This article explains how Tekst uses API calls with SAP Service Cloud to automate case processing and train AI models. Understanding the API flow helps you estimate resource usage and plan your integration.
Real-Time Case Processing Flow
When Tekst processes cases in real-time through SAP Service Cloud, the following sequence of API calls occurs for each case:
1. Case Creation Notification
SAP Service Cloud sends a webhook notification to Tekst when a new case is created. This initial ping triggers the automation workflow.
2. Automation Execution (PATCH Request)
After receiving the notification, Tekst performs its AI-driven automation and updates the case by patching specific SAP fields. This is typically a single PATCH API call that updates fields such as:
- Category or subcategory classifications
- Priority levels
- Routing or assignment information
- Custom fields for sentiment, product identification, or other AI predictions
- Language detection
For a complete list of fields Tekst can update, refer to SAP possible actions and Objects and fields used.
3. Case Closure Notification
When a case is closed in SAP Service Cloud, another webhook notification is sent to Tekst. This signals that the case lifecycle is complete.
4. Feedback Retrieval (GET Request)
After closure, Tekst fetches the complete case data to check if any relevant fields have changed. This GET request retrieves the case with all updated field values to support Tekst's feedback mechanism, which:
- Compares AI predictions with final agent decisions
- Identifies correction patterns
- Feeds data back into the AI training pipeline to improve accuracy
This feedback loop is what enables Tekst models to continuously learn and achieve 90%+ accuracy in routing and classification over time.
API Call Summary per Case
For each case processed in real-time, expect the following API interactions:
The total API calls from Tekst to SAP per case is 2 calls (1 PATCH + 1 GET). SAP sends 2 webhook notifications to Tekst per case (create + close). Plan your API rate limits accordingly based on your expected case volume.
Additional API Operations
-
Bulk Operations for Model Training
During initial setup or model retraining, Tekst uses bulk APIs to fetch historical messages and cases from SAP Service Cloud. The volume of API calls depends on:
- The number of historical cases being analyzed
- The date range specified for training data
- The complexity of your case structure (interactions, attachments, etc.)
Bulk fetching during training typically uses batch endpoints or pagination to minimize the total number of API requests. Contact your Tekst customer success team to understand the expected API usage for your specific training dataset.
-
Message Review via Tekst Platform
When users access the Tekst platform at customer.tekst.com to review messages and AI predictions, Tekst fetches case data using standard GET requests. These API calls are triggered on-demand when:
- Users navigate to specific messages or cases
- Users filter or search through cases
- Users view prediction details and related case information
These GET requests are minimal and occur only when users actively interact with the platform to review individual cases.
Optimizing API Usage
To optimize API consumption with your SAP Service Cloud integration:
- Scope your webhooks: Configure SAP Service Cloud Flows to send notifications only for relevant case types or statuses
- Monitor rate limits: Ensure your SAP API limits can accommodate your expected case volume plus bulk operations
- Plan training windows: Schedule model training during off-peak hours to minimize impact on real-time operations
- Use custom hooks: Implement fine-grained webhook triggers to avoid unnecessary notifications
For questions about API usage specific to your implementation or to discuss optimization strategies, contact your Tekst customer success team.
0 comments
Please sign in to leave a comment.