I tracked my time for a month before adding AI to my workflow, then tracked it again after. The result: 10.5 hours per week recovered — not by working less, but by stopping the manual busywork that used to fill my day. Here is exactly what changed and the specific workflows behind it.
Before I started optimizing, here is where my time was disappearing each week: email drafting and replies (3.5 hours), meeting prep and follow-up notes (2 hours), research for decisions or content (2 hours), status updates and reports (1.5 hours), and miscellaneous writing tasks like Slack messages and briefs (1.5 hours). That is 10.5 hours of mostly cognitive overhead — work that produces outputs but does not move anything important forward.
The pattern I noticed: almost all of it was writing. AI is extremely good at writing. That is why this worked.
If you want a single-session setup for your morning AI routine, see our 20-minute AI morning routine guide — it pairs perfectly with the workflows below.
This is where the biggest gain came from. I used to open every email, read it fully, think about what to say, draft a reply, revise it, send it. For 40–60 emails a day, that adds up fast.
Now: I batch emails into 3 sessions per day. For each session, I copy and paste a batch of email threads into Claude with this prompt:
Here are 8 emails I need to respond to. For each one: 1. Summarize what they need in one sentence 2. Suggest the appropriate action (reply/delegate/archive/snooze) 3. If it needs a reply, draft one (professional, direct, under 100 words) [paste emails]
I review the drafts, edit maybe 20% of them, and send. What used to take 3.5 hours now takes about 45 minutes. Time saved: 2.75 hours. For a more automated version of this workflow, see our guide on building an AI email triage system.
I have about 8 meetings per week. Each used to require 10–15 minutes of prep and 15–20 minutes of follow-up notes. That is 3.5–4 hours of meeting overhead for meetings I was already attending.
Prep is now a 90-second prompt: I paste the meeting agenda and any relevant context, ask Claude for 3 smart questions to ask and the key things I should know going in. Follow-up is equally simple — I paste my rough notes immediately after the meeting and ask for a formatted summary with action items and owners.
Meeting notes (rough): [paste notes] From this, produce: - 3-sentence summary of what was decided - Action items with owners and due dates - Open questions that need follow-up - Any risks or blockers mentioned Format for easy copy-paste into Notion.
Total time savings on meeting overhead: about 2 hours per week. You can automate this end-to-end — see the AI meeting summarizer workflow for a Make.com pipeline that handles transcription and note delivery automatically.
Before I made any significant decision — choosing a vendor, evaluating a new tool, preparing a recommendation — I used to spend 1–2 hours gathering information and structuring my thinking. AI has compressed this dramatically.
For tool evaluations, I paste the list of options and my requirements and ask Claude to create a comparison framework. For decisions, I describe the situation and ask for the key factors to consider, the likely tradeoffs, and what information I should gather before deciding. The research is not replaced — but the framing and synthesis that used to take an hour now takes 10 minutes.
Time saved: about 1.5 hours per week on research overhead.
Status updates are the most boring writing you will ever do. They are also weirdly time-consuming because you have to mentally gather all the context, figure out what is worth mentioning, and format it for the audience.
My workflow now: I keep a running daily log in Notion (just bullet points — takes 2 minutes at end of day). At the end of the week, I paste the week's log into Claude with this prompt:
Here is my work log for this week: [paste log] Write a weekly status update for my manager. Include: - 3 key accomplishments (with impact, not just activity) - What I am working on next week - Any blockers I need help with - One risk on my radar Keep it under 250 words. Professional but not stiff.
Notion AI's built-in summarization can also auto-draft this from your pages — useful if you are already living in Notion. Time saved: about 1 hour per week.
The remaining hour comes from dozens of small writing tasks: drafting a Slack message for a tricky situation, writing a job posting section, creating a one-paragraph brief for a project, writing a LinkedIn post. Each is a few minutes, but they add up to about an hour weekly.
For all of these, the approach is the same: give AI the context, specify the output format and audience, set a length limit. The first draft is almost always usable. You refine the tone and add any specifics the AI could not know.
For the tasks that repeat every week — the status update, the report, the follow-up email template — it is worth going one step further and building an automation. Make.com lets you schedule AI-powered workflows that run without you touching them: pull your Notion log on Friday afternoon, send it to the Claude API, and deliver the formatted status update draft to your inbox. One-time setup, permanent time savings.
Do not try to overhaul everything at once. Pick the one category where you spend the most time on writing overhead. For most people that is email. Spend one week doing nothing but AI-assisted email drafting. Track your time honestly.
Once that is a habit, add meeting follow-up notes. Then status updates. Each one takes a week to become a reflex; within a month, you will have 5–8 hours back and it will feel effortless because each individual habit is small.
The tool matters less than the habit. Whether you use Claude, ChatGPT, or something else, the leverage comes from consistent use on the specific tasks that eat your time — not from occasional experiments when you remember to try it. For more high-leverage prompts to use daily, our list of 10 Claude prompts that transform your workday is a good next read.
The three tools that power this workflow: Claude or ChatGPT (your primary AI assistant), Notion with AI enabled for note-keeping and daily logs, and Make.com for automating any workflow that runs on a schedule. Together these three tools cover writing, knowledge management, and automation — the three levers that account for the entire 10-hour weekly recovery described above.
💡 Ready to build your own AI productivity system? See the tools we actually use →
One weekly email with real-world AI workflows, prompts that actually work, and tool recommendations. No fluff.