TOHOL
TOHOLSolutions
GL3 + GL4 · Operational foundation

From Operational Data to AI-Ready Knowledge

Connect the operational data acquired through GL3 with the GL4 preparation pipeline on HERCULES: curation, documents, chunks, vectors and semantic validation.

Explore the workflow Discuss this use case

An additional application scenario based on the CIRMS operational architecture. Scope and supported integrations need to be confirmed.

The challenge

Build the knowledge foundation before adding AI.

Collecting data is not the same as preparing it for AI. This additional application scenario focuses on CIRMS's GL3/GL4 foundation, keeping data preparation separate from the planned GL5 models and conversational features.

Information in scope

Registered sources
Supported sources selected for the deployment.
Source metadata
Structure, identity and change information.
Operational records
Data acquired through the agreed GL3 processes.
Preparation requirements
Scope, permissions and intended knowledge use.
From information to a reviewed decision

How the workflow is organized.

An application-level outline, not a claim that every source, connector or operational procedure is preconfigured.

  1. 01

    Understand the source

    Refresh source metadata and identify the extraction scope.

  2. 02

    Acquire and curate

    Bring data into the preparation process and apply the configured curation rules.

  3. 03

    Prepare knowledge assets

    Generate documents, chunks and vector representations through GL4.

  4. 04

    Validate readiness

    Review semantic validation and readiness status before downstream AI consumption.

Visible preparation

Understand how source records become knowledge assets.

Reusable knowledge

Prepare structured assets for the intended downstream use.

Separate responsibilities

Keep data readiness distinct from AI reasoning and user-facing answers.

Architecture in context

A practical architectural foundation.

The GL4 preparation path on HERCULES

  1. Stage 0Source metadata refresh
  2. Stage 1Raw data
  3. Stage 2Data curation
  4. Stage 3Document generation
  5. Stage 4Chunk generation
  6. Stage 5Vector embedding
  7. Stage 6Semantic validation
  8. Stage 7Readiness completion

An architecture-level summary, not live execution data. Review the configured readiness criteria and outputs before using the knowledge in an AI application.

Explore the preparation concept
Define the boundaries

What sits behind the use case.

Separate the platform foundation from planned intelligence and the requirements of the target environment.

Operational foundation

GL3 / GL4 & architecture

HERCULES performs GL3 automation and acquisition as well as GL4 AI Data Readiness. This page describes that documented foundation, not a new GL5 feature.

Planned or additional scope

AI & future extensions

AI models, conversational answers and decision assistance consuming the prepared knowledge belong to the planned GL5 layer and require separate evaluation.

Agree before deployment

People, data & permissions

Confirm supported source types, identity/change handling, access permissions and the readiness criteria. A passed preparation stage alone does not guarantee a correct or safe AI answer.

A bounded evaluation

Register a supported, representative source and review the preparation outputs and execution status with the data owner before increasing scope.

Candidate measures: source coverage, preparation failures, knowledge freshness and semantic-validation results against the intended use case.