Synthetic Multi-Turn Conversations for Browser Agent Tasks

1,062 synthetic multi-turn conversations between a user and an AI assistant focus on practical agentic tasks like train ticket booking, dynamic form filling, and payment processing. Created by DataCreator AI, the dataset includes diverse scenarios such as successful execution, context retrieval, tool integration, and failure recovery.

Use Cases

• Train a language model on multi-turn conversation sequences for browser automation tasks like form filling and payment processing.
• Analyze conversation patterns for failure recovery scenarios within agentic workflows.
• Develop and benchmark tool-calling agents using the provided examples of context retrieval and tool integration.

Strengths

• Contains 1,062 synthetic multi-turn conversations, providing a substantial corpus for training.
• Focuses on diverse, practical agentic tasks including train booking, form filling, and payment processing.
• Includes varied scenarios like successful execution, failure recovery, and tool integration.

Limitations

• Dataset is entirely synthetic, which may not fully capture the complexity and noise of real-world user interactions.
• Limited to 1,062 examples, which may be insufficient for training large-scale models without augmentation.
• Specific column structure and features are not detailed, limiting understanding of the data's granularity.

Provenance

Source

DataCreatorAI on Hugging Face.

Collection Method

Synthetically compiled.

Freshness

Last updated 2026-03-22.