LobsterAI
LobsterAI is NetEase Youdao's open-source desktop agent that you command remotely from WeChat, WeCom, DingTalk, or Feishu.
If your day runs through WeChat, WeCom, DingTalk, or Feishu and your real work lives on a desktop, LobsterAI is one of the few agents built for exactly that loop. Open-source code, local sessions, and the remote-dispatch skill system are the reasons to pick it over cloud agents or browser-only copilots. It is not a zero-setup chatbot — you must configure a model provider API key (or a local Ollama install) before the first task — so casual users expecting instant answers should look at a hosted assistant instead.
Verified 5d ago · liveness 75/100 · cite: rightaichoice.com/tools/lobsterai
- Chinese office workers who dispatch tasks from WeChat, WeCom, DingTalk, or Feishu and get the finished file back
- Analysts and ops staff who repeatedly turn local spreadsheets into charts, reports, and decks
- Tech-savvy users who want an always-on desktop agent with local memory and auditable open-source code
- Teams with sensitive documents who want sessions, configs, and memory to stay on the machine
- Users who want a quick-answer chatbot without installing or configuring a desktop app
- Teams whose collaboration stack is Slack, Teams, or email rather than Chinese IM tools
- People who need browser-only access from any device with zero local setup
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip LobsterAI if your team's collaboration stack is Slack, Teams, or email rather than WeChat, WeCom, DingTalk, or Feishu, or if you want browser-only access from any device with no local install.
Running tasks consumes credits, and the changelog repeatedly adds top-up and subscription entry points for when your balance runs low — a 1000-credit unlock offer is aimed at new subscribers.
LobsterAI is free to download and open source, with a paid credit/subscription layer for built-in model access and site deployment, plus a Team plan launched in v2026.8.14 covering team management, usage statistics, and per-member quotas. That structure is cheaper than managed cloud agents that bundle model calls into a flat seat price, but more expensive than configuring it with your own Ollama models, which costs only your hardware.
In short
LobsterAI — LobsterAI is NetEase Youdao's open-source desktop agent that you command remotely from WeChat, WeCom, DingTalk, or Feishu. Best for Chinese office workers who dispatch tasks from WeChat, WeCom, DingTalk, or Feishu and get the finished file back, Analysts and ops staff who repeatedly turn local spreadsheets into charts, reports, and decks, Tech-savvy users who want an always-on desktop agent with local memory and auditable open-source code. Free to use.
What's new in LobsterAI
Checked 5 days agoAcross the latest 5 updates: 4 changelog entries and 1 news mention.
v2026.9.20 优化定时任务微信推送状态与登录稳定性
Scheduled-task WeChat push status was reworked so a report that did not arrive can be re-sent on its own. WeChat/QQ QR login errors and repeated restarts after upgrade were also fixed.
v2026.9.17 强化 OpenClaw 2.0 升级与一键修复
OpenClaw 2.0 upgrade and one-click repair capabilities were strengthened, and config hot-reload now protects in-flight IM conversations and scheduled tasks from gateway restarts.
v2026.9.15 重做首次登录引导与长对话额度
A redesigned first-login walkthrough was introduced, OpenClaw upgrades can now auto-backup and repair old configs and databases, and long-conversation output budgets and summary compression were optimized to reduce truncated replies.
v2026.9.14 升级至 OpenClaw 2.0，新增 Markdown 编辑
LobsterAI upgraded to OpenClaw 2.0 (v2026.8.1) for steadier task execution and channel connections, and added direct Markdown file editing with live preview, autosave, and draft recovery.
网易有道发布《中国办公 Agent 用户行为不完全报告》
NetEase Youdao's LobsterAI published what it describes as China's first incomplete report on office Agent user behavior, disclosing usage data from its desktop agent.
What people actually say about LobsterAI — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
13 mentions across 2 sources (Hacker News, YouTube) · researched Aug 27, 2026.
Average across the 2 sources that answered — each source counts once, not each post.
- +Local desktop deployment keeps your data on your machine—strong for privacy.
- +Remote control via WeChat, Feishu, DingTalk, Telegram fits China-based workflows.
- +Runs 24/7, executing queued tasks even when you're offline.
- +Open-source framework (OpenClaw) offers a rich extension ecosystem.
- +Supports global mainstream AI models—no single-vendor lock-in.
- −Almost no real community usage reports or detailed reviews online.
- −A 'scam alert' video exists, undermining trust for new users.
- −Positive buzz seems driven by one YouTube creator, not organic adoption.
- −Setup requires technical proficiency; not for non-technical users.
- −Remote control relies on Chinese messaging apps, limiting global appeal.
- • No paid tiers published yet, but future monetization is a real risk—users should monitor pricing changes
- • Running resource-intensive tasks may require a capable machine, adding hardware cost
Viability Score
How well maintained and how widely used is LobsterAI? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Runs as a local desktop agent on macOS (Apple silicon and Intel) and Windows, current build 2026.9.20
- Remote task dispatch and result delivery via WeChat, WeCom, DingTalk, and Feishu
- Built-in agent browser you can view and take over alongside a running task
- Encrypted website login storage with opt-in agent auto-login
- Data analysis that reads local spreadsheets and outputs visual HTML reports
- Slide and presentation generation from research and local source material
- Document writing, batch reading, and structured extraction into tables (.docx, .xlsx, .pptx, .pdf)
- Application development from a plain-language requirement to a locally runnable app
- Scheduled tasks that collect and deliver recurring briefings automatically
- Task guidance that lets you add or adjust requirements while a task is running
- Multi-agent collaboration configuration with visible subtask and execution records
- Dual-layer memory covering session context and long-term work preferences
- Expert bundles packaged by work scenario (product, marketing, data)
- Skill and connector hub with MCP JSON config import
- Skills and connectors ecosystem built on the OpenClaw agent framework (OpenClaw 2.0)
About LobsterAI
LobsterAI (有道龙虾) is a desktop-level AI agent from NetEase Youdao that runs on your own Mac or Windows machine rather than in a browser tab. It connects to local files, the terminal, the browser, and local projects, so data analysis, slide decks, document writing, web research, and even small application builds happen inside your real working environment. Current builds are versioned 2026.9.20, with separate installers for Apple silicon Macs, Intel Macs, and Windows. The core loop is remote command of a machine that stays on: send one message from WeChat, WeCom, DingTalk, or Feishu and LobsterAI keeps working on the desktop, then returns the finished artifact — store-manager.html for a small-shop inventory app, 产品增长分析.html for a growth report, .pptx research decks, or .xlsx candidate screening sheets. Underneath sits an OpenClaw-based skill ecosystem (upgraded to OpenClaw 2.0 in the September 2026 releases), expert bundles organized by role (product, marketing, data), scheduled tasks, and a skill/connector hub that includes MCP JSON import and a QiChaCha MCP authorization. A dual-layer memory keeps session context plus long-term preferences such as "give conclusions first." Because the source code is public on GitHub and sessions, configs, and memory stay local, you can audit permissions and tool-call records. This is not a generic chat assistant: it trades setup effort for local execution, Chinese IM control, and self-hosted data.
Behind the Verdict
LobsterAI's differentiator is where the work happens. Every other popular agent asks you to upload files into someone's cloud; LobsterAI runs on the machine that already holds your spreadsheets, contracts, and repos. Sessions, configs, and memory stay on the local disk, tool calls are gated and recorded, and the whole codebase is public on GitHub — so a security-conscious team can actually review what the agent is allowed to touch. The September 2026 changelog shows steady hardening rather than feature theater: the OpenClaw runtime moved to 2.0 (v2026.9.14, 2026.9.17), scheduled-task WeChat push delivery can now be re-sent on its own if a report doesn't arrive, and config hot-reload was reworked so in-flight IM conversations and scheduled jobs aren't killed by a gateway restart. The skill library is the practical selling point. Document handling covers .docx, .xlsx, .pptx, and .pdf; there are developer skills for Playwright browser automation, Remotion programmatic video, Canvas design, and frontend component generation; and skill-creator lets you package your own workflow so it recurs. Expert bundles (product manager, marketing, data) install a group of skills at once instead of one by one. Recent releases added an optional built-in agent browser you can watch and take over in the task pane, encrypted website login storage with opt-in auto-login, Markdown file editing with live preview, task guidance that lets you adjust requirements mid-run, multi-agent collaboration configuration, folder-as-context uploads, and a "my files" hub that unifies local artifacts, cloud shares, and deployed sites. Weaknesses are real. Setup is the entry fee: you need an API key from a provider (OpenAI, Anthropic, Google, DeepSeek, Moonshot, Zhipu, MiniMax, Qwen, Doubao, OpenRouter, xAI, or a local Ollama endpoint) or the packaged model route, and the docs and site are primarily Chinese. It is desktop-only — mobile interaction means messaging it, not using an app. The fast release cadence (80 versions logged) means you will be updating often, and the DeepSeek Harness and OpenClaw 2.0 work is recent enough that early builds needed compatibility patches on some machines. Where it fits: analysts, ops staff, and solo builders who repeatedly turn local spreadsheets into reports, decks, and screening tables, and who want a 24/7 job runner. Where it doesn't: teams standardized on Slack, Teams, or email; anyone who wants browser-only zero-install access; and buyers who want a fully vendor-managed service.
Researching LobsterAI? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas LobsterAI actually fits — and what changes day-one when you adopt it.
Uploads 上月销售数据.xlsx, enables the xlsx and docx skills, and asks for a report with Top-10 products, month-over-month trends, and charts.
Outcome: A formatted Word report with trend and bar charts lands locally in 5-10 minutes instead of the 1-2 hours the manual path takes.
Sends one message describing a small-shop inventory system, then lets the agent draft requirements, write the code, and run it locally.
Outcome: A runnable store-manager.html opens on the machine with an inventory dashboard already wired up.
Messages LobsterAI from WeChat to sort yesterday's meeting notes and send back the minutes.
Outcome: A 项目会议纪要.pdf with discussion highlights and follow-up action items arrives in the chat thread while the machine keeps working.
Use Cases
- Automatically gather web data and generate an HTML analysis report on your desktop
- Remotely command document editing or slide creation from your phone via WeChat or Feishu
- Monitor application or ad-dashboard status 24/7 and act on predefined rules
- Screen a stack of resumes or contracts into a structured .xlsx comparison table
- Turn a plain-language requirement into a runnable local app such as a store inventory system
- Run a daily scheduled briefing (for example a 9am AI news digest) delivered to chat
- Package a recurring team workflow as a custom skill with skill-creator
- Process sensitive files locally without uploading them to a cloud service
Models Under the Hood
as of 2026-09-27
Limitations
- You must configure at least one AI model with an API key before first use (unless you run a local Ollama deployment), so a model provider account is required.
- Desktop-only installation across macOS (Apple and Intel chips) and Windows; mobile interaction is limited to remote control through WeChat, WeCom, DingTalk, and Feishu rather than a mobile app.
- Documentation and site content are primarily in Chinese.
- The project ships updates frequently — the changelog logs 80 versions — so expect regular install/upgrade cycles; several 2026 releases specifically patched upgrade failures, repeated restarts, and Windows installer issues.
- The OpenClaw 2.0 migration arrived in September 2026 and older configs, databases, and memory states may need the built-in repair flow.
as of 2026-10-04
Verification history
We have re-verified LobsterAI 8 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
Showing the 6 most recent of 8 verification passes.
Free to cite with attribution — this page re-verifies continuously.
Where the pricing makes sense
The company stage and team size where LobsterAI's pricing actually pencils out — and where peers do it cheaper.
LobsterAI is free to download and open source, with a paid credit/subscription layer for built-in model access and site deployment, plus a Team plan launched in v2026.8.14 covering team management, usage statistics, and per-member quotas. That structure is cheaper than managed cloud agents that bundle model calls into a flat seat price, but more expensive than configuring it with your own Ollama models, which costs only your hardware.
Setup time & first value
How long it actually takes to get something useful out of LobsterAI — broken out by persona, not the marketing-page minute.
Plan on 10-20 minutes for a first useful task: install the macOS or Windows build, add a model provider API key in Settings → 模型 and pass the connection test, then ask for something small. Ollama users skip the key but need Ollama installed locally first. Connecting WeChat, WeCom, DingTalk, or Feishu for remote control adds a separate bot-configuration step.
Switching to or from LobsterAI
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From a browser-only copilot: install the desktop build and point its skills at the same local files you were uploading to the browser tool.
- →From ChatGPT or Claude desktop chat: keep your existing provider account, add the API key in Settings → 模型, and move recurring jobs into scheduled tasks.
- →From a manual report workflow: replace your spreadsheet-to-deck steps by enabling the xlsx, pptx, and docx skills and describing the output once.
- →From Ollama-only local setups: LobsterAI accepts a local http://localhost:11434 endpoint, so existing local models carry over without a new provider account.
- →From scattered task notes: rebuild recurring flows as scheduled tasks or custom skills via skill-creator so they run without re-prompting.
- ↗To a hosted cloud agent: export your finished artifacts from 我的文件 and re-upload the source files to the cloud workspace.
- ↗To a Slack/Teams-native assistant: keep LobsterAI's local documents and move routine notifications into your team's chat platform.
- ↗To raw model APIs: your prompts and skill descriptions are portable, but the OpenClaw skill packaging and gated permissions are not.
- ↗To a plain automation tool: scheduled tasks that only fetch and deliver reports can move to a cron plus scripting setup; agentic web operations cannot.
- ↗To a different open-source agent framework: because LobsterAI is open source, its skill definitions can be reviewed and adapted rather than rebuilt blind.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “LobsterAI”, and we withheld 6: 6 could not be judged, because “LobsterAI” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about LobsterAI.
Official links
Tools that pair well with LobsterAI
Common stack mates teams adopt alongside LobsterAI, with the specific reason each pairing earns its keep.
OpenAgents
OpenAgents is an Apache-2.0 platform for language agents that analyze data, call 200+ plugins and browse the web.
Glide
GlideOS turns a spreadsheet, file, or prompt into working internal business apps and agents for desktop and mobile.
Genspark
Genspark turns one prompt into cited research, decks, dashboards and no-code agents inside a single AI workspace.
Featured Head-to-Head Comparisons
Lobsterai vs Truleo
If you're in law enforcement needing to connect RMS, jail calls, and body cameras into actionable intelligence, Truleo is the only choice—but it comes at a cost. LobsterAI is a free, open-source desktop agent perfect for individual productivity in China, but it has a recent security bug (unauthorized email access) that demands caution. Different tools for completely different users.
Lobsterai vs Presto Voice
For large QSR chains aiming to boost drive-thru revenue and efficiency, Presto Voice is a proven enterprise solution with high automation rates and upselling capabilities. LobsterAI offers a free, open-source desktop agent for individual users seeking local task automation, but a recent security bug raises concerns. Choose based on your operating context: drive-thru operations vs. personal desktop productivity.
Lobsterai vs Locus Robotics
These tools target completely different domains. Locus Robotics is a proven industrial automation platform for large-scale warehouses, while LobsterAI is a free, open-source desktop assistant for individual productivity. Choose Locus if you need to scale fulfillment operations; choose Lobster if you want a local AI agent to automate PC tasks. The recent LobsterAI security bug may concern enterprise buyers.
Alternatives to LobsterAI
View all
OpenAgents
OpenAgents is an Apache-2.0 platform for language agents that analyze data, call 200+ plugins and browse the web.
Frequently Asked Questions
Categories
Best-of guides
Used LobsterAI? Help shape our editorial sentiment research.