Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Kevin2523/nextAuditAi/llms.txt

Use this file to discover all available pages before exploring further.

Flowise is the visual AI composition layer in NextAudit AI. It lets you build LLM-powered chains and autonomous agents through a drag-and-drop interface, connecting to the self-hosted Ollama inference service and storing vector embeddings in the pgvector-enabled PostgreSQL instance. This keeps all AI computation and data within the stack, with no dependency on external AI providers.

​
Service configuration

Flowise runs from the flowiseai/flowise:latest image. Both the host and container ports are set to FLOWISE_PORT, so the container listens on the same port number it exposes. 
flowise:
  image: flowiseai/flowise:latest
  container_name: flowise
  ports:
    - "${FLOWISE_PORT}:${FLOWISE_PORT}"
  volumes:
    - flowise_data:/root/.flowise
  depends_on:
    postgres:
      condition: service_healthy
  restart: unless-stopped
Flowise waits for the postgres service to pass its health check before starting, ensuring the database schema is available when Flowise initializes.

​
Environment variables

environment:
  PORT: ${FLOWISE_PORT}
  DATABASE_SCHEMA: ${DATABASE_SCHEMA}
  DATABASE_TYPE: postgres
  DATABASE_HOST: postgres
  DATABASE_PORT: 5432
  DATABASE_NAME: ${POSTGRES_DB}
  DATABASE_USER: ${POSTGRES_USER}
  DATABASE_PASSWORD: ${POSTGRES_PASSWORD}
VariableValueDescription
PORT${FLOWISE_PORT}Port Flowise listens on inside the container
DATABASE_TYPEpostgresSelects the PostgreSQL driver
DATABASE_HOSTpostgresDocker service name of the PostgreSQL container
DATABASE_PORT5432Standard PostgreSQL port
DATABASE_NAME${POSTGRES_DB}Database name created during PostgreSQL initialization
DATABASE_SCHEMA${DATABASE_SCHEMA}PostgreSQL schema Flowise uses for its tables
DATABASE_USER${POSTGRES_USER}Database user with read/write access
DATABASE_PASSWORD${POSTGRES_PASSWORD}Credential for the database user

​
Volume

volumes:
  - flowise_data:/root/.flowise
The flowise_data volume stores uploaded files, flow definitions saved locally, and any configuration that Flowise writes to disk. The database-backed metadata (flows, credentials, chat history) lives in PostgreSQL.

​
pgvector integration

The PostgreSQL instance in this stack is built with the pgvector extension enabled. Flowise uses this to store and query vector embeddings generated during document ingestion and retrieval-augmented generation (RAG) flows. The EMBEDDING_SIZE environment variable on the PostgreSQL service controls the vector dimension — this must match the embedding model you configure inside Flowise.
When building RAG flows, use the Postgres vector store node in Flowise and point it at DATABASE_HOST=postgres. The pgvector extension handles similarity search directly in SQL, keeping embedding retrieval within the stack.

​
AI flow types for audit

Q&A chains

Build question-answering flows over ingested audit documents, policy PDFs, or FleetDM inventory exports. Users or n8n workflows can query the knowledge base in natural language.

Document analysis

Run LLM chains over osquery result sets or vulnerability reports to extract structured findings, classify severity, and generate human-readable summaries.

Alert triage

Create agent flows that receive a raw alert from n8n, look up context from the vector store, and return a triage decision with a recommended action.

Policy compliance

Chain FleetDM policy results through an LLM to explain non-compliance in plain language and suggest remediation steps tailored to the specific host configuration.

​
Connecting to Ollama

Inside the Flowise UI, add an Ollama chat model node and set the base URL to http://ollama:11434. This uses Docker’s internal DNS to reach the Ollama service directly without going through the host network. See Ollama: self-hosted LLM inference for the full Ollama service configuration.

​
Accessing the Flowise UI

The Flowise editor is available at: 
http://<host>:${FLOWISE_PORT}
From the UI you can create new flows, manage credentials, configure vector store integrations, and monitor chat sessions. Flows can also be triggered programmatically via the Flowise REST API, which n8n uses to call AI flows as part of larger audit pipelines.

Build docs developers (and LLMs) love

Get started for freeTalk to us