AI Support Copilot Pilot

Client: Prasanna (CIO / Business Lead / Product Owner) POD: Gyde AI POD Duration: 4 weeks (2 sprints) Status: Discovery

---

Engagement Summary

Build an AI support copilot that helps agents understand incoming tickets, retrieve relevant knowledge, draft grounded responses, recommend next-best-action (including escalation), and provide traceability for review.

Phase 1 scope: one support queue, one knowledge source, one escalation mechanism, human-in-the-loop only.

---

POD Composition

Role	Person	Focus
POD Lead	Amit	Architecture, eval strategy, technical direction
AI Engineer	Atharva	Prompts, retrieval pipeline, model integration
Data Engineer	Nancy	Ingestion, chunking, embeddings, data quality
QA	Nishka	Test strategy, golden set, eval harness, adversarial testing
Governance Engineer	Shubham	Threat model, security, responsible AI
Implementation Manager	Shivani	Charter, delivery, client comms, sprint cadence

2nd Level Escalation SPOC: Shubham

---

Document Index

Discovery

• Discovery Call Questions (Technical)
• Discovery Call Questions (PM)
• Discovery Call Invite Email

Team

• Product Overview & Technical Primer

Engagement

• POD Charter -- pending Discovery
• Architecture Sketch -- pending Discovery
• Evaluation Plan -- pending Discovery
• Sprint Plan -- pending Discovery
• Risk Register -- pending Discovery