WorkBuddy 5.9.2: final voice transcripts go straight into the input box

On October 26, 2026, WorkBuddy released version 5.9.2. Its official changelog says final speech-recognition transcripts now go straight into the input box without an extra polishing step. Support for combination shortcuts has also been restored. Release notes

WorkBuddy is an AI workspace for documents, spreadsheets and local tasks, started by entering task instructions. This change affects the step that turns speech into text ready to send: the input box receives the final transcript for the user to check before handing it to the assistant. The announcement provides no accuracy or speed figures for this change. Check whether the text preserves your requirements; removing polishing does not guarantee an error-free transcript.

Suppose you want the assistant to organize event assignments. You say: “Turn tomorrow’s event assignments into a checklist. The budget is 3,000 yuan. Do not contact suppliers.” Before sending, the important details to check are the amount, the timing and the restriction on contacting suppliers.

With the extra polishing step removed, your check focuses on the final transcript. You can restore missing conditions in the input box or replace vague references with specific names before submitting the task. This approach suits people who already know what they want done and prefer dictation for getting it down. The assistant still receives the text you have confirmed.

Restored combination shortcuts also help existing users retain their keyboard habits. Use the bindings currently configured in your client rather than copying someone else’s key combination.

This check is for people who could already start and finish voice input in WorkBuddy. Continue using the recording entrance you have already confirmed works.

1. Check the version first. In the signed-in desktop client, click your avatar at the bottom of the left sidebar, select “Check for updates,” and follow the prompts to install 5.9.2 or later. The avatar menu also opens Settings, which provides general shortcut viewing and configuration. Settings guide
2. Create a task and prepare your requirements. Click “New task” on the main screen, then choose a working directory and model. The directory is where task files are read and saved; if you have not specified one beforehand, the app creates a separate directory for the new task. Task entry point
3. Use the recording method you already know. Start recording using your previously working method, dictate the event-checklist request above, then end recording in the same familiar way. Once the final transcript appears, check the amount, date and negative instructions in the input box, and manually add any needed context. If the previous voice method no longer works after updating, enter the same request as text and report the issue through “Feedback” in the avatar menu. Voice update
4. Send after confirming the text. Sending opens the conversation page, where you can see the work progress. Review the resulting assignment checklist to confirm that it follows the budget and constraints. If it needs changes, continue in the same task with your corrections. Checking input and checking execution results are separate steps worth keeping. Conversation guide