How I Use AI Tools for Work Without Losing My Mind or My Job

Feed meeting transcripts or documents into AI tools like Otter.ai or ChatGPT to get structured summaries. This eliminates hours of note-taking and ensures nothing is missed.

1. 1
   Record or upload the meeting transcript — Use Otter.ai to record Zoom or Teams meetings. It generates a real-time transcript with speaker labels. For documents, copy the text and paste it into ChatGPT. For PDFs, use a tool like ChatPDF that extracts text while preserving structure.
2. 2
   Ask for a structured summary — Prompt: 'Summarize this meeting transcript in 5 bullet points. Include: decisions made, action items with owners, and open questions.' Otter.ai has a built-in summary feature that does this automatically. Review the summary for accuracy—AI can miss nuance or assign wrong action items.
3. 3
   Extract action items into your task manager — Copy the action items from the AI summary into Todoist, Trello, or Asana. I use Todoist and create a task for each item with the owner and due date from the meeting. This ensures nothing falls through the cracks and gives you a single source of truth.
4. 4
   Share the summary with attendees — Paste the AI summary into an email or Slack channel within 2 hours of the meeting. Add a note: 'AI-generated summary, please verify your action items.' This saves everyone time and builds trust in the process. I've seen team morale improve because people feel heard.
5. 5
   Archive summaries for future reference — Save the AI summaries in a shared folder (Google Drive or Notion) with a consistent naming convention: 'YYYY-MM-DD_Topic_Summary'. This creates a searchable knowledge base. When a question comes up months later, you can find the answer in seconds instead of scheduling another meeting.

Use GitHub Copilot or ChatGPT to generate code for repetitive tasks or to debug errors. This reduces coding time by 40% but requires careful review to avoid security or logic issues.

1. 1
   Describe the function you need in plain English — In your IDE with Copilot, type a comment like: '// Function to validate email format using regex' and let Copilot suggest the code. For ChatGPT, prompt: 'Write a Python function that takes a list of integers and returns only the even numbers, sorted descending.' Be specific about language and constraints.
2. 2
   Review the generated code for correctness — Never trust AI-generated code blindly. Check for off-by-one errors, missing imports, and incorrect API usage. I once had Copilot generate a SQL query with a SQL injection vulnerability. Run the code in a sandbox environment first, especially if it touches production data.
3. 3
   Add unit tests for the AI-generated code — Ask the AI to generate unit tests for its own code. Prompt: 'Write 3 unit tests for the above function using pytest.' This catches edge cases the AI might have missed. Run the tests in your CI pipeline. I've found that AI-generated tests often miss null inputs or boundary conditions.
4. 4
   Refactor for readability and performance — AI code tends to be verbose or over-engineered. Simplify variable names, remove unnecessary comments, and optimize loops. For example, Copilot might generate a list comprehension when a simple for loop is clearer. Use your judgment—AI doesn't know your team's coding standards.
5. 5
   Commit with a clear description of what the AI did — In your commit message, note that the code was AI-assisted. Example: 'feat: add email validation (AI-generated with manual review)'. This helps future developers understand the context and sets expectations for code review. Some teams even tag AI-generated code for extra scrutiny.

Use Tome or Gamma to generate slide decks from a simple prompt. These tools create structured presentations with visuals, saving hours of manual formatting.

1. 1
   Define your presentation goal and audience — Open Gamma and type: 'Create a 10-slide presentation for the executive team on Q1 sales performance. Include: revenue vs target, top 3 products, and recommendations.' The more specific you are about the audience, the better the output. Avoid vague prompts like 'make a presentation about sales'.
2. 2
   Let the AI generate the slide structure — Gamma will produce a draft with headings, bullet points, and suggested images. Review the flow—AI often puts the most important slide last. Reorder slides so the key message comes first. I usually move the 'Recommendations' slide to position 2 or 3.
3. 3
   Replace generic visuals with real data — AI-generated charts often use fake data. Replace them with actual charts from Excel or Google Sheets. Use the AI's suggested layout but insert your real numbers. For example, if Gamma shows a pie chart with 25% growth, change it to your actual 18% and adjust the visualization accordingly.
4. 4
   Trim the text for readability — AI tends to write long paragraphs. Condense each slide to 3-5 bullet points maximum. Use the 'Presenter Notes' section for details. This follows the 10-20-30 rule: 10 slides, 20 minutes, 30-point font. I've found that audiences engage more with visuals than walls of text.
5. 5
   Add your personal examples and stories — AI can't know your specific wins or lessons learned. Insert a slide with a real customer story or a challenge your team overcame. This authenticity builds trust and makes the presentation memorable. For instance, replace a generic 'customer satisfaction increased' with 'Client X gave us a 9/10 after we fixed their migration issue.'

Use AI-powered tools like Zapier with ChatGPT or Microsoft Power Automate to automate repetitive data tasks like copying data between apps, formatting spreadsheets, or sending alerts.

1. 1
   Identify a repetitive data task — Look for tasks you do daily or weekly that involve moving data between apps: copying email attachments to Google Drive, updating CRM records from spreadsheets, or formatting dates in a report. For example, I automated the process of saving Slack file uploads to a specific Google Drive folder.
2. 2
   Create a Zapier workflow with AI step — In Zapier, create a new Zap. Add a trigger (e.g., 'New email attachment in Gmail') and an action (e.g., 'Save to Google Drive'). Then add a ChatGPT step to process the data: 'Extract the invoice number and total from this PDF and save them to a Google Sheet.' Test with a real file.
3. 3
   Handle errors with fallback actions — AI steps can fail if the data format is unexpected. Add a fallback action: if ChatGPT returns an error, send a Slack message to yourself with the raw data. This prevents data loss. I've had Zaps fail because the PDF was scanned and not text-searchable—the fallback caught it.
4. 4
   Monitor the automation weekly — Check Zapier's task history once a week to see if any Zaps failed. Adjust the prompt or add data validation steps. For example, if the AI sometimes misreads dates, add a step to format dates using Zapier's built-in formatter before saving.
5. 5
   Scale to more tasks gradually — Start with one automation and run it for a week. Once you trust it, add more. I now have 12 Zaps running that save me about 5 hours per week. The key is to start simple—automating a single data entry task can free up more time than you expect.

Use AI tools like Miro AI or ChatGPT to generate and categorize ideas for projects, content, or problem-solving. This accelerates the initial creative phase and helps overcome writer's block.

1. 1
   Set a clear brainstorming goal — Prompt: 'Generate 20 ideas for blog posts about remote team management. Focus on practical tips for new managers.' The more specific the goal, the more useful the ideas. Avoid open-ended prompts like 'give me ideas about work'—you'll get generic fluff.
2. 2
   Review and categorize the ideas — Copy the AI-generated ideas into a tool like Miro or a simple spreadsheet. Group them by theme (e.g., 'communication', 'productivity', 'tools'). Delete any that are irrelevant or impractical. I usually keep about 60% of AI suggestions—the rest are too vague or unrealistic.
3. 3
   Combine and refine the best ideas — Take the top 3-5 ideas and ask the AI to expand on each: 'For idea #3, provide a detailed outline including key points and examples.' This gives you a solid starting point for execution. For instance, if the idea is 'weekly async standups,' ask for a step-by-step guide on how to implement them.
4. 4
   Add your own domain knowledge — AI lacks context about your team's dynamics, past projects, and constraints. Augment the ideas with your own experience. For example, if the AI suggests 'use Slack for team updates,' add a note that your team prefers email due to time zones. This makes the plan actionable.
5. 5
   Create an action plan from the top ideas — Turn the refined ideas into a project plan with tasks, owners, and deadlines. Use the AI to generate a timeline: 'Based on these 3 ideas, create a 4-week implementation plan with weekly milestones.' Then assign tasks in your project management tool.