← Assistants & Chatbots
Assistants & Chatbots · General-purpose assistant

DuMate

Baidu desktop general-purpose productivity agent that can operate software and browsers, process files and automate multi-step office workflows.

7.4OUT OF 10 · LOW CONF.

DuMate review

Updated 2026-08-28

Product & capabilities

Baidu describes DuMate as a desktop AI agent for enterprise employees and professional users. It can see screen content, manipulate applications, browse the web, process local files and chain multi-step workflows. The execution stream exposes the process, including file handling, scripts and browsing, which gives users more auditability than opaque one-shot automation.

The product uses a resource-point system across agent tasks and Skills. Enterprise administrators can allocate seats and shared resource pools, manage users/groups and configure login/security controls. This makes DuMate an organizational automation platform rather than only a personal desktop copilot. 1

Pricing & access

DuMate currently has Free, Pro and Max personal tiers. Baidu advertises first-month continuous-subscription promotions of ¥9.9 for Pro and ¥69.9 for Max; normal purchase-page pricing can vary and should be checked in-product. Paid plans consume points as the common usage unit. Enterprise pricing is seat-plus-resource based; extra enterprise capacity is documented at ¥1,500 per 250,000 resource points, with 250,000 as the minimum add-on. 23

Evidence & trust

Independent exact-product review and benchmark evidence remains very limited, which constrains comparison with more mature desktop-agent products without constituting negative evidence by itself.

Agent Finder published a hands-on review in May 2026 after testing coding, research and file workflows. It found DuMate genuinely capable inside Baidu's ecosystem while identifying English-language usability as a major constraint. This is useful qualitative evidence, but it is one publication and predates some current billing documentation.

Baidu's current service rules state that, after security processing and strict de-identification, collected service data and AI responses may be evaluated, analyzed and used for model training. Organizations with sensitive data therefore need to examine enterprise privacy/security terms carefully rather than assuming all interaction data is excluded from model improvement. 45

Who it's for

DuMate is best suited to Chinese-speaking professionals and enterprises that want one agent to span browser, desktop software, files, scripts and internal systems. It is less suitable for English-first teams seeking a globally documented ecosystem, a broad independent review corpus or transparent cross-market benchmarking.

Strengths & weaknesses

Strengths

  • True desktop/action agent rather than chat-only assistant.
  • Cross-application workflow automation and file/script handling.
  • Visible execution stream improves auditability.
  • Enterprise seats, shared resource pools and admin controls.
  • Free plus paid personal tiers and documented enterprise capacity model.

Weaknesses

  • Very limited independent exact-product evidence.
  • Strong Chinese-language/Baidu ecosystem orientation.
  • Usage-point economics can be harder to forecast than flat subscriptions.
  • Data-use language requires careful enterprise privacy review.
  • No independent benchmark validating reliability on long-horizon workflows.
SOURCES

Sources & references

5 sources
  1. Official sourceBaidu Intelligent Cloud — DuMate documentation
    OFFICIAL
  2. Official sourceBaidu Intelligent Cloud — DuMate personal billing
    OFFICIAL
  3. Official sourceBaidu Intelligent Cloud — DuMate enterprise billing
    OFFICIAL
  4. Official sourceBaidu Intelligent Cloud — DuMate enterprise service rules
    OFFICIAL
  5. SourceAgent Finder — DuMate Review
    EDITORIAL REVIEW
MAJORBaidu launched DuMate as a general-purpose desktop productivity agent.

News & updates about DuMate

Recent AiToolMap coverage linked to this product.

Related AI tools

Products matched by shared tasks, audience, workflow, product type and capabilities.

AI
Construct Computer
Construct Computer is a thoughtfully designed AI-employee workspace for small teams because it combines a persistent cloud computer, inspectable memory, background work, integrations, reusable workflows and agent-built internal apps. The limitation is not feature breadth but assurance: v2 is still in active development, public third-party evidence is thin, workflows remain structurally limited, and the agent can take external side effects without a mandatory per-action approval gate.
WHY RELATED · Shared task: Workflow automation · Similar input/output: Natural-language instructions → Files
7.7/10AITOOLMAP RATINGRATING CONFIDENCE · LOW
AI
happycapy
Happycapy is a compelling browser-first agent workspace that removes much of the setup friction from running persistent AI agents, automations and coding/research workflows in an isolated cloud sandbox. Its current pricing is competitive and its Product Hunt reception is strong, but independent evidence remains concentrated in a small number of community surfaces and mobile functionality is not fully identical to the desktop agent-native computer.
WHY RELATED · Shared task: Workflow automation · Similar input/output: Natural-language instructions → Files
7.8/10AITOOLMAP RATINGRATING CONFIDENCE · LOW
AI
MuleRun
MuleRun is a credible always-on AI-agent platform built around a dedicated cloud computer, persistent memory, scheduled work and enterprise connectors. It is more ambitious than a chatbot and has strong early hands-on reports, but the product is still young: output consistency on long-running workflows remains a known work in progress, credit economics require active monitoring, and independent review volume is still limited.
WHY RELATED · Shared task: Workflow automation · Similar input/output: Natural-language instructions → Files
6.5/10AITOOLMAP RATINGRATING CONFIDENCE · LOW