Time-Waster Detection for Companion & Conversational AI Agents (human-verified micro dataset)
The Ultimate Fraud Detection Dataset: Perfect for AI builders, dataset engineers, cyber security professionals and conversational AI developers.
Version updated 22nd May 2025.
For a limited time only, enjoy promotional pricing at $199, this dataset is easily valued between $999 - $1,999 dependant on industry application and use case- get in early before our prices increase.
This human-verified, micro conversational dataset is designed to train AI agents and LLMs to recognize when a user is wasting tokens or about to churn without converting. It is ideal for fine-tuning routing models, building companion AI moderation systems, or enhancing customer service chatbots and agent networks as well as identifying fraudulent behaviour before it takes root and causes any damage.
The dataset captures the full behavioral flow:
- User engagement
- Micro-escalation detection
- Repeated payment refusals
- Meet-up pressure / disengagement patterns
- Structured agent intervention (Soft Exit + Hard Block signals)
Includes:
- Fully annotated CSV dataset
- Clean conversational flow for model training
- Companion AI + customer service use case ready
- Human-verified labels + categories
This is the first Gumroad release from Life Bricks Data’s human-agent conversational dataset collection, designed to support next-gen AI agent builders.
Format: CSV&JSON + Dataset README + Annotation Guide & Data Dictionary.
Size: 167 text entry messages (micro dataset high value)
License: Commercial license for dataset integration & fine-tuning experiments
Disclaimer: This dataset is anonymized and contains no personal identifying information. It is intended for professional AI research, dataset augmentation, and model fine-tuning use cases. The dataset is from the companion service industry and as such our best attempts have been made to annotate [EXPLICIT] content wherever possible as long as it does not affect the flow and underlying context of the conversation.
Human-verified micro dataset to train AI agents & chatbots to detect time-wasters before they burn tokens. Includes full conversational flows, escalation points, and Soft Exit / Hard Block labels.