Terminal Agents Suffice for Enterprise Automation
Patrice Bechard, Orlando Marquez Ayala, Emily Chen, Jordan Skelton, Sagar Davasam, Srinivas Sunkara, Vikas Yadav, Sai Rajeswar
2026-04-02
Summary
This paper investigates whether really complex AI agents are actually needed to automate tasks businesses do, or if simpler agents can be just as good, or even better.
What's the problem?
Currently, there's a trend towards building very sophisticated AI agents that use things like special protocols and can even interact with websites like a person. However, these agents are expensive to build and run, and it's not clear if all that complexity is truly necessary to get the job done. The core question is: are these complicated systems actually providing a significant benefit over simpler approaches?
What's the solution?
The researchers tested a different approach: they created AI agents that only have access to a basic computer terminal and the ability to read and write files. These agents interact with business software directly through the software's underlying programming interface, instead of trying to mimic a human user. They then compared how well these 'terminal agents' performed on real-world business tasks against the more complex agent types.
Why it matters?
The findings suggest that you don't always need a super-advanced AI to automate tasks. A simpler agent, combined with a powerful AI model, can often achieve the same or better results. This is important because it means businesses could potentially save money and effort by using less complex automation solutions, making AI more accessible and practical for everyday use.
Abstract
There has been growing interest in building agents that can interact with digital platforms to execute meaningful enterprise tasks autonomously. Among the approaches explored are tool-augmented agents built on abstractions such as Model Context Protocol (MCP) and web agents that operate through graphical interfaces. Yet, it remains unclear whether such complex agentic systems are necessary given their cost and operational overhead. We argue that a coding agent equipped only with a terminal and a filesystem can solve many enterprise tasks more effectively by interacting directly with platform APIs. We evaluate this hypothesis across diverse real-world systems and show that these low-level terminal agents match or outperform more complex agent architectures. Our findings suggest that simple programmatic interfaces, combined with strong foundation models, are sufficient for practical enterprise automation.