NLP Chatbot for SAP Dashboards
Budget: $250 – $750 USD
I need an NLP layer that sits inside SAP Business One as well as the web-based dashboards we publish through Power BI, Tableau, QlikView and other vendors. The goal is simple: let any employee type or speak a plain-language question—“What is the current month’s inventory level for Product A?”—and receive an immediate, trusted answer pulled straight from our underlying databases.
Key points of the build
• Natural Language Processing is the heart of the product. It must understand everyday business phrasing, map it to the correct tables or views, generate the SQL (or other query language) on the fly, and return concise results or visual summaries.
• Direct, seamless integration with SAP B1 is mandatory. The chat panel should feel native, obey user permissions already defined in SAP, and leave no footprint that could jeopardise upgrades.
• The same component must be embeddable in multiple BI front-ends. If the dashboard vendor supports custom visuals or HTML containers, the chatbot should drop in with minimal configuration and call back to the same inference engine.
• Responses need to be fast and auditable. Each answer should show the query it executed, so power users can validate logic and compliance teams can trace data lineage.
• Security, SSO and role-based access are non-negotiable; any exposed API must authenticate through our existing identity provider.
Deliverables
- Embeddable chat widget (SAP B1 add-on + JavaScript component for dashboards)
- Backend NLP service with documentation, source code and deployment scripts
- Connection layer for SAP B1 schemas and at least one BI tool to prove portability
- Test suite covering sample questions, edge cases, and performance benchmarks
- Handover session so our in-house IT team can extend the intent library
The first milestone will be a working proof-of-concept answering inventory and sales queries in a sandbox system. From there we will iterate on additional intents, security hardening and UI polish.
Key points of the build
• Natural Language Processing is the heart of the product. It must understand everyday business phrasing, map it to the correct tables or views, generate the SQL (or other query language) on the fly, and return concise results or visual summaries.
• Direct, seamless integration with SAP B1 is mandatory. The chat panel should feel native, obey user permissions already defined in SAP, and leave no footprint that could jeopardise upgrades.
• The same component must be embeddable in multiple BI front-ends. If the dashboard vendor supports custom visuals or HTML containers, the chatbot should drop in with minimal configuration and call back to the same inference engine.
• Responses need to be fast and auditable. Each answer should show the query it executed, so power users can validate logic and compliance teams can trace data lineage.
• Security, SSO and role-based access are non-negotiable; any exposed API must authenticate through our existing identity provider.
Deliverables
- Embeddable chat widget (SAP B1 add-on + JavaScript component for dashboards)
- Backend NLP service with documentation, source code and deployment scripts
- Connection layer for SAP B1 schemas and at least one BI tool to prove portability
- Test suite covering sample questions, edge cases, and performance benchmarks
- Handover session so our in-house IT team can extend the intent library
The first milestone will be a working proof-of-concept answering inventory and sales queries in a sandbox system. From there we will iterate on additional intents, security hardening and UI polish.