Hi everyone,

Something unexpected changed for me over the last few weeks, and I want to share it in its rough-but-ready form — because you can take the best of it and start using it now to:

• overcome email overwhelm, and
• gain an actually helpful intelligent assistant — one that knows how to:
  • protect your precious deep work time,
  • triage email by urgency, your key relationships, and the projects you’re actively working on, and
  • apply the principles of cognitive high performance to you.

Here’s my day now:

• Mornings belong to my most important work. I don’t open email. By 7am a briefing has already checked everything — my inboxes, the systems I monitor — and most mornings it says: nothing urgent for you today; go do your deep work. That one sentence has removed a tension I’d carried for years: the fear that focusing means missing something important.
• At 12:30, email stops being an infinite todo list anyone can add to. Instead it becomes a channel checked against my key relationships and active projects, condensed into a single batch with a fixed size: “about 30 minutes today; the time is blocked; drafts are waiting.” When the block arrives, I review, rewrite in my own words, send. Done.
• At 4:30, the day ends well. The shutdown briefing does one last check for anything urgent or from key relationships — and then the day closes. Most days, my shutdown ritual doesn’t include email at all anymore.

The result surprised me. I’m more responsive now, not less. The people who matter most get faster, better replies than when I was checking “constantly” — and my deep work is more protected than it has ever been (and deeper as a result).

Why this helps — a little of the science

This design isn’t arbitrary — each piece leans on a body of research:

• Checking email less often reduces stress. In a randomized experiment, people limited to checking email three times a day reported lower daily stress than when checking freely — and that lower stress carried through to better overall well-being (Kushlev & Dunn, 2015). Three briefings a day is essentially that prescription — with the willpower requirement removed.
• Interruptions are more expensive than they feel. Interrupted work still gets done — but at the cost of significantly more stress, frustration, and effort (Mark, Gudith & Klocke, 2008). And even brief switches leave “attention residue” that degrades your thinking on the task you return to (Leroy, 2009). Every peek at the inbox you don’t take during deep work compounds.
• Email-free stretches measurably deepen focus. When researchers cut workers off from email for a week, they switched windows less, stayed on tasks longer, and — by heart-rate monitoring — were less stressed (Mark, Voida & Cardello, 2012). A protected morning is a small, sustainable daily version of that.
• An honest caveat: batching alone isn’t enough. Simply batching email hasn’t reliably reduced stress in field studies (Mark et al., 2016) — plausibly because you’re still wondering what’s landing. And that worry is real: the mere expectation of having to monitor messages raises anxiety and measurably harms well-being — yours and your family’s (Becker et al., 2021). This is precisely what a trusted watcher changes: focused stops meaning exposed.
• Days that end, restore. Psychologically detaching from work at day’s end is among the most reliable predictors of recovery and next-day energy (Sonnentag & Fritz, 2015), while the felt pressure to keep responding undermines recovery and sleep (Barber & Santuzzi, 2015). The 4:30 close is what makes real detachment possible — without the “did I miss something?” tax.

The unlock was not “AI + email.” I think we’ve all tried that — some of us have even had it forced on us. It’s just another surface to check. The unlock was giving it two things: awareness of my time (when I’m at my best, when I’m open for collaboration, when the day should end) and what matters to me (my key relationships, my current priorities). With those, AI turns out to be a remarkable chief of staff — but only once it fully knows what it’s protecting.

The critical step is no longer technical. It’s being able to say clearly what your time is for. If you can, every tool in this new wave works better for you. If you can’t, most of them become one more surface to check and struggle to direct.

I walk through the how and why of my current setup as field notes in this video. If you’d rather skip the details and just hear the overview, it starts here.

​

My own setup has some extra requirements: I run multiple email accounts (for clear communication channels), I wanted local processing wherever possible, and I wanted to keep full control of any tasks or timeblocks it creates using the real-time features of MakeTimeFlow. You may not need any of that to get started. Here’s the quick-start recipe:

• First, make sure you understand the privacy restrictions on any email channel you might connect, and how to set the privacy settings for your AI agent properly. Here’s Anthropic’s guidance on using Claude Cowork safely. One caution worth knowing by name: email contents are an untrusted source, so letting them into your prompts opens the door to prompt injection — another reason your agent should draft, never send.
• If your privacy requirements allow:
  • Use Claude Cowork.
  • Use a connector to attach your email — and your calendar, if you want the time awareness.
  • Create a project (it keeps files and context together).
  • Ask it to create a tasks.md file for your active tasks, and a priorities.md file for your priorities — including your key relationships and current projects.
  • Then ask it to create three scheduled daily briefings. Use each topic below as a starter prompt, and let the AI help you build more detailed prompts tailored to you:
    • 7:00am — Morning briefing. Scan email for anything urgent that requires an immediate response, using priorities.md and tasks.md as guides for what matters. If nothing is urgent, just say so. If something is, explain what it is and why, and draft a response.
    • 12:30pm — Early-afternoon briefing. Scan email for anything that requires a response today. For anything that does, estimate the time it will take me to review and reply, and create a timeblock in available time on my calendar so I can respond in a single batch. Provide relevant context (previous emails, tasks, priorities), and draft a response for anything that needs one.
    • 4:30pm — Shutdown briefing. Scan email for any VIP or urgent responses. If there are none, just say so. If there is something, explain what it is and why, and draft a response.

If you want to run this with MakeTimeFlow, it gets a bit easier. You connect our MakeTimeFlow MCP server and plugin, which gives your agent training in the principles of cognitive high performance and shares (if you give it permission) your current success stories, tasks, and timeblocks, along with tools for time grounding and time calculations. Email me if you’d like help getting started with this.

Yours in flow,
Narath.