L5 Knowledge Center#

Path: /knowledge

The knowledge center is the platform for accumulating, searching and managing enterprise AI knowledge assets. Knowledge runs consistently through L1-L6: assessment/decision/planning/delivery/operations processes are automatically archived as knowledge assets.

Overview#

Path: /knowledge/overview

The dashboard shows: total assets and distribution, knowledge base status, recently added assets, AI memory usage.

Knowledge Base Management#

Path: /knowledge/bases

Steps:

1. Click New Knowledge Base to create
2. Fill in name and description
3. Upload documents (PDF/Word/Excel/plain text etc.)
4. The system automatically chunks and embeds them
5. Check embedding task status

Knowledge base detail (/knowledge/bases/{id}): document list, embedding task status, search testing.

Embedding task states:

State	Description
pending	Queued
processing	Processing
completed	Completed
failed	Failed
cancelled	Cancelled

Embedding task types:

Type	Description
full	Full embedding
rechunk	Re-chunking
update	Incremental update
delete	Delete embeddings

Knowledge Search#

Path: /knowledge/search

Vector-retrieval based search:

1. Type a natural-language query in the search box
2. The system performs semantic search and returns relevant document fragments
3. Results include relevance scores and source references
4. Filter by knowledge base, asset type, tags

Knowledge Asset Management#

Path: /knowledge/assets

Asset types:

Type	Description
document	Document
project_memory	Project memory
scenario_memory	Scenario memory
pattern	Pattern
ai_asset	AI asset
methodology	Methodology
best_practice	Best practice
decision_record	Decision record
meeting_note	Meeting minutes
report	Report
template	Template
code_sample	Code sample

Asset domains:

Domain	Description
enterprise_knowledge	Enterprise knowledge
project_knowledge	Project knowledge
scenario_knowledge	Scenario knowledge
methodology_pattern	Methodology & patterns
ai_assets	AI assets

Asset states:

Asset visibility (three tiers):

Visibility	Who can see	Typical use
private	Creator only (even tenant admins can't see it; unauthorized access gets 404)	Personal drafts, unfinalized content
project	Project members	Internal project sharing
tenant	All tenant members	Published, accumulated knowledge

🔒 Deletion/editing of private assets is likewise creator-only; admins cannot view or dispose of others' private assets (eliminating the "managed but invisible" privilege path).

💡 Asset publishing (draft → active or visibility raised to tenant) is done by management roles; the system records published_by / published_at; exports are gated by visibility — invisible assets never appear in export results.

Asset relation types:

Relation	Description
references	References
derived_from	Derived from
implements	Implements
generated_by	Generated by
related_to	Related to
depends_on	Depends on
supports	Supports
uses	Uses
replaces	Replaces
supersedes	Supersedes
produced_by	Produced by

Relation sources: system / ai / manual

Automatic Capture Mapping#

The system automatically creates knowledge assets when these business objects are created:

Source entity	Asset type	Asset domain
assessment	document	enterprise_knowledge
assessment_report	report	enterprise_knowledge
decision_scenario	scenario_memory	scenario_knowledge
decision_report	report	scenario_knowledge
planning_plan	methodology	methodology_pattern
planning_artifact	document	methodology_pattern
delivery_project	project_memory	project_knowledge
delivery_task	project_memory	project_knowledge
delivery_milestone	project_memory	project_knowledge

Tag categories:

Category	Description
industry	Industry tags
domain	Domain tags
technology	Technology tags
type	Type tags
custom	Custom tags

Asset detail (/knowledge/assets/{id}): content and metadata, relation graph, version history, citation chain.

Knowledge Graph#

Path: /knowledge/graph

Visualizes relations between knowledge assets:

• Nodes: knowledge assets
• Edges: relation types
• Supports zoom, drag and filtering

Knowledge Packages#

Path: /knowledge/packages

Knowledge packages are structured knowledge collections.

Collection types:

Type	Description
manual	Manually created
auto	Auto-generated
system	Built-in

Eval Templates#

Path: /knowledge/eval/templates

Manage evaluation templates, supporting:

• Template creation and editing
• Template runs (evaluation execution)
• Run result viewing
• A/B testing
• Dataset management
• Report generation

Related pages:

Page	Path	Description
Template list	/knowledge/eval/templates	Template management
New template	/knowledge/eval/templates/new	Create a template
Template detail	/knowledge/eval/templates/{id}	View and edit
Run list	/knowledge/eval/runs	Evaluation run records
Run detail	/knowledge/eval/runs/{id}	Run result details
Report list	/knowledge/eval/reports	Evaluation reports
Dataset list	/knowledge/eval/datasets	Evaluation datasets
Dataset detail	/knowledge/eval/datasets/{id}	Dataset content
A/B testing	/knowledge/eval/ab-test	A/B comparison

Enterprise Memory#

Path: /knowledge/memories

Enterprise memory is the AI's working memory: key knowledge fragments of the enterprise, referenced by AI Agents during conversations and reasoning.

Steps:

1. Browse the enterprise memory list
2. Add entries manually or let AI generate them automatically
3. Memory types include: decision preferences, business rules, org structure, etc.
4. AI automatically retrieves relevant memories in Copilot conversations to assist answers

Knowledge Audit#

Path: /knowledge/audit

Audit log of all knowledge center operations:

• Asset create/modify/delete
• Embedding task execution
• Relation changes
• Search query records