WeCom and WorkBuddy Connect, Making AI a Coworker in the Chat List

In the version described by ifanr, a WorkBuddy entry appears in WeCom's sidebar. After claiming access, users can give tasks directly in a chat window. A fuller WorkBuddy enterprise workbench can also be downloaded and logged into with a WeCom identity. The bot appears in WeCom contacts, where users can chat with it or switch to the workbench to choose models and add connectors.

The report's APPSO team tested several office scenarios. For monthly reports that employees submit through WeCom, users previously had to gather articles, data and analysis before feeding them to an AI and then pasting results into a form. With the integration, the report says, a monthly report can be assigned to WorkBuddy in one sentence inside the WeCom chat window, and WorkBuddy can call on work context to produce the result.

In a smaller task, WorkBuddy was asked to create a work group. It returned the relevant colleague information and then showed a WeCom group-creation confirmation box with the group name and three members listed. After confirmation, the new group was created. The report presents this as an example of the bot handling a routine administrative step directly inside WeCom.

The test also involved a topic-selection agent for an editorial team. Instead of exporting chat records and uploading them to another AI, the team asked WorkBuddy inside WeCom to review past discussions about topic choices and organize them into a reusable Skill. It then asked WorkBuddy to combine existing group topics with the newly organized selection criteria, search for recent news and recommend up to ten topics worth following, with news facts, reporting angles, sources and timestamps, and with labels showing which topics were already discussed and which were new.

For a multi-person project, the team used WorkBuddy to prepare a reader meetup. The activity plan was in a document, arrangements were discussed in a group, and progress was tracked in a spreadsheet, with colleagues updating information and changing workflows. WorkBuddy generated a detailed progress report covering confirmed changes, task progress, unapproved suggestions and pending decisions. It also organized the information into a project board.

The report says WorkBuddy then checked which tasks carried delay risks and reminded the responsible people. Because it is connected to WeCom's messaging and collaboration functions, it could send messages and create group chats within its authorization, returning task results to colleagues. The report describes the flow as reading context, producing output and returning results to the workflow without copy-pasting. It adds that AI can send messages, create groups, save files to a cloud drive and push items to responsible people, moving work into the next stage.

The report also notes that real enterprise work is more complex than the tested scenarios. Different departments use different systems, employees have different permissions, and business data and customer information cannot be read freely by AI. To address management needs, administrators can use WeCom's existing organizational structure to allocate WorkBuddy access, set access permissions, manage usage and purchase AI compute, according to the report. Employees do not need to find AI tools and register accounts individually, and enterprises can decide who can use the service and what information can be accessed.

For companies that have built CRM, ticketing, inventory and other business systems, the report says existing workflows do not need to be rebuilt. Through interfaces, connectors and permission settings, AI can gradually join existing work steps. Routine project organization and collaboration can be handled by the WorkBuddy bot in WeCom, while tasks that require local computer files or more specialized work can continue in WorkBuddy's standalone workbench.

The integration is aimed at placing AI inside the work environment employees already use, according to the report. It describes the goal as reducing the need to explain background, move materials and switch between software, so work that is already running can continue. The report says users should not have to spend time learning to work with AI; instead, AI should learn how to work with them.