AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

ROGA: Scaling Generalist Agents for Office Productivity Tasks via Tool Generation

Mugeng Liu, Xiaojun Ma, Yuhang Xie, Qin Chen, Xuanzhe Liu, Yun Ma

ICLR 2026 · Multi-agent systems and games › LLM agents · approach: Software engineering agent benchmarks

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

The paper argues that reactive automatic tool generation (ATG) agents fail in stateful office environments and proposes ROGA, a tool-generation approach for generalist agents on office productivity tasks.

Problem

Current automatic tool generation agents, built on reactive paradigms, do not adapt well to realistic office environments that need long-term reasoning and stateful interaction.

Approach

Scales generalist agents for office productivity by generating tools automatically, moving beyond reactive tool generation toward long-horizon, stateful adaptation; specific mechanism not stated in the abstract.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use ROGA instead of best ATG baselineGPT-4.1 agents · OSWorld (office tasks)Pass@1: 18.18 -> 31.82Improved
use ROGA instead of best ATG baselineGPT-4.1 agents · WindowsAgentArena (office tasks)Pass@1: 26.19 -> 28.57Improved
use ROGA instead of OWL on GAIA-OfficeGPT-4.1 agents · GAIA-OfficePass@1: 73.08 -> 76.92Improved
add Active World Modeling (ablation)ROGA · OSWorld (office tasks)Pass@1: 20.00 -> 31.82Improved
add Persistent Symbolic Memory (ablation)ROGA · OSWorld (office tasks)Pass@1: 19.09 -> 31.82Improved
add Situated Self-Correction (ablation)ROGA · OSWorld (office tasks)Pass@1: 23.64 -> 31.82Improved
add Capability Lifecycle management (ablation)ROGA · OSWorld (office tasks)Pass@1: 29.09 -> 31.82Mixed
add Capability Lifecycle management (tool reuse) (ablation)ROGA · reasoning stepsPass@1: 1.54x increase in steps without itImproved
add Persistent Symbolic Memory (ablation)ROGA · WindowsAgentArena (office tasks)Pass@1: 21.42 -> 28.57Improved
use ROGA instead of specialized SheetAgentGPT-4.1 · TableBenchPass@1: 51.47 -> 56.09Improved
use ROGA instead of specialized SheetAgentGPT-4.1 · SheetCopilotBenchPass@1: 23.53 -> 25.91Improved
use ROGA on non-office math/knowledge tasksGPT-4.1 agents · Math500 / MMLU-ProPass@1: 70.00 vs 71.00 (OWL) on Math500; 76.00 vs 78.00 (AutoAgent) on MMLU-ProNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy