How ServiceNow and Outshift built federated AI discovery

TL;DR

Governed AI interoperability: Outshift and ServiceNow created a joint architecture that enables federated agent discovery and cross‑vendor workflow execution.
Open‑standards coordination: Using AGNTCY’s ADS and open standards, agents collaborate across clouds without custom integrations.
Faster, safer change verification: Automated end‑to‑end checks compress risk, testing, and approvals, reducing outages and accelerating CAB workflows.

Industry

Technology & Software

Company Size

15,000+ employees

Solution

Open standards (A2A, MCP, OASF) and AGNTCY Directory Service integrated with the ServiceNow AI Control Tower (AICT) governance control plane.

Summary

From agent sprawl to governed interoperability

Enterprises can only unlock meaningful automation when independent AI agents can discover, trust, and collaborate across business divisions. Outshift by Cisco and ServiceNow built an open standards reference blueprint to move enterprise AI out of isolated departmental silos and into a unified, governed, and federated workflow execution plane—compressing change verification from weeks to minutes. Outshift and ServiceNow demonstrated that multi-vendor AI agents can interoperate safely across cloud and organizational boundaries, executing end-to-end tasks with shared governance and standardized communication. By combining ServiceNow's centralized AI Control Tower (AICT)with Outshift's open source AGNTCY Directory Service, the teams successfully established a scalable blueprint for the modern digital enterprise. This achievement proves that practical, secure, and highly auditable agentic operations are possible today without custom-built API integrations, laying the foundation for a scalable Internet of Agents.

Challenge

Mitigating the 66% to 80% of network downtime caused by process bottlenecks

AI agents promise a future of autonomous enterprise operations, capable of handling everything from routine IT service tickets to complex, critical network architecture adjustments. However, this multi-agent software is notoriously difficult to coordinate outside of its native, single-vendor platforms. Each department—from Network Operations using Cisco tooling to IT Service Management using ServiceNow—quietly deploys its own specialized models. This division creates agent sprawl, duplicated capabilities, and broken workflows that halt the moment they cross organizational borders. A prime example is the Change Advisory Board (CAB) process. According to industry research, human error contributes to 66%–80% of all network downtime incidents. To prevent outages, change managers must manually coordinate multiple teams, run isolated tests, and assemble massive documentation packets. Historically, connecting independent AI agents to automate this process required expensive, brittle, custom-coded integrations. Outshift and ServiceNow needed to prove that multi-vendor AI agents could securely collaborate under a shared governance umbrella to automate these high-risk operations safely.

Outcome

Intelligent governance meets federated autonomy

The teams addressed these challenges by deploying a dual-layer architectural blueprint that cleanly separates central governance from distributed autonomy. ServiceNow AI Control Tower (AICT) anchors the governance plane, providing unified visibility, policy enforcement, and compliance oversight across every AI agent operating in the enterprise.

Outshift supplied the open‑standards data plane through AGNTCY Agent Directory Service (ADS), an open‑source, Linux Foundation–backed registry that enables agents to dynamically discover one another, establish trust, and collaborate across clouds using standardized protocols. Together, these layers successfully automated the entire cross-departmental change verification process by orchestrating specialized agents:

Intake: A ServiceNow Quality Assessor agent analyzes incoming firewall requests and drafts missing fields automatically.
Correlation: A ServiceNow Risk Analysis agent mines historical Configuration Management Database (CMDB) tickets to find patterns of past failures.
Coordination: A Cisco Network Specialist agent queries a live Network Digital Twin (NDT) to identify secondary downtime risks.
Calibration and testing: Cisco Test Planner and Test Executor agents fork a safe copy of the network topology, apply the proposed changes, and run automated testing scripts.
Data management: A Supervisor agent compiles the results into a fully audited, CAB-ready compliance packet.

By replacing manual, weeks-long coordination loops with software-defined automation, this project proves the feasibility of running distributed, multi-vendor agent systems with the exact same reliability and security expected of enterprise networks.

Probabilistic intelligence only delivers in the enterprise when it’s wrapped in deterministic governance — and that has to be true across every vendor in the portfolio, not just our own. Our work with Cisco makes that concrete: AI Control Tower as the enterprise control plane, with Cisco’s network specialists composing alongside ServiceNow AI Agents under one shared policy and audit fabric. That is how agentic operations cross the line from a single demo to workflows you can actually run a business on.
Nenshad Bardoliwalla, Group Vice President, AI Products, ServiceNow

Resources

BLOG

ServiceNow and Outshift by Cisco are untangling the multi-agent enterprise

- A partnership blueprint validates governed, federated operations — Outshift by Cisco + ServiceNow jointly prove a reference architecture for Federated Discovery & Distributed Workflow Execution, demonstrating CAB‑class change verification compressed from weeks to minutes. - Open standards provide the interoperability substrate — Open standards such as A2A, MCP, and AGNTCY's Directory Service and OASF (Linux Foundation, open source) enable multi‑vendor agents to compose workflows across org and cloud boundaries without glue code. - Control‑plane and data‑plane layers interoperate cleanly — ServiceNow AICT enforces governance centrally, AGNTCY’s open‑source components deliver the federated discovery data plane, and ServiceNow + Cisco ANV agents execute workflows grounded in CMDB and Network Digital Twin state for auditable, policy‑aligned automation.

VIDEO

ServiceNow and Outshift by Cisco: Governed Federated Discovery

This demo shows how ServiceNow and Cisco enable centralized governance and distributed AI autonomy using open, vendor‑neutral standards. Agents discover each other through AGNTCY Directory Service + OASF and execute governed workflows across clouds and orgs. ServiceNow AI Control Tower (AICT) acts as the governance control plane, enforcing enterprise and industry compliance across every AI asset.

VIDEO

ServiceNow and Outshift by Cisco: Distributed Workflow Execution workflow

AI agents take on the entire network change verification workflow — quality checks, risk analysis, impact assessment, and full test execution. See how ServiceNow and Cisco ANV agents collaborate to produce CAB‑ready evidence in minutes, not weeks.

WHITEPAPER

Governed Federated Discovery Whitepaper

How ServiceNow AI Control Tower and AGNTCY Agent Directory Service Deliver Federated, Governed Multi-Agent Workflows at Scale

WHITEPAPER

Multi-Agent Change Verification: From Ticket to CAB in Minutes, Not Weeks

How Multi-Vendor AI Agents, a Network Digital Twin, and Open Standards Transform Enterprise Change Management Workflows

Our Work

Our Collaborators

Company

Apply

Connect

Blog

Categories

Tools

Resource Hub

Apps & Services