An Ataccama RDM implementation starts with business requirements and results in an RDM configuration. Throughout the implementation, consultants produce numerous artefacts, including validation rules, migration plans, test cases and technical documentation. Although these artefacts serve different purposes, they all describe the same solution.
Traditionally, creating and maintaining these artefacts requires navigating business requirements, XML configuration files and project documentation. Large Language Models (LLMs) can accelerate this process by generating implementation artefacts and analysing the completed configuration.
The following sections demonstrate practical examples of how AI can support different stages of an Ataccama RDM implementation.
AI Support Across the RDM Implementation Lifecycle
| № | Project stage | Project task | Input | Coverage | AI support |
|---|---|---|---|---|---|
| 1⃣ | Design | Validation Rule and Lookup Design | Business requirements, data model, sample data, DQ rule templates | Full draft | Generate validation rules, recommend validation type |
| 2⃣ | Design/ Migration | Bulk Import/Export Plan Generation | Data model description, single-table import/export plan templates | Full draft | Generate bulk import/export plans by combining the data model with existing plan templates. |
| 3⃣ | Implementation | Component Configuration (e.g. XML Reader step) | Existing RDM XML files | AI support | Generate XML Reader step configurations and extract metadata for consolidated views of datasets, validation rules, workflows, permissions and relationships. |
| 4⃣ | Testing | Functional and Data Quality testing. | Business requirements, parsed configuration, DQ rule templates, primary key definitions, sample data | Full draft | Generate positive and negative test cases, CSV import files and expected validation results. |
| 5⃣ | Documentation | Documentation & Knowledge Transfer | Business requirements, parsed configuration, project decisions | AI support | Generate DSD sections, migration strategies, deployment and rollback runbooks |
| 6⃣ | Documentation | Architecture Diagrams & Templates | RDM configuration, diagram conventions, existing diagrams/templates (optional), draw.io XML template (optional) | Full draft | Generate editable draw.io XML templates, PNG architecture diagrams, workflow diagrams |
| 7⃣ | Testing / Support | Troubleshooting & Root Cause Analysis | Parsed RDM configuration, import/export logs, workflow logs, error messages | AI support | Analyse logs and parsed configuration, identify root causes, explain validation and workflow failures, recommend corrective actions, and suggest configuration improvements. |
Coverage:
- Full draft – AI generates a complete first draft of the deliverable, requiring consultant review and refinement before implementation.
- AI support – AI assists with analysis, implementation or documentation tasks but does not generate a complete deliverable.
1. Validation Rule and lookup Design
Input
- Validation Rule Reference – existing RDM validation examples.
Expression
Enabled
Error Message
Attribute
matches('^AR[0-9]{6}$', ACCS_ROLE_CD) OR ACCS_ROLE_CD IS NULLtrue
Invalid Access Role Code. Format mismatch: expected AR followed by 6 digits.
ACCS_ROLE_CD
-
Implementation Guidelines
-
-
Sample Data – representative records from the reference dataset
| Dataset Name | Category Code | Category Name | Effective From | Effective To | Status |
|---|---|---|---|---|---|
| Product Category Codes | PC000101 | Consumer Electronics | 01/01/2025 | 31/12/2899 | Active |
| Product Category Codes | PC000102 | Home Appliances | 01/01/2025 | 31/12/2899 | Active |
| Product Category Codes | PC000103 | Office Supplies | 01/01/2025 | 31/12/2899 | Pending |
| Product Category Codes | PC000104 | Outdoor Equipment | 01/01/2025 | 31/12/2899 | Inactive |
| Product Category Codes | PC000105 | Personal Accessories | 01/01/2025 | 31/12/2899 | Active |
Prompt
Analyse the sample data using the supplied DQ rule templates and the example validation rule.
Generate:
- lookup recommendations;
- validation expressions following the provided template;
AI-Generated Output
Suggested Lookup Relationships
| Field | Recommended Lookup Dataset | Allowed Values |
|---|---|---|
| Status | Status Codes | Active, Inactive, Pending |
Suggested Validation rules
| Expression | Enabled | Error Message | Attribute |
|---|---|---|---|
CATEGORY_CD_STRT_DT <= CATEGORY_CD_END_DT OR CATEGORY_CD_STRT_DT IS NULL OR CATEGORY_CD_END_DT IS NULL /* Ensures that the start date is not later than the end date when both values are provided (non-empty). */ | true | Invalid Effective From. Date order violation: must not be after Effective To. | CATEGORY_CD_STRT_DT |
CATEGORY_CD_STRT_DT <= CATEGORY_CD_END_DT OR CATEGORY_CD_STRT_DT IS NULL OR CATEGORY_CD_END_DT IS NULL /* Ensures that the end date is not earlier than the start date when both values are provided (non-empty). */ | true | Invalid Effective To. Date order violation: must not be before Effective From. | CATEGORY_CD_END_DT |
matches('^PC[0-9]{6}$', CATEGORY_CD) OR CATEGORY_CD IS NULL | true | Invalid Category Code. Format mismatch: expected PC followed by 6 digits. | CATEGORY_CD |
matches('^[A-Za-z0-9()\\- &,]+$', CATEGORY_NM) OR CATEGORY_NM IS NULL | true | Invalid Category Name. Format mismatch: allowed letters, digits, spaces, ampersand (&), parentheses (), comma (,), and hyphen (-). | CATEGORY_NM |
2. Bulk Import/Export Plan Generation
AI can generate complete Bulk Import and Export Plans by combining the implemented data model with existing plan templates.
Input
- Data Model – implemented RDM data model.
- Bulk Import/Export Plan Template – existing project template defining the required structure and conventions.
Prompt
Generate Bulk Import and Bulk Export Plans for the provided RDM data model using the supplied project template.
Reuse the template structure and generate plans for all datasets
AI-Generated Output
| Source Artefact | Generated Artefact |
|---|---|
| 
|
| 
|
3. Component Configuration (e.g. XML Reader step).
AI can assist with configuring RDM components in ONE Desktop by identifying the required configuration parameters for individual processing steps. XML Reader is used here as an example.
Input
- Existing RDM XML files
Prompt
Generate an XML Reader step to extract validation rules from the provided RDM XML configuration.
Include all required XML Reader steps, XPath expressions and output attributes.
AI-Generated Output
| Source Artefact | Generated Artefact |
|---|---|
| 
|
The extracted configuration objects are consolidated into a tabular representation, allowing consultants to analyse validation rules, permissions, workflows and other configuration elements across the entire RDM implementation.
4. Functional and Data Quality testing.
Once the RDM configuration has been parsed, AI can assist with implementation review and test preparation.
Implementation Review
AI can validate the implementation against business requirements and implementation standards.
Input
- Parsed RDM configuration
- Business requirements
Prompt
Review the parsed RDM configuration against the supplied business requirements and implementation standards.
Verify that:
- lookup dataset permissions are assigned to the same roles as the corresponding main datasets;
- all primary key attributes are configured as mandatory;
- all primary key attributes are configured as non-editable;
- workflows are configured for all datasets and include email notifications for create, update and delete operations using the correct email templates;
- email templates are consistent, with matching email subjects and message bodies;
- validation rules are complete and consistent, including standardised validation message formats.
Test Generation
AI can generate test assets based on the parsed RDM configuration.
Input
- Parsed RDM configuration
- Business requirements
- Sample data
Generate functional test assets for the supplied RDM implementation based on the parsed configuration and business requirements.
Prompt
Generate import-ready CSV files for functional and Data Quality testing of the supplied RDM implementation.
- positive test records;
- negative test records;
- import-ready CSV files for each test scenario.
AI-Generated Output
| Category Code | Expected Result | Validation Result |
|---|---|---|
| PC000103 | Valid | Record passed all validation checks and is expected to be imported successfully. |
| PC123 | Invalid | Category Code must consist of the prefix PC followed by exactly 6 digits. Only 3 digits were provided. |
| PC000106 | Invalid | Category Name contains unsupported characters (@). |
| PC000107 | Invalid | Effective From (01/01/2026) is later than Effective To (31/12/2025). |
| PC000108 | Invalid | Effective To (31/12/2024) is earlier than Effective From (01/01/2025). |
5. Documentation & Knowledge Transfer
The parsed configuration can be used to generate project documentation directly from the implementation.
Input
- Parsed RDM configuration
- Ataccama documentation template for the selected module
- Customer documentation template or documentation from a previous project
- Project decisions
- Official Ataccama documentation
Prompt
Generate the documentation for the supplied RDM implementation using the provided documentation templates.
- Data Solution Design (DSD)
- Migration documentation
- Deployment and rollback runbooks
- Knowledge articles
- Technical summaries
Because these artefacts are generated from the implemented configuration, they remain consistent with the solution.
6. Architecture Diagrams & Templates
AI can generate technical diagrams and reusable templates directly from the implemented RDM configuration.
Input
- Parsed RDM configuration
- Existing project diagrams (optional)
- Diagram conventions
- draw.io XML template (optional)
Prompt
Generate a PNG diagram illustrating the project stages and AI support throughout the RDM implementation lifecycle.
Generate an editable draw.io XML diagram using the supplied draw.io template.
AI-Generated Output
- PNG diagrams
- draw.io XML templates
7. Troubleshooting & Root Cause Analysis
AI can analyse RDM configurations together with import/export logs and workflow execution logs to identify configuration issues, explain validation failures and recommend corrective actions.
Input
- Import and export logs
- Workflow execution logs
- Plans
Prompt
Identify:
- the root cause of each reported issue;
- configuration inconsistencies;
- recommended corrective actions.
AI-Generated Output
| Finding | Recommendation | Expected Benefit |
|---|---|---|
| Repeated Alter Format transformations | Merge compatible transformations into fewer steps where possible. | Improved readability and fewer processing stages. |
| XML Reader extracts multiple object types independently | Consider extracting shared XML structures once and reusing intermediate outputs. | Reduced XML parsing overhead. |
| Generic step names (e.g. Join 3, Alter Format 2) | Rename steps to reflect their business purpose. | Easier troubleshooting and maintenance. |