Sonos, EXL, and AWS bring agentic AI into ITSM: what managers need to know
Sonos is working with EXL and AWS to embed agentic AI into its IT service management workflows. The goal: streamline decisions, speed up resolutions, and reduce risk across core operations. For leaders, this signals where enterprise IT is heading-AI as an active operator inside processes, not a bolt-on tool.
What changed
The collaboration integrates agentic AI into Sonos ITSM so systems can analyze context, decide on next steps, and act-while staying inside policy and control limits. Think automation that doesn't just execute tasks, but reasons through them.
"Our collaboration with EXL and AWS has been instrumental in enhancing our ITSM processes," said Jim Hauser, head of IT infrastructure at Sonos. "The innovative agentic AI solution developed through this collaboration not only streamlines our operations but also enhances the experience we deliver to our customers. This initiative demonstrates the power of combining expertise from multiple partners to achieve transformative outcomes."
"We are excited to collaborate with Sonos and AWS on this forward-looking initiative," said Narasimha Kini, executive vice president, banking & capital markets and diversified industries at EXL. "This collaboration is a testament to our commitment to creating solutions that embed agentic AI directly into our client workflows to deliver tangible business value. It's a powerful story of how partners can accelerate innovation and address critical business needs."
"IT service management is one of the functions most poised to benefit from agentic AI," said Chris Sullivan, vice president, Americas Channels & Alliances at AWS. "Our scalable and secure AI services, coupled with EXL's expertise, empower Sonos to innovate faster, deliver personalized experiences, and redefine the future of smart audio technology."
Why it matters for management
- Faster recovery, fewer escalations: AI agents can triage incidents, retrieve context, and execute safe actions, trimming MTTR and easing load on senior engineers.
- Operational intelligence built-in: Patterns across tickets, changes, and telemetry get surfaced early so you can fix systemic issues before they spread.
- Lower operational risk: Policy-aware automation reduces human error in repetitive tasks and standard changes while improving auditability.
- Better customer and employee experience: Quicker answers and fewer handoffs lead to higher CSAT/ESAT without adding headcount.
What agentic AI changes inside ITSM
- Proactive prevention: Correlates logs, alerts, and past incidents to flag likely failures and suggest fixes before impact.
- Intelligent triage and routing: Classifies tickets, extracts key signals, and routes to the best resolver or automates resolution directly.
- Knowledge into action: Searches runbooks, change history, and KBs, then executes approved steps or drafts remediation for review.
- Policy-aware change automation: Automates standard changes, validates preconditions, and rolls back safely on failure.
- Continuous learning loop: Learns from resolved tickets and outcomes to improve accuracy and reduce noise over time.
Guardrails leaders should insist on
- Clear decision boundaries: Define which actions agents may take autonomously vs. require human approval.
- Audit and traceability: Every decision, prompt, tool call, and data source should be logged and reviewable.
- Data security and access controls: Enforce least privilege for agents; segregate sensitive data; encrypt in transit/at rest.
- Policy and compliance alignment: Map agent workflows to change, incident, and problem policies; keep CMDB as the source of truth.
- Quality and drift monitoring: Track false positives, action success rates, and model drift; retrain with governed datasets.
- Cost control: Monitor per-ticket AI costs, tool usage, and inference latency; set budgets and alerts.
KPIs to track from day one
- Mean time to acknowledge (MTTA) and mean time to resolve (MTTR)
- First contact resolution rate
- Change failure rate and unplanned rollback rate
- Autonomy rate (tickets or actions resolved by agents within guardrails)
- Proactive incident avoidance (alerts addressed before user impact)
- CSAT/ESAT for IT services
- Cost per ticket and cost per change
90-day execution playbook
- Weeks 1-2: Identify top 5 incident types by volume and impact; extract runbooks and policies.
- Weeks 3-4: Define guardrails; decide which steps can be fully automated vs. require human-in-the-loop.
- Weeks 5-6: Integrate with your ITSM (ServiceNow, Jira Service Management, etc.), CMDB, monitoring, and identity systems.
- Weeks 7-8: Pilot on one high-volume use case (e.g., password resets, disk cleanup, cache flushes) with clear success criteria.
- Weeks 9-10: Review safety logs, failure modes, and user feedback; tighten prompts, tools, and policies.
- Weeks 11-12: Expand to a second use case; start a weekly ops review to manage performance, cost, and risk.
Leadership takeaways
- Agentic AI moves ITSM from ticket handling to decision automation with accountability.
- Success depends on crisp guardrails, high-quality runbooks, and measurable outcomes.
- Start narrow, automate safely, and scale based on evidence, not hype.
About the companies
EXL is a global data and AI company listed on Nasdaq that helps enterprises reinvent business models and drive outcomes using data, AI, and industry expertise. Learn more at exlservice.com.
Sonos builds a connected audio system for homes and portable use, serving more than 17 million households across over 60 countries. Explore products and updates at sonos.com.
AWS provides scalable cloud and AI services that support secure, high-performance enterprise workloads. For an overview of AI services, visit AWS AI & ML.
Next steps for IT leaders
- Assign an AI+ITSM owner with authority over process, data, and tooling.
- Prioritize two automatable use cases tied to business impact and risk reduction.
- Set transparent governance: human-in-the-loop thresholds, audit, and rollback.
- Publish a living runbook and track the KPIs above in your weekly ops review.
For practical guidance on planning and operating AI-enabled ITSM, see the AI Learning Path for IT Managers.
Your membership also unlocks:
