My morning AI brief arrives at 8:00 a.m. with my meetings, who I’m meeting, why we’re meeting, what we should discuss, the weather, and considerations for managing my energy. Same categories each day. I start with a usable overview.

I don’t have a time-saved number for it, and I won’t spec the build. Pick your categories, use data you’re authorized to use, and check for anything missing or stale.

I teach a live masterclass around three skills that matter while the tools change. Clear inputs define the job. Workflow thinking turns the job into steps you can repeat. Curiosity is how you improve when the result misses. The gap I keep coming back to is knowing how to use what you already have.

Which skill is your bottleneck? Practice that one on a real task.

Clear inputs

When I don’t know what to ask, I flip the script. I give a rough goal and add, “Ask me questions before we proceed.” Answering concrete questions is easier than writing a perfect brief from a blank page.

Before: “Write me a newsletter.”

After: “I want a newsletter. Ask me questions before we proceed.”

For a newsletter, those questions can pull out the audience, the tone, the length, the scope, and the visuals. What would you have said if it had just asked? Repeat the interview and you learn which details to put in the first message next time.

Give the output a boundary. I use limits like “answer in one sentence,” “put this in a one-page document,” and “use only these source files.” Sometimes I say to reference only selected files. A clear limit describes the successful deliverable and leaves you less to sort through.

Rewrite one disappointing prompt with a single output constraint and a single source constraint. Then check that the output actually followed the sources. The boundary clarifies the assignment. It does not certify the facts.

In my client automation work, much of the time goes into cleaning data, organizing files, labeling information, and putting it in the right place before the AI does anything useful. A focused set of relevant documents beats a pile of unrelated PDFs. Context preparation is the work. Prepare a small, clearly labeled source folder for one recurring task, with only the files that task needs. Clean files still need a check, and a messy folder does not explain every wrong answer.

Ask Me Questions Before We Proceed

Workflow thinking

This one starts with a familiar move. Break the goal into steps. Name the input, the repeatable steps, and the output.

I use it for a short podcast script. Input: a topic. Steps: a consistent outline, plus voice and story guidance I’ve already established. Output: a draft. I keep a customized assistant for those recurring episode scripts. Personal stories have to be examples I supply and can verify.

A meeting summary is a midpoint. Does that summary just sit there? The workflow I teach takes a transcript, extracts action items, adds them to a task system, and prepares a follow-up email for review. The next steps land where the work happens, admin handoffs included. Check the tasks and the email before anything is assigned or sent.

Map it on paper: input, intermediate steps, destination, human approval. Get consent before transcribing, and limit access. Pilot one meeting. This is a class example. I don’t have measured savings from a client deployment to show you.

Discover a sequence that helps, then streamline it. I have no time-saved figure, and several models agreeing is not a fact check. Test one proposal with a second model. Write down the useful differences. Document the sequence if it earns a repeat. Automate after that.

I teach three levels. Assistance is the tool working with you, back and forth. Workflows connect repeatable steps into an outcome with less manual effort. Agents take on delegated work for you, without you starting every step. Get one task reliable, turn that conversation into a workflow, and only then look at more autonomy. Set permissions, review points, and monitoring if you delegate, and plan on checking the work.

A repeatable process can be the whole next step. For a morning brief, sketch the template and start with calendar events plus meeting prep.

Curiosity

Before you abandon a weak answer, spend one more exchange. Ask which sources it used, which files it touched, what assumptions it made, and why it chose that approach. Correct the brief or the sources and try again. Test whatever it tells you against the evidence. That reply is not a window into the model’s insides. Curiosity feeds the next attempt. You still have to check the result.

Have you been repairing the same email by hand every time? If you rewrite an AI email outside the chat, the tool never sees the wording you cut, the tone you wanted, or the structure you kept. Compare one draft with the final email. Save the differences as a short checklist of wording, sentence structure, tone, and one approved example. Put that checklist in the workflow. The notes do not retrain the model, and they do not follow you into a new chat unless you bring them.

Pick the skill that’s slowing you down and use it on one real task today. If yours is clear inputs, open an unfinished email, a trip plan, or a first draft and add, “Ask me questions before we proceed.” Answer what comes back. Save the details you should have given it upfront.

Recommended for you

Want the full system? 25X is the flagship productivity system we teach.

Explore 25X →

You may also Like

Last Updated: November 18, 2026

Read More

Last Updated: November 17, 2026

Read More

Last Updated: November 3, 2026

Read More

ABOUT THE AUTHOR

Thanh Pham

Founder of Asian Efficiency where we help people become more productive at work and in life. I've been featured on Forbes, Fast Company, and The Globe & Mail as a productivity thought leader. At AE I'm responsible for leading teams and executing our vision to assist people all over the world live their best life possible.


Leave a Reply


Your email address will not be published. Required fields are marked

{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}