feature to broaden awareness inside the AGNTCY ecosystem.

The bottom line: - DataWell is betting that the next AI infrastructure race is not just about connecting agents, but about understanding the hidden relationships that govern what those agents do once they start working together.