Skip to content

Search is only available in production builds. Try building and previewing the site to test it out locally.

Functional architecture

This diagram shows, on a single page, how Graphicient turns raw data into answers. Read it from the bottom to the top: data comes in at the bottom and reaches the users at the top.

Functional architecture of Graphicient

How to read the diagram

The diagram has three columns:

  • Data value chain (left) — the 4 stages the data goes through: Feed, Ingest, Compute and Access
  • Functional architecture (center) — the building blocks of the platform
  • Tech stack (right) — the technologies used

The four stages

StageWhat happensBuilding blocks
FeedYour data is collected where it already livesDatasources (databases, files) and Datalakes (Databricks, Snowflake)
IngestThe data is structured by the ontology and loaded into a graphOntology & ETL and Graph DB & versioning DB
ComputeThe platform processes the graphCore app and App modules (BRE, LLM, Impact, KPI)
AccessUsers get the resultsDashboard & chat with LLM and Use case factory

The building blocks, in simple terms

  • Datasources and Datalakes — where your data comes from: company databases, files, or data platforms such as Databricks and Snowflake
  • Ontology & ETL — the ontology describes your business domain; the ETL (extract, transform, load) takes the data from the sources and loads it into that structure
  • Graph DB & versioning DB — the graph database stores your knowledge graph; the versioning database keeps track of its history
  • Core app — the heart of the platform, on which all modules rely
  • App modules — the specialized features: BRE (business rules), LLM (the AI that understands your questions), Impact (impact analysis) and KPI (indicators)
  • Dashboard & chat with LLM — the screens used day to day: dashboards to explore the graph, and a chat to ask questions in natural language
  • Use case factory — ready-to-use business use cases (UC) built on top of the platform

The tech stack in short

  • React — the user interface
  • Mistral AI — the language model
  • Python — the core app and the data processing
  • Neo4j and MongoDB — the databases
  • AWS — the hosting

Where you find it in Graphicient

The four building blocks of the platform cover these layers: GraphForge for the ontology and the data loading, Explorer for the dashboards, KPI Forge for the indicators, and the AI Chatbot for the chat.