AI at work
Choose a work task, try the sample, and check the result.
A useful place to start
A traceable decision log and action list that keeps unknown owners and dates visible.
Start this lessonPractice with
Practice files, reusable prompts, and review checks come with each lesson. Check the prerequisites before starting.
Use the sample files first. Follow your employer’s AI policy before adding workplace information. Review every result before sharing it.
Follow the lessons in order, using their fictional inputs and review checks.
Spreadsheets and numbersintermediate
A traceable exceptions file that preserves duplicate and missing-value problems rather than hiding them.
ChatGPT Work on the web with file analysis and file output available
Meetings and decisionsintermediate
A one-page briefing that distinguishes meeting purpose, historical context, new questions, and unresolved timing.
ChatGPT Work on the web with uploaded text files and document output
Hiring and onboardingintermediate
An interview preparation document linking every question and probe to a source competency.
ChatGPT Project with uploaded job brief, rubric, and project instructions
Meetings and decisionsbeginner
A traceable decision log and action list that keeps unknown owners and dates visible.
Claude or ChatGPT conversation with pasted synthetic text
Workplace communicationbeginner
A short update that makes one request and adds no unapproved promise.
Claude or ChatGPT text conversation
Compare how people review the same source. Agree who checks the result before using it.
Use ChatGPT for a work task you can inspect from beginning to end. Pick an outcome, read the sample inputs, and follow the walkthrough before substituting approved workplace material. The goal is a result you can check against its source, with uncertainty still visible when the source does not settle a question.
Start with meeting notes or an email draft if you are new to this way of working. Then try a task that uses files, a structured output, or reusable context. Each lesson names its prerequisites and provides fictional inputs, an illustrative result, and checks specific to the task. You do not need to connect an inbox or use real employee records to practice.
Example outputs and completed product tests are different things. Read the testing information before relying on detailed instructions, and treat an unverified step as something that still needs validation. If a feature is unavailable in your account, use the troubleshooting guidance rather than assuming another product surface behaves the same way.
For team practice, compare how people review the same input. Agree who confirms facts, approves an external action, and maintains the source material. A copied prompt or downloaded file shows an interaction, not that a workplace task succeeded.
