Skip to content

Your First Task: From One Sentence to an Acceptable Deliverable ​

After installing WorkBuddy, the most practical question is: WorkBuddy How to Use It? This page answers it in 12 minutes — starting from the interface, you'll create your first task with one sentence, watch it execute autonomously, and finally get a ready-to-use deliverable in the results area on the right: a report, spreadsheet, or PPT file, rather than a block of text you have to copy yourself.

This is the final lesson of the "Quick Start" beginner tutorial for WorkBuddy, and also the first hands-on task lesson. You don't need to prepare any materials: the official documentation provides five example tasks, and copying one is enough to run through the whole flow. Based on WorkBuddy v5.9.3.

What You'll Be Able to Do After This Lesson ​

  • Understand the WorkBuddy interface introduction: what each of the four areas — toolbar, sidebar, conversation area, and results area — is responsible for
  • Create a new task with one sentence (an official example prompt) and run it successfully
  • When you need to work with local files, first select a workspace, then use @ references, paste screenshots, or drag and drop files to add context
  • Find the generated file in the "Deliverables" tab of the results area and locate it on your local disk
  • Send the deliverable out: share a link, send it to WeChat / WeCom, or upload it to the cloud

Your Current Struggles ​

  • It's installed, but when you open it you see nothing but areas and buttons, and you don't know where to click first
  • The AI tools you used before were all chat boxes: one question, one answer, and you still had to copy the answer into a document and format it yourself — what happened to "do the work for me"?
  • The task finished running, but where exactly is the generated file stored? Can you send it directly to a colleague?

When to Use This Technique ​

When you need to:

  • Turn a one-sentence requirement into a directly deliverable file (analysis report, spreadsheet, PPT)
  • Have AI actually execute on your computer (organize files, process data), not just offer suggestions
  • Walk through "submit requirement → watch execution → verify deliverable" in full for the first time, to confirm this tool can really do the work

And you don't want to:

  • Squeeze out answers one sentence at a time in toothpaste-style dialogue, then manually copy, paste, and assemble a document
  • Learn a pile of concepts before daring to start
  • Finish running a task but can't find where the output file is

🎒 Preparation Before Starting ​

  • Completed 1.2 Installation: Main Flow for Windows and Mac, WorkBuddy is installed and you're logged in by scanning the WeChat QR code
  • A requirement you want to hand to AI — if you don't have one, that's fine too: Step 3 has five official examples, and copying one is enough
  • (Optional) A local document you'd like it to process: a sales data spreadsheet, meeting minutes, or your Downloads folder all work
  • 12 minutes of reading time

💡 Does it cost credits?

Conversations in WorkBuddy consume credits; without a subscription you're on the trial version with 500 basic credits per month (free). For your first task, just run one lightweight example. See 2.6 Credits and Pricing for the full rules.

Core Idea ​

WorkBuddy How to Use It — the one-sentence mental model: a conversation is a task. You state the goal in one sentence in the conversation area, WorkBuddy autonomously plans the steps and actually executes them (it can read and write local files), and finally delivers a file in the results area — that's the "from one sentence to an acceptable deliverable" in this page's title.

It's not the same kind of thing as the chat-style AI you're familiar with; the official documentation provides a comparison table:

Traditional AI ConversationWorkBuddy
Can only converse and offer suggestionsCan actually execute tasks
Requires manual file operationsOperates local files automatically
Single-step simple tasksMulti-step complex tasks
Outputs text repliesDelivers acceptable results

So for your first task, there are really only three actions you need to take:

StageWhat You DoWhere You Do It
① Submit requirementDescribe the goal in one sentence (optional: choose workspace, add context)Conversation area input box
② Watch executionAfter sending, watch it work in real time; you can add messages at any timeConversation area
③ Verify deliverableSwitch tabs to find the file, open it to check, and share it outResults area on the right

Let's follow these three steps below.

Follow Along ​

Step 1: WorkBuddy How to Use It — First Get to Know the Four Areas of the Interface ​

Why: WorkBuddy is a desktop workbench, not a web chat box. Spend 1 minute memorizing the interface map first, and you'll know "where to do it" for every step that follows.

The official introduction to the working interface describes four areas:

AreaLocationWhat It Manages
ToolbarTop of the screenEntry points for common operations such as WorkBuddy, Edit, and Window
SidebarLeft sideTask list (grouped by folder), with support for search and management; the user avatar is shown at the bottom
Conversation areaCenterCore interaction area: task title bar, message list, and input box
Results areaRight sideTwo types of content, Overview and Artifacts, which can be switched and viewed in the same area

You should see: Open WorkBuddy's main interface, and the four areas should match one to one. When you open it for the first time there are no tasks yet, so the task list in the sidebar is empty — the next step is to create your first one.

  • I can point out: enter requirements in the center conversation area, view results in the right results area

Step 2: Create a new task and get an independent workspace ​

Why: WorkBuddy's logic for creating a new task is "each conversation corresponds to an independent task" — tasks do not affect each other, and each maintains its own independent workspace and context. Creating a task first is like setting up a clean workbench.

Steps:

  1. Click "New Task" in the main interface
  2. If no directory was specified in advance, the system automatically creates an independent working directory — you can ignore this the first time and go straight to the next step

You should see: You enter the task creation page, with the cursor in the input box, waiting for you to describe your requirement.

💡 Where did the tasks sent from your phone go?

Tasks sent from mobile (via the WorkBuddy mini program / App, WeChat assistant, and other remote channels) are viewed in "Assistant"; this page only covers tasks on the PC client. For an overview of the six remote channels, see 5.1 Assistant Overview.

Step 3: Describe the task in one sentence ​

Why: The official positioning of WorkBuddy is "its core strength is understanding natural language" — you only need to describe your requirement in one sentence, and it can plan and execute autonomously. For your first task, there is no need to write a long prompt.

The official documentation provides five example tasks; copying them verbatim will work:

Example task (official original text)What this one practices
"Help me generate an analysis report from this sales data, including charts"Data file → report + charts
"Organize the images in the Downloads folder and categorize them by date"Batch organization of local files
"Generate a PPT from these meeting minutes"Document → presentation
"Analyze this resume, extract the key information, and generate a table"Information extraction → table
"Research AI industry trends for 2024 and write a report"Produce a report from scratch

For your first run, there are two ways to choose:

  • Easiest: Just use "Research AI industry trends for 2024 and write a report", which does not depend on any local files
  • If you want to practice file handling: Prepare a data spreadsheet of your own, drag it into the input box as in step 4, then use "Help me generate an analysis report from this sales data, including charts"

Typing is not the only way to enter input: click the voice button on the input box to dictate your requirement, and the speech is converted to text in real time and filled into the input box (holding left ⌥ on Mac or ALT on Windows also invokes it); after speaking you can continue editing.

Not sure how to write it? The task creation page offers a set of prompt templates by scenario (daily office work, code development, design and creativity). Hover the mouse over a template, and the input box instantly shows a gray preview of that template's text — the gray text is only a preview; it does not enter the draft, the undo history, or get sent; only after you confirm it is suitable and click the template will the content actually be filled into the input box.

You should see: There is a complete task description in the input box.

  • There is already a requirement in the input box (copying the official example counts)

Step 4: Choose a workspace and add context (optional) ​

Why: If the task involves processing files on your computer (organizing folders, analyzing spreadsheets), you must first tell WorkBuddy "where to read the files". Purely generative tasks (writing a report from scratch) can skip this step and be sent directly.

Choose a workspace: Click "Choose Workspace" at the bottom left of the input box and select the directory to use for the current task. Official note: once selected, WorkBuddy will read and process files based on that directory; you can also start the task without selecting one, and it will generate results in the default directory.

The official guidance also suggests setting up separate spaces by task type (for example, "Invoice organization", "Weekly reports", "Photos to process"), which makes management easier and reduces the risk of misoperation.

Add context (all optional):

MethodHow to do it
Reference contextType @ to reference files, documents, rules, and other information
Paste screenshotCtrl/Cmd + V to paste an image directly from the clipboard
Upload filesClick the upload button, or drag files anywhere on the page (up to 50 files)
Additional notesState the goal, scope, constraints, and expected output in the task description

To make a one-sentence description more reliably correct in a single run, the official guidance suggests stating all four things first:

  • Goal: What content to generate, or which files to process
  • Input: The data to analyze and the documents to reference
  • Output format: Whether you want Word, Excel, PPT, or another format
  • Constraints: Style requirements, word count limits, deadlines, and so on

You should see: After selecting the directory, the workspace you chose is shown at the bottom left of the input box; uploaded files appear as cards in the attachment area of the input box (file name, icon, upload status), and they are not sent automatically after being added — you can keep adding text.

  • (For tasks that process local files) I have set the workspace to the target folder

Step 5: Send the task and watch WorkBuddy execute autonomously ​

Why: This step best shows the difference between an Agent and a chatbot — after sending, there is no transition page like "Preparing to execute…"; WorkBuddy goes straight into the conversation page, and the takeover and execution process is displayed in real time.

Steps: Click send.

You should see:

  • The new task appears in the task list on the left
  • The conversation area continuously shows the execution process and output content, and each model reply area displays the running time
  • The results area on the right starts showing artifacts, all files, changes, previews, and other content based on the task type

While it runs, you don't have to just wait — three things can be done at any time:

  • Append messages: You can keep sending messages while the model is executing; new messages go into the queue above the input box and are processed one by one after the current round of output finishes; messages in the queue support editing, deletion, and drag-to-reorder
  • Switch tasks: Running tasks are shown in the left sidebar, and you can switch to view their progress at any time; you can also create other tasks at the same time to advance them in parallel
  • View next-step suggestions: After a round of conversation completes, suggested prompts for the next step appear above the input box; click one to fill it in (you can pick another or close them)

💡 Don't panic if a "loop execution confirmation" pops up

If WorkBuddy detects that a task may execute in a loop, it pauses before the tool call and shows a confirmation prompt: choose "Continue execution" to keep running; choose "Terminate request" to terminate only this request without affecting the created task, and you can still continue sending messages in the original task afterwards.

  • The task is executing, and its process is visible in the conversation area

Step 6: Review deliverables in the results area ​

Why: "Acceptable deliverables" is the line dividing WorkBuddy from a chat AI: the task doesn't just reply with a message—it generates files. Acceptance = find the file, open and check it, and send it out as needed.

The results area is organized by tabs, with one tab for each of the four result types:

TabWhat to look at
DeliverablesFiles and deliverables generated by the task (documents, spreadsheets, PPTs, reports, etc.)
Web pagesWeb pages, page prototypes, or web apps generated by the task
Workspace filesContents of files in the current workspace
ChangesFile modifications made by this task

For your first task, follow these three steps:

  1. Find the file: The generated target file is in the "Deliverables" tab; for other files produced by the task, check "Overview - Workspace Files"
  2. Verify the file: Clicking "View All Deliverables" opens a new tab on the right; click the "⋯" to the right of a deliverable card to operate on the file:
"⋯" menuPurpose
Open folderOpen the local folder containing the file
Open in default appOpen the file with the system default app
Add to chat boxAdd the content to the input box as a reference
Copy pathCopy the file's local path
Save asSave to a specified local location
Upload to cloudUpload to Library / Tencent Docs / ima Knowledge Base / Lexiang Knowledge Base
  1. Send it out: While previewing a deliverable file, click the share icon in the top-right corner—send a link (the recipient views it via the link), send the file to WeChat, or send the file to WeCom. The latter two require the corresponding WeChat / WeCom client to be installed and logged in on the current device

When uploading to the cloud, Tencent Docs, ima Knowledge Base, and Lexiang Knowledge Base require authorization first; if not yet authorized, clicking the corresponding option will guide you through it.

You should see: Open the generated file in the "Deliverables" tab, and its content matches the requirements you stated in Step 3. If you're not satisfied, go back to the chat area and keep talking (e.g., "add a pie chart too"); it will continue editing based on the current conversation context, without you re-describing the background.

  • I found the file in the "Deliverables" tab and used "⋯ → Open folder" to see its location on the local disk

For detailed steps, refer to the official docs: Start Your First Task, Create a Task, Viewing Results.

Checkpoint ✅ ​

  • I can say what each of the four interface areas is for: toolbar (top operations), sidebar (left task list), chat area (middle interaction), results area (right deliverables)
  • Successfully created and executed the first task with one sentence (the official example verbatim or my own need)
  • Found the generated file in the "Deliverables" tab and can use "⋯ → Open folder" to locate its local position
  • I know where to see other files produced by the task: "Overview - Workspace Files"
  • Can send deliverables out: share a link / send to WeChat / WeCom, or upload to the cloud

Pitfall reminders ​

❌ Treating WorkBuddy as a chatbot: ask one line, copy one line ​

Symptom: You only look at text replies in the chat area the whole time, and in the end you copy the text into Word yourself for formatting.

Why it's wrong: The official comparison table is clear—traditional AI chat "can only converse, offer suggestions, and output text replies," while WorkBuddy "can actually execute tasks and deliver acceptable results." The deliverables are in the results area on the right, as ready-made files.

Correct approach: State the goal fully in one sentence (goal / input / output format / constraints) and let it generate the file directly; when done, go to the "Deliverables" tab to accept, and use "⋯ → Open in default app" to check the content.

❌ The task finished—where are the deliverables? Staring only at the chat area ​

Symptom: The task shows as completed, the chat area is full of process text, but you just can't find the "file."

Why it's wrong: Files are not in the chat area. Per the official wording: the generated target file is in "Deliverables"; other files produced by the task are viewed in "Overview - Workspace Files."

Correct approach: Look at the results area on the right—the "Deliverables" tab for the delivery file, "Workspace Files" for process files, "Changes" to see which files it modified. Can't find the local location? Click "⋯ → Open folder" or "Copy path."

❌ Asking it to organize files without selecting a workspace ​

Symptom: You sent "Organize the images in the Downloads folder by date," but it worked in the default directory and your Downloads was left untouched.

Why it's wrong: Per the official docs—only after selecting a workspace will WorkBuddy "read and process files based on that directory"; if you start a task directly without selecting one, it generates results in the default directory (when no directory is specified at task creation, the system automatically creates an independent working directory).

Correct approach: Before sending, click "Select workspace" at the bottom-left of the input box and choose the directory to process (e.g., Downloads). Also, the official advice: back up important files before processing them.

❌ Sending a vague requirement directly, only to find the direction was wrong after it finished ​

Symptom: "Help me write a report" finished, but the format and framing are all wrong, and the credits were spent.

Why it's wrong: One sentence doesn't mean no information. The official advice is to prioritize the four elements: goal, input, output format, and constraints—these determine the direction of its "autonomous planning."

Correct approach: Self-check the four elements before sending; for multi-step tasks where you fear drift, switch the mode to "Plan": it produces a plan first and executes after you confirm—the official wording is "correct the direction within tens of seconds, rather than discovering the approach was wrong only after the result finishes." For mode details, see 2.4 Three Execution Modes.

Advanced tips (optional) ​

🔧 Switch among the three execution modes as needed. In the + menu on the left of the input box: Default (Agent) executes directly and can modify files; Plan produces a plan first and executes after confirmation; Ask only answers questions without executing. Default is fine for everyday tasks; for how to choose, see 2.4 Three Execution Modes.

🔧 Add messages anytime during execution—no need to just wait. New messages enter the queue above the input box and are processed in order after the current round ends; the queue supports editing, deleting, and drag-to-reorder. For the full picture of multi-turn adjustments, rollback, and references, see 2.1 Conversation Advanced.

🔧 Voice, files, and audio are all input methods. The voice button (Mac ⌥ / Windows ALT), dragging files (up to 50), pasting screenshots (Ctrl/Cmd+V), and audio-to-text are all done in the input box—for the full rundown, see 2.3 Input Methods Explained.

🔧 After running one task successfully, go pick your real work. Full walkthroughs like weekly reports, Excel cleaning, PPT generation, and meeting minutes are all in Scenario Walkthroughs.

Lesson summary ​

✅ WorkBuddy usage—you've gone through the whole flow: create a task → describe the need in one sentence (optional: select a workspace, add context) → send → watch execution in the chat area → accept deliverables in the results area.

✅ The four interface areas are each in place: toolbar (top operations), sidebar (task list), chat area (state needs, watch execution), results area (accept deliverables).

✅ The three actions of "acceptance": find the file in the Deliverables tab, use the "⋯" menu to locate / save as / upload to cloud, and use the share icon to send a link or send via WeChat / WeCom.

✅ The four elements of a one-sentence requirement: goal, input, output format, and constraints—state them all and it runs accurately.

Key actions recap:

Key actionRecommended approachBasis
Create a taskOn the main interface, click "Create Task"; one conversation, one taskOfficial "Create Task" bar documentation
Describe your needsOne sentence + four elements (goal / input / output format / constraints)Official task creation documentation
WorkspaceSelect a directory before processing local files; organize spaces by task typeOfficial task creation / "Create Task" bar documentation
RunningYou can append messages to the queue; multiple tasks can run in parallelOfficial task conversation documentation
AcceptanceGet delivered files from the Artifacts tab; for other files, see "Overview - Workspace Files"Official first task / result viewing documentation

Next lesson preview ​

Your first task ran successfully; the next stop is daily usage: Daily Usage: Turn WorkBuddy into your office companion.

You will learn:

  • Conversation advanced: multi-turn adjustments, rollback, and references—how to make it keep refining unsatisfactory results
  • Task Management: task status, global search, and batch operations
  • Input Methods Explained: voice, file drag-and-drop, audio-to-text, and prompt templates
  • How to choose among the three execution modes (default / plan / Q&A only), and how to switch models
gaodong.aiAboutPrivacyContact

Independent tutorials, not affiliated with any product mentioned. Product names and trademarks belong to their owners. Check official pages for prices and policies.