Lexul AI Support Knowledge Base

Audience: Lexul customer support AI agent, onboarding team, support team, and documentation maintainers.
Use: Host this folder in Fumadocs. Configure the AI agent to retrieve only the needed group page or individual ### Knowledge Record: chunks.

What changed

This package splits the previous single chatbot training page into standalone feature-group routes. This improves retrieval efficiency because the AI agent can load only the relevant feature group instead of reading the full knowledge base.

Recommended retrieval flow

Classify the user's request into one or more intents from Schema and Routing.
Match the request to a feature group page.
Retrieve the top one to five ### Knowledge Record: chunks from that page.
Answer using Agent Instructions and the retrieved records.
Escalate when the record's escalation_rule applies or when retrieved records do not confirm the requested behavior.

Page map

Page	Purpose
Agent Instructions	Short system-style behavior profile for the support bot.
Schema and Routing	20-field documentation schema, intent labels, record format, and chunking rules.
Feature Groups	Index of feature-group knowledge pages.
Source Index	Lexul.com and help.lexul.com URLs used by the training records.
Ingestion and Evaluation	Metadata recommendations and chatbot evaluation checklist.

Feature group retrieval sizes

Use this table to decide how much context to retrieve. Token counts are approximate and assume roughly four characters per token.

Feature group	Records	Approx. page tokens
Customers	10	~7,118
Assets	7	~5,005
Sales	4	~2,968
Deals	2	~1,145
Contracts	6	~4,072
Inventory	6	~5,177
Equipment	2	~1,462
Inspections	3	~1,909
Receipts	2	~1,385
Timesheets	3	~2,173
Users	2	~1,416
Time Tracking	5	~3,337

Packaging notes

Generated: 2026-06-23
Content format: Fumadocs-compatible MDX
Chunking target: one ### Knowledge Record: per vector record
Best production use: retrieve by feature group first, then by record-level semantic similarity