# Empowering Users to Get More Done With Agent Mode > Source: > Published: 2026-06-04 15:20:52+00:00 # Empowering Users to Get More Done With Agent Mode The future of AI is not single-modality chat; it is in powerful agentic capabilities. Today, we are excited to introduce [ Agent Mode](https://arena.ai/agent?ref=arena.ai), designed to help everyone from everyday users looking to get more done to entrepreneurs looking to maximize agentic efficacy across complex use cases. While traditional chat requires you to break a complex task into numerous prompts, Agent Mode autonomously builds a plan and uses its built-in tools to accomplish the entire multi-step workflow in one go, like building a website or running deep research. Now you can automate tasks without having to switch modalities or require additional prompting, in a powerful sandbox testing environment. The release of Agent Mode also reinforces our core promise to be the best place to test the latest AI models, based on real-world use cases. It provides the same simple and intuitive interface tens of millions of people use on [ Arena.ai](http://arena.ai/?ref=arena.ai), increasing the flexibility to test and measure the agentic intelligence of different model labs. Agent Mode offers a suite of advanced tools, including web search, image generation, coding and technical assistance, file attachments, and a powerful sandbox/bash environment for testing and iteration. Arena now unlocks a whole new set of use cases beyond basic chat. Instead of breaking complex tasks into numerous isolated prompts, Agent Mode is smart enough to accomplish tasks on its own to get a job done, with minimal follow-ups from the user. With this new release we also offer a new leaderboard methodology geared towards understanding the performance of multi-component agents based on organic user traces. **Agent Mode for Users: Complex Tasks, Simplified** Agent Mode provides value to both everyday users and entrepreneurs by offering a single, powerful tool to accomplish complex tasks and get more work done. By combining research, coding, and analysis into one connected experience, users can tackle ambitious projects—like building a small business website, running deep research, or planning a product launch—with significantly less friction and manual coordination. It also means being able to put in one prompt to execute a multi-stage workflow, rather than continuing to have to re-prompt, saving time and effort to get to the answer or deliverable faster. We have seen our community use Agent Mode in a myriad of different ways, taking advantage of some of the most critical workflows being addressed. The specific breakdown we’ve seen so far is as follows: **Coding Domination (29%):** The task distribution is dominated by coding, which makes up a cumulative 29% of the task mix.**Research & Planning (11% each):** The next most popular categories belong to research-based tasks including looking up information, planning or brainstorming.**Workflow Automation (4%):** Workflow automation only makes up 3.9% of tasks, yet represents a promising category for generating value for end users.**Remaining long-tail use cases:** There's a number of long tail use cases, including data analysis, translation, and media analysis. Another key insight is the levels of agency we see users giving the agents. The majority of initial messages index on delegation of certain tasks rather than handing over complete control. We see this trend continue for follow-on messages, where twice as many users end up tightening controls rather than loosening them, essentially treating the model like an employee, and taking more of a “hands-on” manager-type role. We’re excited to see the evolution of the user/agent relationship, especially as the models continue to evolve and improve, and as agentic workflows move from early adopters to the mainstream. **Measuring Real-World Utility** Agent Mode now answers the urgent question of how frontier models perform in agentic, real-world contexts. Agent Mode provides the signals of real-world utility into how various models handle multi-step workflows — something that is increasingly critical in today’s AI landscape. The data generated from all of these real-world tasks will power a public [ agent leaderboard](http://arena.ai/leaderboard/agent?ref=arena.ai), providing a new standard for measuring AI advancement. We’re excited to have our community contributing to the leaderboard via the signals generated from their real-world agentic usage using the leading AI models. For more information, read our [that explores the methodology behind the rankings.](https://arena.ai/blog/agent-arena-methodology) __Agent Arena Leaderboard blog post__**Getting Started** The release of Agent Mode establishes a rigorous new standard for measuring intelligence and accuracy in AI—giving everyone the tools to evaluate and understand true agentic capabilities head-to-head. Join us in defining how the next generation of AI is benchmarked at [ arena.ai/agent](http://arena.ai/agent?ref=arena.ai). **Frequently Asked Questions** With this launch, we wanted to highlight a few of the common questions that may arise: **How do I access Agent Mode?** From the Arena homepage at [ Arena.ai](http://arena.ai/?ref=arena.ai), click on the dropdown menu item in the top left, and change the option from the default “Battle Mode” to “Agent Mode.” You can also access it directly at [.](http://arena.ai/agent?ref=arena.ai) __arena.ai/agent__**What is different about Agent Mode compared to the other modes?** The current chat experiences of Battle, Direct, and Side-by-Side are limited to isolated, single-modality interactions. Agent Mode changes that by introducing a more dynamic workflow experience that helps shoulder the manual work for you. **What built-in tools does Agent Mode use?** Agent Mode has access to a powerful suite of tools including web search, image generation, file upload, coding assistance, and a sandbox/bash environment that gives the agent the autonomy to execute tasks. Additional tools and functionality will continue to be added over time. **When should I use Agent Mode vs other modes?** Agent Mode combines research, coding, analysis, and iteration into one connected experience, allowing you to move fluidly between tasks without losing context. This means if you have a more complex, multi-step workflow, Agent Mode would be an ideal fit. With Agent Mode you can complete tasks such as building and launching a small business website, planning a product launch, or doing deep research into a topic of your choosing. Conversely, if you have simple tasks that could be accomplished via chat functionality, Agent Mode likely isn’t necessary. Anything regarding simple research, search, or basic knowledge queries that are single step are better suited for other modes such as Battle Mode, Side-by-Side, or Direct. **What is the Agent Leaderboard and why is it important?** The [ Agent Arena leaderboard](http://www.arena.ai/leaderboard/agent?ref=arena.ai) is the first agentic evaluation built entirely from live behavioral signals: turn-by-turn natural language feedback, explicit task success labels given by users, artifact download events, and other user behavior signals. This feedback is directly aggregated from millions of turns from hundreds of thousands of real agentic workflow sessions from Agent Mode, with no curated prompts or paid evaluators. The result is a wholly new class of evaluations: direct evidence of how agents perform in the hands of real people in the natural course of their work, not in controlled conditions. The Agent Arena leaderboard tells us the downstream utility of AI agents to the people using them. **For more information on Agent Mode, check out our **__Help Center__