{"slug": "building-an-escalation-root-cause-agent-with-gemini-and-adk", "title": "Building an Escalation Root-Cause Agent with Gemini and ADK", "summary": "A customer service team lead at Amazon built an escalation root-cause analysis agent using Google's Agent Development Kit (ADK) and Gemini, deployed on Cloud Run. The agent automatically classifies escalation summaries, flags repeat patterns, and drafts coaching notes, aiming to scale manual review processes. The developer also created a batch runner script for processing CSV files of cases and plans to integrate with ticketing systems via MCP in the future.", "body_md": "*Gen AI Academy APAC — Track 1 (AI Agents with Gemini, ADK, and Cloud Run)*\n\nI lead a customer service team of 25 agents at Amazon, handling both\n\nbuyer-side and marketplace seller support. A big part of my job is\n\nreviewing escalated cases — calls or chats where a customer asked for\n\na supervisor — and figuring out *why* they escalated in the first\n\nplace. Was it a policy gap? A training issue? A system limitation\n\nnobody flagged?\n\nRight now, that review is manual. Every escalation gets read, tagged,\n\nand turned into a coaching note by a human — usually me, or one of my\n\nleads. It works, but it doesn't scale well, and patterns across dozens\n\nof cases are easy to miss when you're reviewing them one at a time\n\nbetween everything else on your plate.\n\nSo for Track 1 of the Gen AI Academy APAC program, I built an agent\n\nthat does the first pass of this analysis automatically: read an\n\nescalation summary, classify the root cause against a standard\n\ntaxonomy, flag whether it looks like a repeat pattern, and draft a\n\ncoaching note — the same way I would, just faster and more\n\nconsistently.\n\nThe agent takes a case summary like this:\n\nCustomer requested a refund for a damaged item outside the standard\n\nreturn window. Agent denied it citing policy; customer says a rep\n\nlast month approved a similar exception for someone else.\n\nAnd returns a structured analysis:\n\n```\n{\n  \"root_cause_category\": \"policy_misapplication\",\n  \"severity\": \"medium\",\n  \"is_likely_repeat_pattern\": true,\n  \"pattern_reasoning\": \"Inconsistent policy application across agents\n  suggests a training or documentation gap rather than an isolated\n  error.\",\n  \"coaching_note\": \"...\"\n}\n```\n\nIt's built on **Google's Agent Development Kit (ADK)** with **Gemini**\n\nas the underlying model, and deployed as a live service on **Cloud\nRun**. The agent has one tool — a lookup function for the standard\n\nFor batch review, I also built a runner script that processes a whole\n\nCSV of cases at once and outputs a summary report — closer to how I'd\n\nactually use this day to day, reviewing a batch of the week's\n\nescalations rather than one at a time.\n\nA few things stood out going through this:\n\nThe repo includes a `README.md`\n\nwith full setup instructions if you\n\nwant to run it yourself, plus a sample CSV of synthetic escalation\n\ncases to test against (no real customer data used anywhere in this\n\nproject).\n\nThe natural extension is Track 2 — instead of feeding it a CSV\n\nmanually, connect it directly to a real ticketing system via MCP so\n\nit can pull escalations and analyze them without a manual export\n\nstep. That's the version I'd actually want running against my team's\n\nqueue.\n\n*Built for the Google Cloud Gen AI Academy APAC program, in\npartnership with Hack2Skill.*", "url": "https://wpnews.pro/news/building-an-escalation-root-cause-agent-with-gemini-and-adk", "canonical_source": "https://dev.to/aadilmajeedlone/building-an-escalation-root-cause-agent-with-gemini-and-adk-5b23", "published_at": "2026-08-21 21:20:28+00:00", "updated_at": "2026-08-21 21:44:10.036102+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-products", "developer-tools"], "entities": ["Amazon", "Google", "Gemini", "Agent Development Kit", "Cloud Run", "Hack2Skill", "Gen AI Academy APAC"], "alternates": {"html": "https://wpnews.pro/news/building-an-escalation-root-cause-agent-with-gemini-and-adk", "markdown": "https://wpnews.pro/news/building-an-escalation-root-cause-agent-with-gemini-and-adk.md", "text": "https://wpnews.pro/news/building-an-escalation-root-cause-agent-with-gemini-and-adk.txt", "jsonld": "https://wpnews.pro/news/building-an-escalation-root-cause-agent-with-gemini-and-adk.jsonld"}}