2 days · From prompt mastery to your first AI agent · No coding required
Choose the right model, write better prompts, verify AI outputs. Ends with a Prompt Competition judged by AI — anchored on the Case Context Summarizer your teammates already endorsed.
Understand autonomous agents, then build a Guided SOP Assistant for your market in Claude Cowork. Ends with Agent Design Showcase + AI Judge.
Every example, exercise, and build target in this workshop is grounded in the 9 best ideas curated from your cohort's submissions. Together they cover 6 of the 10 GS themes — Routing, Synthesis, Agent Guidance, Forecasting, Training, and SLA & Notification Triggers.
30 min → <10 min per case. Datalake tables already exist, multi-market, no engineering dependency. Anchors the prompt-engineering exercise.
Theme 2: Synthesis & Summarization · ID teamRAG over your SOP / KB corpus — the canonical "answer ONLY from this document" pattern. Every market handles SOP friction; this is the build target.
Theme 3: Agent Guidance & Resolution · TH Chatbot leadAlready shipped by a Regional TQM lead with Cursor. Five GS-style scenarios, real-time scoring, no backend. The proof point: you can ship in the same shape.
Theme 8: Training & Simulation · Enablement / Regional TQMSieve true positives vs false positives so IRT meets the 30-min first-email SLA. Rule-based classification — ideal CD build.
PH Training leadCall transcript → structured case notes. Reduces AHT for IRT; same problem family as Case Context Summarizer.
PH Training leadGuided SOP suggestions inline while the agent handles a live chat. Top Idea — directly connected to the Day 2 build target.
TH Vendor Mgmt leadWFM forecasting automation — shortens forecast prep from days to hours. Top Idea + CD-Perfect. Highest single ROI item in the cohort.
WFM teamReal-time auto-tagging for P0–P2 cases so resolution SLA isn't broken by handover gaps. Top Idea — newly elevated by sponsor review.
PH Training leadManage scheduled callbacks so customers don't fall through the cracks on missed re-contact promises. Top Idea — cross-market applicability.
SG/MY Senior OpsGS Operations Managers and Senior Managers from SG/MY, ID, TH, PH, VN, KH/MM — owners of case-handling AHT, FCR, and SLA outcomes
Specialists and Asst. Managers building chatbot flows, KB content, n8n automations, and the Cyborg apps themselves
Quality auditors, training leads, and regional TQM — pressure-test hallucination, scoring, and evidence trails
Forecasting, dashboards, OKR consolidation — the analytics + strategic-projects bridge to country GMs and GSTF
Persona + Chain-of-Thought + structured output, applied to a real D365 case from your market — your reusable template for IRT, MIWI, fraud, and DSAT write-ups.
Live Claude Cowork project with steering rules (citation required, no inventing SOPs), a saved Skill, and a chained quality-check Skill that flags low-confidence answers.
Haiku / Sonnet / Opus matched to GS tasks — PAC tagging vs case summary vs Safety severity decisions, with cost-per-case math at 400k MIWI scale.
PII (PAX/DAX/MEX) handling, hallucination detection on policy answers, citation requirements — the L1 / L2 / L3 trust ladder for GS workflows.
One-page reference: Steering · Skills · Hooks · MCP · Cowork mapping · GS-specific examples for each — when to use what, when to ask GTS / GSTF.
Strategic one-pager for your market's next agent — including a GSTF roadmap check so you know you're filling the interim gap, not duplicating a future build.
Use AI better today — choose the right model, write better prompts, verify the output.
Understand autonomous AI, then build one for your market's queue.