Introduction
WINGS is part of the Wipro Intelligence suite, using AI agents to run IT service delivery, moving support from reactive ticket handling to proactive operations. At a large client, a global consumer goods company with 80,000 employees, WINGS agents resolved 65% of IT service desk tickets end to end: password resets, access requests, software installs, VPN issues and common laptop problems. In one month, the resolution share fell to 48%. The client's CIO is concerned because Wipro's contract includes productivity commitments tied to agent performance. The question asks what is going on. I'll use a structured root cause analysis.
For AI agent performance drops, consider four layers: the inputs (what tickets arrive), the agent (models, prompts, tools), the environment (systems the agent acts on) and the measurement (how resolution is counted). A sudden change usually points to one layer changing.
Step 1
Clarifying Questions (3 minutes)
Why it matters: Definitions can shift. Expected answer: Tickets closed by agents without human touch, not reopened within 72 hours. Impact on approach: Reopen rates matter too.
Why it matters: Sudden drops suggest a change event. Expected answer: Sudden drop in week two of the month, then stable at the lower level. Impact on approach: Look for a specific change around that week.
Why it matters: Denominator effects. Expected answer: Volume up 12%; more tickets about a new collaboration tool. Impact on approach: Check mix.
Why it matters: Agent or environment changes. Expected answer: WINGS moved to a newer LLM version; the client migrated identity management to a new single sign-on provider. Impact on approach: Two candidate change events.
Practice similar questions
Subscribe to access the full answer
.png){kind=logo}
