UiPath AI Trust Layer, Audit Trails, BYOLLM & Context Grounding Setup — Complete Platform Prerequisi
Everything you need to configure BEFORE building your first agent — in one session. We set up the AI Trust Layer (enable agents, configure guardrails, select LLM providers), explore the full audit trail system (every LLM call logged with input, model, region, and user), walk through Bring Your Own LLM (Azure AI Foundry, AWS Bedrock), create storage buckets, build a Context Grounding index, trigger programmatic index sync, and query the index using Gen AI activities — all without touching Agent Builder yet.
📌 What you'll learn in this session:
🛡️ AI Trust Layer — Complete Setup
Creating and deploying AI Trust Layer policy under Automation Ops governance
Enabling: agents, coded agents, app features, autopilot, document understanding, Studio Web
Enabling/disabling specific LLM providers — Anthropic, OpenAI, Gemini
Guardrails configuration — PII detection at platform level
Feature settings — saving input/output for audit compliance
Deploying policy to tenant — linking policy to your UiPath instance
📊 Audit Trails — Enterprise Differentiator
"Audit is the one key feature that differentiates a normal product from an enterprise product"
Every LLM call logged: product, timestamp, feature, tenant, user, model, region, status
Viewing input data passed to LLMs for each call
Filtering by tenant, product, and model
Usage summary — monitoring AI unit consumption across products
🤖 Model Hub & Bring Your Own LLM
UiPath-managed models — 30+ models across regions (US, EU, Singapore)
BYOLLM — connecting enterprise LLMs from Azure AI Foundry, AWS Bedrock
When BYOLLM is needed — organizational policy requires data to stay on their infrastructure
Custom model configuration — connection setup and authentication
OpenTelemetry integration — pushing GenAI logs to organization's monitoring platform (Datadog, Splunk, etc.)
📦 Storage Buckets & Context Grounding
Creating storage buckets — Orchestrator cloud vs Azure vs AWS storage options
Uploading files and creating folder structures
Creating Context Grounding indexes from storage buckets
Monitoring ingestion — succeeded/failed/skipped file tracking
On-demand sync and scheduled ingestion options
🔄 Programmatic Index Management
Update Context Grounding Index activity — triggering sync from RPA automation
Wait for Index Completed activity — ensuring sync finishes before agent queries
Critical warning: agents FAIL if they query an index during active ingestion
Fire-and-forget pattern — update triggers sync but doesn't wait
Combining both activities for safe, automated index maintenance
🔍 Context Grounding Search — Testing Without an Agent
Search Context Grounding activity — querying the index from an RPA workflow
Configuring max results and threshold values
Threshold = confidence filter — only results above the threshold are returned
Viewing results: source document, page reference, content snippet
"This is exactly how the agent queries the index during execution"
⏱️ Timestamps:
0:00 - Session Start & Screen Sharing
2:00 - AI Trust Layer — Creating Policy in Automation Ops
5:00 - Enabling Agents, Coded Agents, GenAI Capabilities
8:00 - Enabling/Disabling LLM Providers (Anthropic, OpenAI, Gemini)
10:00 - Deploying AI Trust Layer Policy to Tenant
12:00 - Admin Panel — Usage Summary & LLM Call Monitoring
15:00 - Audit Trails — "The Enterprise Differentiator"
18:00 - Viewing Input Data for Every LLM Call
20:00 - Model Hub — 30+ Managed Models & BYOLLM
24:00 - Azure AI Foundry & AWS Bedrock Integration
26:00 - OpenTelemetry — Pushing Logs to External Monitoring
28:00 - Storage Buckets — Creating with 3 Storage Options
32:00 - Creating Context Grounding Index from Storage Bucket
36:00 - Ingestion Monitoring — Succeeded/Failed/Skipped
38:00 - Update Context Grounding Index Activity — Programmatic Sync
42:00 - Wait for Index Completed — Why This Is Critical
45:00 - Warning: Agent Fails During Active Ingestion
48:00 - Context Grounding Search Activity — Querying Without Agent
52:00 - Threshold Values — Confidence-Based Filtering
55:00 - Q&A: Community Edition Availability, External Connectors Preview
58:00 - Tomorrow Preview: Agent Builder, Temperature, Tokens & Guardrails
👨🏫 Trainer: Pathrudu Chintakayala
UiPath MVP (2024/25/26)
7x UiPath Certified
Agentic Automation Expert & Community Speaker
#UiPath #AITrustLayer #AuditTrail #BYOLLM #ContextGrounding #StorageBuckets #ModelHub #OpenTelemetry #AgenticAutomation #GenAI #UiPathMVP #PlatformSetup #IndexSync #UiPath2025