How sellinaxie/enterprise-agentic-ai-platform works
V8D is an authenticated enterprise AI architecture and risk-assessment platform. Its main workflow accepts an assessment, retrieves grounded knowledge, executes deterministic or agentic analysis, applies runtime risk controls, and persists results for product views or human review. A separate knowledge workflow parses text or bounded document uploads, then chunks, embeds, and stores material for vector and GraphRAG retrieval; an opt-in policy connector reuses ingestion. Groups also capture identity, provider portability, evaluation, telemetry, and the frontend. Some route, workflow, and agent implementations were not sampled, so uncertain connections are omitted or marked unverified.
Main entry points: assessments.py Assessment orchestration · ingestion.py Knowledge ingestion · retrieval.py Evidence retrieval · knowledge.py Knowledge endpoints
Components 24
Assessment product
- FastAPI application
app/main.py - Versioned API routes
app/api/router.py - Assessment endpoints
app/api/routes/assessments.py - Assessment orchestration
app/services/assessments.py - Execution selection
app/services/execution_router.py - Assessment agents
app/agents - Assessment workflows
app/graph - Assessment persistence
app/repositories/assessments.py
Knowledge and retrieval
- Knowledge endpoints
app/api/routes/knowledge.py - File parsing
app/ingestion/service.py - Knowledge ingestion
app/rag/ingestion.py - Knowledge storage
app/repositories/knowledge_chunks.py - Evidence retrieval
app/rag/retrieval.py - GraphRAG enrichment
app/knowledge_graph - Policy connector
app/connectors/policy_repository.py
Identity and governance
- Human review queue
app/repositories/runtime_reviews.py - Authentication and RBAC
app/identity - Runtime risk gates
app/runtime/gate.py
Model and runtime platform
- Model providers
app/providers - Assessment evaluation
app/evaluation - Runtime telemetry
app/runtime/telemetry.py
Product interface
- Authenticated product UI
frontend/app - Public showcase
frontend/app/demo
Other components
- Analyst / reviewer
Connections 27
- Analyst / reviewer → Authenticated product UI: uses
README.md - Authenticated product UI → FastAPI application: calls API
README.md - FastAPI application → Versioned API routes: mounts routes
app/main.py - Versioned API routes → Assessment endpoints: includes
app/api/router.py - Assessment endpoints → Assessment orchestration: dispatches
- Assessment orchestration → Execution selection: selects execution
- Execution selection → Assessment agents: routes execution
- Execution selection → Assessment workflows: routes execution
- Assessment orchestration → Evidence retrieval: retrieves evidence
- Assessment orchestration → Assessment persistence: persists results
- Assessment orchestration → Runtime risk gates: checks risk
- Assessment orchestration → Human review queue: queues review
- Knowledge endpoints → File parsing: accepts uploads
app/api/routes/knowledge.py - File parsing → Knowledge ingestion: delegates ingestion
app/ingestion/service.py - Knowledge ingestion → Knowledge storage: stores chunks
app/rag/ingestion.py - Knowledge endpoints → Knowledge ingestion: ingests documents
app/api/routes/knowledge.py - Knowledge endpoints → Evidence retrieval: searches evidence
app/api/routes/knowledge.py - Evidence retrieval → Knowledge storage: searches vectors
app/rag/retrieval.py - Knowledge ingestion → GraphRAG enrichment: enriches graph
- Evidence retrieval → GraphRAG enrichment: combines retrieval
- Policy connector → Knowledge ingestion: reuses ingestion
app/connectors/policy_repository.py - Assessment endpoints → Authentication and RBAC: requires principal
README.md - Authenticated product UI → Human review queue: supports review
README.md - Model providers → Assessment orchestration: supplies generation
- Assessment evaluation → Assessment orchestration: evaluates outcomes
- FastAPI application → Runtime telemetry: exports signals
README.md - Public showcase → Analyst / reviewer: presents walkthrough
README.md