Dev48
Language
  • About
  • Services
  • Industries
  • Technologies
  • Articles
  • Contacts
Book a call
    Home/Articles/Standing out among more than 30 vendors omdia highlights sensetime raccoon work
Dev48

© 2026 · All rights reserved.

Standing Out Among More Than 30 Vendors: Omdia Highlights SenseTime Raccoon Work as an Enterprise AI Workplace Benchmark Through End-to-End Delivery

Фото: tim zhang (Unsplash) — https://unsplash.com/photos/a-large-building-with-a-sign-on-top-of-it-JEFitDlpAME?utm_source=dev48&utm_medium=referral

Standing Out Among More Than 30 Vendors: Omdia Highlights SenseTime Raccoon Work as an Enterprise AI Workplace Benchmark Through End-to-End Delivery

Source: SenseTime

Recently, Omdia, a globally recognized technology market re earch and con ulting firm, relea ed *Global Enterpri e-Grade Agentic Worker – China Market 2026*, providing a y tematic review of China' enterpri e workplace agent market and more than 30 vendor . Among general-purpo e product , Sen eT…

September 26, 2026

Omdia has established a benchmark for enterprise agents, with SenseTime Raccoon Work emerging as an enterprise AI workplace model through its delivery capabilities.

Recently, a globally recognized technology market research and consulting firm Omdia released "Global Enterprise-Grade Agentic Worker – China Market 2026", providing a systematic review of China's enterprise workplace agent market and more than 30 vendors:

Among general-purpose products, SenseTime Raccoon Work stood out alongside Qianwen Office, Baidu Dazi, and Tencent WorkBuddy as one of the four representative vendors selected for detailed independent analysis in the report.

Evaluated Across Five Dimensions

Raccoon Work Scores Significantly Above the Market Average

Omdia identified Raccoon Work as a representative general-purpose enterprise-grade Agentic Worker vendor in the China market, describing its product evolution as "moving from conversational recommendations toward complete task delivery." Across five dimensions—data and document accuracy, long-chain task reliability, delivery adaptability, native desktop environment integration, and OPC ecosystem operations—Raccoon Work scored significantly above the market average:

OPC ecosystem operations and scaling: 4.9 points, substantially ahead of the market average of 2.1 points. The report specifically highlighted the Level 9 standard pilot, the first batch of Level 6 certificates, more than 8,000 workflows, and events held in over 40 cities. It concluded that Raccoon Work is using standards and certification to turn AI workplace capabilities into a reusable industry ecosystem.

Data and document accuracy: 4.8 points, compared with the market average of 3.3 points. It supports multi-table cleaning, calculation, and parsing, delivering finished analytical outputs rather than raw formulas.

Delivery result adaptability: 4.7 points, compared with the market average of 3.1 points. Output structures can be retained, edited locally, and reused collaboratively, while complex layouts are less likely to break.

Native desktop integration: 4.6 points, compared with the market average of 2.9 points. It supports authorized folders, browser automation, Quick Bar, write-back, and scheduled tasks across desktop, web, and iOS environments, and can directly read more than 20 local file formats.

Long-chain task reliability: 4.5 points, compared with the market average of 2.8 points. It maintains continuity throughout the entire process, from task decomposition and tool invocation to information transfer and final delivery.

From Conversation to Delivery

Raccoon Work Completes the End-to-End Process for Complex Tasks

Long-running tasks are a critical test of delivery capability. Citing a next-generation long-horizon evaluation, the report notes that only a small number of products can fully complete professional workflows involving hundreds of steps across multiple applications.

The report states that SenseTime Raccoon Work has completed the full chain from "evidence acquisition and analytical reasoning to deliverable generation and action execution," directly producing outputs that are editable, collaborative, and reusable. In a bidding scenario, for example, the system parses tender documents, breaks down scoring criteria, generates materials in parallel, and performs template and scoring QA. Internal practice shows that it can compress a process that originally required multiple people working for several days into a matter of several hours for one person.

Supporting real-world deployment is SenseTime Raccoon Work's three-tier architecture spanning individuals, enterprises, and industries. On the individual side, it has served more than 20 million users cumulatively, with over 3 million weekly active users; daily active users have recently grown by an order of magnitude. On the enterprise side, it has been deployed by more than 8,000 customers, including leading companies such as China Mobile, Kingsoft Office, Ant Group, and Lenovo. On the industry side, Raccoon Work turns high-frequency capabilities into reusable templates through OPC standards and certification, industry editions, private deployment, and integrated appliances. In domestic environments, it is also compatible with openKylin and Kylin OS.

The next phase of office AI will be defined not by the length of its feature list, but by whether tasks are actually completed, completed correctly, and manageable. Omdia selected SenseTime Raccoon Work for in-depth analysis as one of four representative vendors—essentially validating its approach of combining desktop execution, enterprise connectivity, and productivity governance.

Going forward, SenseTime Raccoon Work will continue to use real-world workflows as its benchmark, connecting general office applications, industry editions, and domestic deployment. This will reduce the need for individuals to switch between applications and give enterprises the confidence to entrust it with core workflows.

← All articles

More in AI & Machine Learning

All →
Building Production Agents with Jev and LangGraph
LangChain

Building Production Agents with Jev and LangGraph

LangSmith Custom Apps: Build custom interfaces around your agent data
LangChain

LangSmith Custom Apps: Build custom interfaces around your agent data

For months, OpenAI’s agent swarms have been attacking online databases to find obscure factsПресса
OpenAI

For months, OpenAI’s agent swarms have been attacking online databases to find obscure facts

Tesla finally moves to electrify trucking after a decade of work and delaysПресса
Tesla

Tesla finally moves to electrify trucking after a decade of work and delays

New in LangSmith: Engine v2, Managed Deep Agents, Fine-Tuning, and more
LangChain

New in LangSmith: Engine v2, Managed Deep Agents, Fine-Tuning, and more

Tesla poised to scale production of heavy-duty Semi trucks with opening of Nevada factoryПресса
Tesla

Tesla poised to scale production of heavy-duty Semi trucks with opening of Nevada factory

More from SenseTime

Premier Li Qiang Visits WorldSkills Expo, Where SenseTime Presents Its Technologies and Applications
SenseTime

Premier Li Qiang Visits WorldSkills Expo, Where SenseTime Presents Its Technologies and Applications

He used SenseTime Seko to turn a nightmare into a hit short series
SenseTime

He used SenseTime Seko to turn a nightmare into a hit short series

No. 1 in China's Neocloud Market! SenseCore Leads Across the Board
SenseTime

No. 1 in China's Neocloud Market! SenseCore Leads Across the Board

SenseNova U1.5 Technical Report: Toward Native Unified Visual Intelligence
SenseTime

SenseNova U1.5 Technical Report: Toward Native Unified Visual Intelligence