Governed natural-language access to enterprise ERP information.
AJSMGPT translates supported business questions into controlled, validated Oracle ERP queries—without giving a language model unrestricted SQL access.
- Role
- Independent system design and development
- Environment
- Internal enterprise ERP workflow
- Stack
- FastAPI · Oracle · Qdrant · LLMs
Response time is a performance target. Organisation-wide time and cost savings have not yet been formally measured.
ERP information was available, but not naturally accessible.
Employees often depend on manual ERP navigation, predefined reports, or technical SQL support to retrieve operational information.
A general-purpose model generating unrestricted SQL would create unacceptable accuracy and security risks. The real challenge was building useful natural-language access while keeping behaviour controlled, testable, and reversible.
Natural-language interaction around deterministic controls.
The model assists interpretation; verified routes and database safeguards govern execution.
- 01
Business question
A user asks for operational information in everyday language.
- 02
Controlled routing
Supported intent and parameters are matched to verified logic.
- 03
Validated query
Read-only Oracle access runs through constrained query patterns.
- 04
Safe answer
Results are formatted for the user with traceable system behaviour.
A safe answer carries evidence of the route behind it.
purchase_last_n_by_materialParameters validated · Read-only access · Row limit applied
Sanitized demonstration values would appear here.
Route approved · Query constrainedThis demonstration contains no client queries, supplier identities, database values, or production information.
New behaviour earns its way into production.
Publishing is a separate human-controlled action, with versions, audit history, disable support, and rollback.
- 01Open
- 02Candidate generated
- 03Tested
- 04Human approved
- 05Regression passed
- 06Ready to publish
- 07Published
The assistant is useful because its freedom is intentionally limited.
Read-only database access
The system retrieves information; it cannot modify operational ERP records.
Verified query routes
Supported questions use reviewed tables, joins, filters, and calculations.
Regression protection
Approved behaviours are retested before new versions can be published.
Human release control
Testing does not equal publishing; an authorized person makes the final release decision.
Independent engineering across AI, enterprise data, evaluation, and release.
I designed and developed the system independently. Domain users supported business-rule validation and query-correctness feedback.
Controlled scope is part of the design.
- Only supported and approved question patterns should be treated as reliable.
- New business rules require domain validation, testing, and controlled publishing.
- The under-three-second figure is a target, not a universal production guarantee.
- Public examples are sanitized; client identity, schema details, and production data remain confidential.
Need trustworthy AI access to complex business information?
I’m available for selected enterprise-assistant, retrieval, and governed-AI projects.