AI for Productivity

Claude is the strongest for written communication, summarizing long documents, and structured thinking tasks where you need a clear output. ChatGPT is faster for quick questions, brainstorming, and research tasks with no specific format requirement. Gemini integrates directly with Google Workspace, which makes it the default choice if most of your work happens in Docs, Gmail, and Sheets. The best tool is the one that fits where your work already lives.

The key is applying AI to tasks with high volume and low judgment requirements first. Email drafting, meeting summaries, first-draft documents, and research summaries are all high-volume and low-stakes enough that the output needs light review rather than heavy editing. Apply AI to tasks that require deep judgment (strategic decisions, relationship communication, nuanced analysis) only after you have established where the output quality meets your standards.

The most effective workflow is not using AI to read your email, it is using AI to write faster replies. When you open an email that needs a response, paste it into Claude or ChatGPT with your intended answer in a bullet point, and ask for a three-sentence professional reply. This is faster than writing from scratch and produces cleaner output than typing while distracted. For email summaries, Gemini's Gmail integration handles this without leaving your inbox.

Indirectly, yes. AI reduces context-switching by letting you clear shallow work (emails, summaries, admin) faster, which creates longer uninterrupted blocks for deep work. It also helps you prepare for deep work sessions: use it to pre-plan the specific questions you need to answer, the specific output you need to produce, and the specific information you need. Starting a deep work session with a clear micro-agenda reduces the time spent figuring out what to do next.

Use AI for structure and first drafts, use your own judgment for final output. Give the AI an outline or bullet points, ask for a first draft, then rewrite the parts that are not quite right. This is faster than writing from a blank page and produces better output than editing AI prose without rewriting it. The editing mindset ("this is good enough") produces worse results than the rewriting mindset ("I am using this as raw material").

AI can help you think through prioritization, but it cannot do it for you because it does not know your context. The useful workflow is to paste your task list into Claude and tell it: your top goal for this week, the deadline pressures you are under, and which tasks require energy versus which ones are administrative. Ask it to identify which three tasks have the highest leverage on your goal. Use it as a thinking partner for the decision, not a decision-maker.

The most consistently reported gains are in writing-heavy work: emails, reports, and documentation that used to take 30-60 minutes per piece take 10-15 minutes with AI assistance. Research and synthesis tasks that involved reading and summarizing multiple sources are faster by a similar ratio. The tasks where AI saves the least time are ones that require ongoing context about your specific situation (strategic decisions, relationship management, nuanced judgment calls) because the setup time to brief the AI approaches the time saved.