Q4 Milestone 3 - Use LLM to automatically extract data for client onboarding - Phase 4: Model Integration

[matildemaccio]

Objective:

Implement a new endpoint in the API that leverages a Language Model (LLM) to extract and structure onboarding data from user inputs. The endpoint should process unstructured text, generate structured data, and save it in the database.

---

Tasks:

1. API Endpoint Development:

   • Create a new API endpoint that accepts raw user input related to onboarding (e.g., location, capacity, installation date).
   • Ensure the endpoint interacts with the LLM API to process the input text and extract the relevant data.

2. Model Integration:

   • Integrate the selected Language Model (LLM) API into the system.
   • Pass user input text to the LLM API and retrieve the structured data response.
   • Ensure the API correctly interprets and extracts details such as location, capacity, and installation date.

3. Database Interaction:

   • Map the structured JSON data to the corresponding database fields.
   • Implement logic to save the extracted onboarding data into the database.
   • Ensure proper error handling in case of incomplete or ambiguous data, preventing database inconsistencies.

4. Testing and Validation:

   • Write unit tests to verify that the API endpoint correctly calls the LLM, structures data, and interacts with the database.
   • Perform integration tests to ensure the entire flow, from raw input to database storage, works seamlessly.
   • Handle edge cases where the LLM may fail to extract certain data or produce unexpected results.

---

Deliverables:

• A fully functional API endpoint that processes onboarding-related user input and saves the extracted data into the database.
• JSON schema for the structured data generated by the LLM.
• Test coverage for the new functionality, ensuring robust error handling and integration success.