NetGent: Agent-Based Automation of Network Application Workflows

TL;DR

NetGent addresses the data-generation bottleneck for networking ML by using natural-language prompts to define abstract, non-linear workflows that are compiled into NFAs and executed via a compile-then-replay pipeline. A State Synthesis component maps prompts to concrete states $\hat{s}= (\textit{detectors}, \textit{code})$, which are cached in a State Repository and replayed deterministically by a State Executor, dramatically reducing token costs and enabling scalable experimentation. The system achieves robustness to UI drift by regenerating only the affected states, and evaluation across 50+ workflows shows diversity, repeatability, and efficiency across domains like video streaming, conferencing, social media, and scraping. NetGent thus provides a scalable foundation for realistic networking datasets, combining language-based synthesis with reliable, reusable execution, and complements existing emulation tools in the field. The work enables more robust ML research in networking by delivering repeatable, diverse traffic traces with reduced reliance on costly LLM interactions.

Abstract

We present NetGent, an AI-agent framework for automating complex application workflows to generate realistic network traffic datasets. Developing generalizable ML models for networking requires data collection from network environments with traffic that results from a diverse set of real-world web applications. However, using existing browser automation tools that are diverse, repeatable, realistic, and efficient remains fragile and costly. NetGent addresses this challenge by allowing users to specify workflows as natural-language rules that define state-dependent actions. These abstract specifications are compiled into nondeterministic finite automata (NFAs), which a state synthesis component translates into reusable, executable code. This design enables deterministic replay, reduces redundant LLM calls through state caching, and adapts quickly when application interfaces change. In experiments, NetGent automated more than 50+ workflows spanning video-on-demand streaming, live video streaming, video conferencing, social media, and web scraping, producing realistic traffic traces while remaining robust to UI variability. By combining the flexibility of language-based agents with the reliability of compiled execution, NetGent provides a scalable foundation for generating the diverse, repeatable datasets needed to advance ML in networking.

Figures (2)

• Figure 1: NetGent runtime loop progresses from the initial to the end state
• Figure 2: ESPN workflow: log into Disney+, select the account, enter the PIN if required, navigate to ESPN, play the first video, and advance the playback slider to the five-minute mark.