Step 1

• Industry-specific OOTB models and scenarios
• Tools for FAQ and customer requests development

Decision making on the list of the intents served by VA, additional data and integration requirements, if any

Step 2

• Additional data (FAQ content, etc.) is provided
• Spitch solution is deployed in the cloud or on premises
• Client’s employees test VA

Pilot operations conclusion. Decision making on the list of intents/scenarios corrections and models fine tuning, if necessary

Step 3

• Models/scenarios fine-tuning
• Spitch solution is integrated with backends, if necessary

Presentation of the results and scale-up plan. Decision making on go to operations

The client can scale up and manage VA

Step 1

• The test data is uploaded
• The AI models are adapted (if necessary)

Assessment of the transcriptions and definition of the evaluation criteria with the customer

Step 2

• The first markers and scripts are demonstrated

The evaluation results are discussed with the customer and adjustments implemented

Step 3

• Fine-tuning for scripts and markers

Presentation of the results and know-how transfer / training for customer’s own adjustment and extensions

The customer can upload and analyze data independently

Step 1

• Stand-alone verification (no integration)
• Embedded verification for agent conversations, (CRM and other integrations required)
• Authentication for self-services

Decision making on the integration architecture and enrolment process

Step 2

• Spitch solution is deployed and integrated for voiceprints enrolment purposes
• The first voiceprints are being created in live conversations or from archive recordings

Step 3

• Models fine-tuning for FAR/FRR targets
• Spitch solution is integrated for verification purposes

Presentation of the results and scale-up plan

The client can scale up and manage the VB system