Introduction

The "8. NLP Grouped by Type" report from the "CDR" report group allows you to consolidate the credit consumption of AI/NLP components used in wolkvox Studio, grouping them by component type (e.g., routers, classifications, summaries, cache, copilot, QA, etc.). It is useful for auditing consumption, detecting which components are demanding the most credits, comparing trends between components, and supporting decisions for optimizing flows (routing points) and cost control.

Report Information

• TYPE_NLP: Indicates the type of AI component/functionality that generated consumption in the wolkvox Studio routing points. This column may display different values depending on the operation's configuration; in the example image, some of the values observed include: nlp-MrWizard-chat (MrWizard component in chat), nlp-MrWizard-voice (MrWizard in voice), getvars (variable retrieval/reading), model-router (routing/route selection), nlp-wolkvox-inten (intent detection), nlp-classifications (classifications), nlp-text-summary (text summary), nlp-voice-summary (voice summary), nlp-voc-autoqa (automated QA/VOC), nlp-Copilot-default (default copilot), nlp-cache (cache layer), conversations-chat-61074 / conversations-chat-61091 / conversations-chat-61083 (conversation modules/instances for chat identified by suffix), their variants conversations-chat-61074-cache / conversations-chat-61091-cache (cache associated with that instance), conversations-voice-61267 (conversation module/instance for voice), chatbot (bot component), nlp-QAi (AI-assisted QA), webhook (execution/consumption associated with webhook integrations).
• NLP_AMOUNT: Number of events/executions recorded for that TYPE_NLP within the queried range. Useful for assessing the volume of component usage (not cost; it's volume).
• CREDITS_SEND: Total credits consumed in "send/request" actions associated with that component (e.g., when the flow "triggers" an evaluation, query, or process).
• CREDITS_ANSWER: Total credits consumed in the "response/result" of the component (e.g., the return of the model/process once executed).
• TOTAL_CREDITS: Total sum of credits for the component in the period: CREDITS_SEND + CREDITS_ANSWER. This is the key column for prioritizing optimization and detecting the types with the greatest impact on consumption.