Your manager needs an update on seats and the start time for a customer training session. You have two versions of the registration count and a chat message proposing a later start. Give both WorkBuddy modes the same records and see which brief you can use.
The exercise leaves you with two replies and a plain text scoring record. You need a signed-in desktop app, enough credits to run tasks, and two available model modes. With one available mode, you can still run the exercise and score the response. The training records below are fictional. Model settings
Start a fresh task in Fast mode
Create an office task and select Fast in the model menu by the input area. Record the mode and any model name the interface displays. Send this entire block as one message. New task and model controls
These are fictional training records. The reporting cutoff is November 18 at 15:00.
A | Nov 18 12:00, registration sheet. 46 registered and 38 confirmed attendance.
B | Nov 18 14:00, updated registration sheet. 48 registered and 40 confirmed attendance. Actual attendance has yet to be counted.
C | Nov 18 13:30, venue email. The room is confirmed to seat 45.
D | Nov 18 14:20, project lead's message. The start time is confirmed at 09:00.
E | Nov 18 14:45, group chat. The speaker asks for 09:30. The lead has yet to confirm the change.
F | Nov 18 14:50, equipment note. Li Ning is following up on the adapter. Delivery time is undetermined.
Use these six records to brief a manager. Include Registration and venue, Start time, and Follow-up. Use the latest registration sheet and distinguish attendance confirmations from actual attendance. Calculate remaining seats using confirmed attendance and show the basis. Keep the confirmed start and the proposed change separate. Give the known follow-up owner and timing. Cite each key fact by record letter and flag missing information.
Wait for the full brief and copy it into a text document under Fast, run 1. Record the elapsed time shown with the reply. If you use a stopwatch instead, start it when you send the prompt and stop when the complete reply is visible. Use the same timing method for every attempt. Task timing
Run the same prompt in Balanced mode
Start a second task, select Balanced, and send the same block. Use the same device and task settings. The fresh conversation keeps the first answer out of the second run. Save the second reply and time under Balanced, run 1.
One run per mode is enough to practice scoring. For an everyday choice, run each mode another one or two times in fresh tasks, alternating the order. Record errors and timeouts too. Offline evaluation
Score each response against five checks
- The updated counts are 48 registered and 40 confirmed. The source is B.
- Actual attendance has yet to be counted, as B states.
- On a confirmed-attendance basis there are 5 seats left, because 45 minus 40 equals 5. This uses B and C.
- The confirmed start is 09:00. The proposed 09:30 start is still awaiting approval. The sources are D and E.
- Li Ning is following up on the adapter and its delivery time is undetermined. The source is F.
Copy one scoring block for each reply. Save the filled blocks in a plain text file such as Training Brief Comparison.txt.
Mode and displayed model name:
Run number:
48 registered and 40 confirmed (pass/fail):
Actual attendance has yet to be counted (pass/fail):
5 seats left using confirmations and sources B/C (pass/fail):
09:00 confirmed and 09:30 proposed, sources D/E (pass/fail):
Li Ning follows up on the adapter, delivery time undetermined, source F (pass/fail):
Elapsed time:
Credits attributable to this task (enter unavailable if unclear):
WorkBuddy's website has a credit usage history under your profile's plan and usage area. Record a credit figure only when you can match it to the task. Mark it unavailable if other work makes it impossible to identify. View usage
Choose based on the checked results
A passing reply gets all five checks right. If both modes pass, compare their elapsed times and any available credit figures across repeat attempts. If a brief treats 09:30 as confirmed, record the exact error and check how the second mode handles the same question. If both miss key facts, clarify the prompt and give both modes the revised version. One exercise measures performance on this task.
To use the method for your own work, choose a recurring brief and a set of records you can verify yourself. Write a few expected facts first so you have something concrete to check after each run.
Terms used in this exercise
- Model mode. A WorkBuddy setting for the model configuration used by a task.
- Prompt. The task instruction you send, including any records pasted with it.
- Calculation basis. The count or definition used for a calculation. Here, the remaining seats are based on confirmed attendance.
- Credits. The units WorkBuddy uses to report account usage.