Agentic AI: Xurrent's Q2 2026 Release

Sera AI has been part of how Xurrent works for years, routing and classifying requests, drafting knowledge articles, and assisting users directly as a virtual agent, with 91% of customers already running it in production. Today, we're extending that foundation with four new capabilities that span the full range of how enterprise IT teams work with AI. Whether your team wants AI woven into the interface they already use, or you're ready to deploy autonomous agents across your workflows, this release meets you where you are.

Sera AI Assist

Most enterprise teams aren't ready to abandon their service desk interface for a blank chat prompt. They shouldn't have to.

Sera AI Assist brings the power of an LLM conversation directly into the specialist experience your team already knows. When a specialist opens a request, the Sera AI Assist panel is right there, ready to summarize the issue, surface related problems, or look for similar requests with a single click. When they need to go deeper, they ask. In natural language, without switching tools or leaving the ticket.

Draft a response. Query the system. Get context on a CI. The capabilities are there when a specialist needs them, and out of the way when they don't. For teams that want to move forward without a forced paradigm shift, this is how you get there. The interface your specialists trust doesn't go away. It gets smarter.

Sera AI Studio

Good AI is built, not just deployed. Sera AI Studio gives IT teams the controls to configure, test, and tune how their virtual agent behaves before any of it touches end users.

The Agent Instructions tab lets administrators define exactly how the virtual agent understands and routes requests: company terminology, behavioral rules, escalation logic, sensitivity handling. It's a structured markdown file that shapes how the agent thinks about your environment specifically, not a generic one.

The Golden Set tab is where you validate it. Teams define expected outcomes for specific prompts. "I need to reset my password" should surface the VPN Password Reset Guide, for example. Then they run scoring tests against the live agent. Each row returns a pass or fail, a response time, and an AI-generated suggestion for improvement. A configurable scoring threshold flags anything that falls below your acceptable accuracy bar, and a notification fires if a re-run produces a lower score than the previous one. That last detail matters: it catches regressions before your users do.

The result is an AI deployment your organization can actually trust, because you tested it, tuned it, and know how it performs.

Xurrent MCP Server

For organizations operating with a broader AI ecosystem, including custom agents, preferred LLMs, and automation frameworks built outside of Xurrent, the MCP server opens the platform to all of it.

The Model Context Protocol server lets customers connect any external AI model to Xurrent, whether that's OpenAI, Google, Anthropic, or one built in-house. External agents don't just get access to Xurrent data. They inherit the Shared Policy and Data Layer, operating under the same governance, audit trail, and visibility as everything else running on the platform. No matter where the intelligence comes from, how it behaves inside Xurrent is governed consistently.

This is what open AI interoperability looks like when it's done right. Not a free-for-all. One platform, one governance layer, open to whatever your team wants to bring.

Sera AI Agents

Sera AI has been assisting the IT team for years. The agents that launched today enable autonomous action.

Xurrent's AI Agents arrive as digital members of the IT team. They're onboarded like any specialist, assigned to teams, given specific roles, and held to the same accountability standards as the humans working alongside them. The Agent Studio dashboard shows you exactly how they're performing: how many times they've run this week, their average success rate, and how much time they've saved.

Each agent comes with configurable skills. A Triage Agent can run duplicate detection, CI linking, completeness gating, and impact assessment, with each skill carrying its own confidence threshold and action type so you decide how much autonomy the agent gets before it acts versus suggests. A Triage Agent can be working an active request right alongside your human team, asking the end user the right clarifying questions and routing based on the answers. Every action is logged, auditable, and governed by the same policies that apply to everyone on the platform.

This is only safe to do because of what Xurrent was built on. The Shared Policy and Data Layer, the Service Catalog and Data Model, the governed foundation that's been in place for a decade, is what makes agentic AI trustworthy in an enterprise environment. Deploying autonomous agents without that kind of architecture isn't innovation; it's exposure. Xurrent customers don't have to choose between AI productivity and organizational control. They get both.

IT teams don't just manage the work anymore. They direct the workforce that does it. Get started with Xurrent today.