SHARKPARTY
Case studies
Case study · Platform · AI

Auth0 Guide.

An AI-powered chatbot that answers your questions about Auth0 in natural language—grounding every response in Auth0’s own Documentation, Blog, and Community, with links back to the source. For Enterprise tenants, it reads Security Center to help monitor, troubleshoot, and harden security.

Role—Staff Engineer & Tech Lead, AI Frameworks & Services Status—Early Access · US Public Cloud
ReactRxJSNode.jsgRPCProtobufmTLSPythonLangGraphRAGLLMs
// What it does
01
Answers, with receipts

Ask “How do I set up passkeys?” in natural language and get a summarized answer plus links to the exact Documentation, Blog, and Community sources it drew from.

02
Context-aware prompts

Guide suggests questions relevant to where you are in the Dashboard—under Branding it offers “What is Universal Login?”, and so on.

03
Security Center incidents

For Enterprise tenants, Guide reads Security Center metrics to help troubleshoot alerts and recommend remediation from the matching incident playbook.

04
Playbook remediation

When a bot surge trips an alert, Guide surfaces Bot Detection playbook steps—like enabling CAPTCHA on affected flows—and points you at the relevant tenant logs.

// How it works
01
Understand

Guide processes your message to determine intent—a product question, or a tenant-security one.

02
Retrieve

It searches Auth0 sources—Docs, Blog, Community, and (for Enterprise) Security Center data—for relevant material.

03
Synthesize

It generates a summarized response grounded in what it found, with links back to every referenced source.

04
Verify & refine

You ask follow-ups, validate against the provided links, and rate responses to close the feedback loop.

// Architecture

A streaming path from question to grounded answer.

Event-driven UI

A React frontend orchestrates the message flow with RxJS—streaming, retries, and resilient failure handling.

Node.js gateway

A gateway brokers the conversation and fans requests to backend services.

gRPC over mTLS

Auth0’s first bidirectional gRPC layer, defined in Protobufs and secured with mutual TLS between gateway and services.

LangGraph agent

A Python service where LangGraph orchestrates the agent framework and the RAG-backed LLM.

The result is a RAG-backed chat that makes Auth0’s advanced capabilities discoverable rather than merely present—and, for Enterprise tenants, turns a security alert into a concrete set of next steps instead of a dead end.

// Scope—sharp on purpose
Can answer
  • What is Universal Login?
  • How do I enable integrations?
  • How do I set up passkeys?
  • What triggered my bot protection alert threshold? (Enterprise)
Won’t answer
  • What connections do I have enabled?
  • How many users logged in last month?
  • Vim or Emacs?
  • Anything with personal or sensitive data

// Grounded in Auth0-authored content with source links on every response—messages aren’t used for LLM training or shared with third parties, and Guide doesn’t touch account or tenant data beyond Security Center metrics you’ve opted into.

// Want the deep version?

Happy to talk through the hard parts.

Email Kam More work