Simple AI Techniques to Enhance Daily Productivity

I've spent 25 years building software, and the thing that took me the longest to learn had nothing to do with code. Most of the time we lose in a day doesn't come from the hard work. It comes from the small stuff done badly: rewriting the same email three times, second-guessing a Slack message before you hit send, staring at a calendar that looks like a losing game of Tetris. AI hasn't fixed my whole day. It fixed those small things, and that's turned out to matter more than I expected.

I'm not talking about theoretical AI productivity techniques here. These are the three things I actually use, every day, and where I've seen them help and where they haven't.

Rewriting Emails and Slack Messages Without the Second-Guessing

I write a lot of messages in a day, emails to clients, Slack replies to a team, notes to people I've never met in person. The tone has to shift depending on who's reading it, and typing fast doesn't leave much room to catch a clunky sentence or a line that reads more clipped than I meant it to.

I run drafts through Claude before sending anything that matters. Not to write the message for me, but to check grammar and make sure the tone lands the way I intend. A quick note to a coworker doesn't need the same voice as a message to a client, and it's easy to get that wrong when you're moving fast. Having something catch that before I hit send has saved me from more than one message I'd have wanted to walk back.

Here's roughly the prompt I use:

Review this message for grammar and tone. I want it to sound
professional but not stiff. Point out anything that could come
across the wrong way to [coworker/client], and suggest a fix.
Don't rewrite the whole thing unless something is actually broken.

[paste your draft]

That last line matters. Early on I let AI rewrite everything, and messages started sounding less like me and more like a press release. Asking it to flag problems instead of rewriting on autopilot keeps my own voice in the message, which is the whole point.

If your problem is less "how do I sound" and more "my inbox is a landfill," that's a different tool. Something like SaneBox works further upstream, sorting and prioritizing what lands in your inbox before you ever open it, rather than cleaning up what you write.

A Second Set of Eyes on Code

I also use AI for code review, and this is where "trust but verify" earns its keep. AI is genuinely good at catching things a tired brain skims past: an unclosed edge case, a variable name that doesn't match its use, a function doing three jobs when it should do one. It's a fast, patient reviewer that never gets tired on the fifth pull request of the day.

What it isn't is a replacement for understanding your own code. Software is 90% thinking and 10% typing. AI can speed up the typing and even sharpen the thinking, but it can't decide what the system actually needs to do. That's still the developer's job. I've caught AI confidently suggesting a "fix" that solved the wrong problem because it didn't have the full context of why the code was written that way. If I hadn't understood the logic myself, I'd have shipped it.

A prompt I lean on for review:

Review this code for bugs, edge cases, and readability. Don't just
tell me it looks fine, tell me what you'd change and why. Flag
anything that seems inconsistent with the rest of the function.

[paste your code]

Used this way, AI code review is a genuine boost to daily productivity. Used as a substitute for actually reading your own code, it's a liability with good grammar.

Turning Your Calendar Into a Daily Briefing

The third habit is the one that's changed my mornings the most. I have a custom GPT that pulls my calendar and gives me a daily briefing: what's on the schedule, where meetings are stacked too close together, and what actually needs my attention versus what can slide. Instead of opening my calendar cold and reacting to whatever's first, I start the day already knowing where the pressure points are.

This connects to something I believe about how work should get built in general: work backwards. Figure out what the day actually needs to produce, then work out how to get there, instead of just moving through whatever shows up first. A calendar briefing does that same thing for time management. It turns "what's on my calendar" into "what does today actually require of me," which is a different and more useful question.

If you want to try building something similar, a simple starting prompt looks like this:

Here's my calendar for today: [paste or describe it]. Give me a
short briefing: flag any back-to-back meetings with no buffer,
call out anything that looks like it needs prep beforehand, and
tell me what the one or two priorities should be if the rest of
the day gets eaten by meetings.

You can run this manually each morning with ChatGPT or Claude, or build it into a recurring custom GPT once you know the format you like. If you'd rather not build the automation yourself, platforms like CustomGPT.ai are built for exactly this, turning a prompt like the one above into something that runs on its own schedule.

Mistakes to Avoid

• Treating AI edits as final. Trust but verify applies to writing and code both. Read what it suggests before you send it or ship it.

• Pasting sensitive information into rewriting tools. Check your company's data policy before running client details or proprietary code through any AI tool.

• Accepting code review feedback you don't understand. If AI flags something and you can't explain why it matters, dig in before you fix it. You're still responsible for the code.

• Using one generic prompt for everything. A prompt built for a formal client email won't serve a quick Slack message well, and vice versa. Adjust the ask to the context.

• Letting a daily briefing replace actually looking at your calendar. It's a head start, not a substitute for judgment about your own day.

None of this requires a technical background. It requires picking one small, repeated frustration, a message you rewrite too many times, a bug you keep missing, a calendar you keep getting ambushed by, and handing that one thing to AI first. Once it's working, the next one is easier to spot.

Try the prompts above this week. Swap in your own message, your own code, your own calendar. The goal isn't to use AI for everything. It's to stop losing time to the small stuff so there's more of it left for the work that actually needs you.