ChatGPT for work productivity

{{whyLabel}}: You cannot optimize what you haven't measured; identifying repetitive tasks ensures you apply AI where it has the highest ROI.

{{howLabel}}:

• List every task you perform more than twice a week.
• Categorize them into 'Data Entry', 'Communication', 'Content Creation', or 'Analysis'.
• Rate each task by 'Cognitive Load' (1-5) and 'Time Spent'.

{{doneWhenLabel}}: [A spreadsheet or list exists with at least 10 categorized and rated tasks]

{{whyLabel}}: AI excels at structured data and drafting but fails at high-stakes emotional intelligence or physical presence.

{{howLabel}}:

• Mark tasks as 'AI-Ready' if they involve processing text, summarizing, or generating structured data.
• Mark tasks as 'Human-Only' if they require final legal approval, deep empathy, or physical interaction.
• Focus your system-building only on the 'AI-Ready' list.

{{doneWhenLabel}}: [The task map has a clear 'AI-Ready' filter applied]

{{whyLabel}}: This reduces prompt length by up to 60% by providing permanent context, preventing the AI from being generic.

{{howLabel}}:

• Layer 1 (Context): Define your role, industry, and typical projects.
• Layer 2 (Instructions): Set rules like 'Always use Markdown', 'Be concise', and 'Ask for missing info before answering'.
• Layer 3 (Tone): Choose a professional persona (e.g., 'Analytical Consultant' or 'Creative Strategist').

{{doneWhenLabel}}: [Custom Instructions are saved in your AI settings]

{{whyLabel}}: Structured prompts yield 3x better results than conversational 'chatting'.

{{howLabel}}:

• Use the acronym: Context, Objective, Style, Tone, Audience, Response format.
• Practice by rewriting a simple request (e.g., 'Write an email') into a CO-STAR prompt.
• Ensure every prompt specifies the 'Response format' (e.g., 'a 3-column table').

{{doneWhenLabel}}: [One complex task is successfully completed using a CO-STAR prompt]

{{whyLabel}}: Reusing successful prompts saves hours of 're-prompting' and ensures consistency across projects.

{{howLabel}}:

• Create a page in a note-taking app or a simple Markdown file.
• Store templates for recurring tasks: 'Meeting Summary', 'Email Draft', 'Code Review', 'Data Analysis'.
• Use placeholders like [INSERT TEXT HERE] for easy copy-pasting.

{{doneWhenLabel}}: [A library with at least 5 reusable prompt templates is accessible]

{{whyLabel}}: Automated transcription and summarization eliminate the need for manual note-taking and ensure no action items are missed.

{{howLabel}}:

• Select a generic 'AI Notetaker' that integrates with your video conferencing tool.
• Set it to automatically join meetings and generate a 'Summary' and 'Action Items' list.
• Create a prompt in your library to 'Refine meeting notes into a project update'.

{{doneWhenLabel}}: [First meeting is automatically summarized with clear action items]

{{whyLabel}}: Drafting emails is often the biggest 'time-sink'; AI can handle the structure while you provide the intent.

{{howLabel}}:

• Use a browser-based AI sidebar or a dedicated email plugin.
• Feed the AI 3-5 bullet points of your intent.
• Use the 'Reply with [Tone]' feature to generate drafts in seconds.

{{doneWhenLabel}}: [Five consecutive emails are drafted using AI assistance]

{{whyLabel}}: Standard summaries are often too vague; this method ensures high information density without fluff.

{{howLabel}}:

• Prompt the AI to identify 5 missing entities from a summary and re-write it to include them.
• Repeat this 3 times until the summary is dense and highly informative.
• Use this for long reports, whitepapers, or competitor analysis.

{{doneWhenLabel}}: [A 10+ page document is summarized into a high-density 1-page brief]

{{whyLabel}}: Privacy is critical in 2025; local models allow you to process confidential data without it leaving your machine.

{{howLabel}}:

• Install a local AI runner (e.g., Ollama or LM Studio).
• Download a medium-sized model (e.g., Llama 3 or Mistral).
• Use this specifically for internal financial data or private client notes.

{{doneWhenLabel}}: [A local AI model is running and responding to a test query offline]

{{whyLabel}}: True productivity comes from 'Agentic' workflows where AI triggers automatically based on events (e.g., a new email or form submission).

{{howLabel}}:

• Choose an automation tool (e.g., n8n for open-source or Make for ease of use).
• Create a simple 'Trigger -> AI Action -> Output' flow.
• Example: 'When a new lead arrives in Gmail -> AI summarizes their LinkedIn profile -> Post to Slack'.

{{doneWhenLabel}}: [One automated workflow is active and running without manual input]

{{whyLabel}}: Specialized agents perform better than general-purpose ones for niche tasks like 'Brand Voice Checking' or 'Code Debugging'.

{{howLabel}}:

• Use the 'Create a GPT' feature (or equivalent system prompts).
• Upload your company's style guide, past successful reports, and specific SOPs as 'Knowledge'.
• Test it with a task it was specifically designed for.

{{doneWhenLabel}}: [A specialized AI agent is created and tested with 3 sample tasks]

{{whyLabel}}: You need objective data to decide which AI tools to keep and which to discard.

{{howLabel}}:

• Metric A: Time saved per week (Target: >4 hours).
• Metric B: Quality of output (Scale 1-5, Target: >4).
• Metric C: Friction/Stress level (Target: Lower than baseline).

{{doneWhenLabel}}: [A tracking sheet with these 3 metrics is ready]

{{whyLabel}}: Immersion is the fastest way to find system weaknesses.

{{howLabel}}:

• For 7 days, attempt to use AI for every 'AI-Ready' task identified in Phase 1.
• Do not skip the system even if it feels slower at first (learning curve).
• Log every 'Friction Point' where the AI failed or was too slow.

{{doneWhenLabel}}: [7 days of work are completed using the new system]

{{whyLabel}}: Systems naturally accumulate 'bloat'; this step ensures only high-value workflows remain.

{{howLabel}}:

• Review your friction log from the 7-day sprint.
• Keep: Workflows that saved time and maintained quality.
• Tweak: Workflows that were useful but needed better prompts.
• Delete: Workflows where the human effort of prompting exceeded the value of the output.

{{doneWhenLabel}}: [Finalized list of permanent AI workflows is established]