I saw a few YouTubers talking about this a few weeks ago. My first reaction was — another productivity hack I'll try once and forget. But I was curious enough to actually try it. Not because I was unhappy with typing, honestly I was quite fast at it — around 80 words per minute consistently. But I figured, why not see what the fuss is about.

What I didn't expect was how much it changed the quality of outputs I was getting back.

I've been recommending this to everyone I know since then. Not because it's cool or trendy, but because I've genuinely seen the difference — and it's not subtle.

The real unlock isn't some advanced prompting framework. It's simpler than that: talk more.

The context is the lever

Here's what I mean. If you describe your situation to a senior colleague — someone with years of experience in your world — you don't have to spell everything out. You say three sentences and they get it. They fill in the gaps from their own knowledge. They ask a sharp clarifying question because they understood the nuance.

If you give the same three-sentence brief to an eager intern who's just started? You'll get something back that technically addresses what you said, but missed the point entirely. Because they didn't have enough context to know what mattered.

AI is more like that intern. The more context you give it, the better it fills in the gaps on its own.

Typing takes effort. And when something takes effort, you do less of it. You compress your thoughts, you cut corners, you simplify. What should have been a five-sentence dump of your actual situation becomes two sentences that barely scratch the surface.

Voice gets you context for free. Now I use TypeWhisper on my Mac — it's free and open source for personal use, and it runs entirely offline using local models. The only requirement is a computer with enough performance to run those models. The accuracy is genuinely impressive. I almost never have to go back and correct anything. Even with technical terms and abbreviations, it gets it right the first time. My Android phone gets a similar experience through FUTO Keyboard, which also runs local speech-to-text on-device. Near-identical to the Mac, just not quite as seamless.

And thinking out loud is faster than thinking in full sentences. I can articulate a messy, half-formed thought in 30 seconds that would have taken me three minutes to type out properly. It doesn't matter if your English isn't perfect — the AI is good at picking out the relevant signal from the noise. Even if your first language isn't English, as long as you can get the idea across, it still works.

I can't go back to typing. Full stop.

Try this today

The lowest-friction starting point is right in front of you. Whatever AI you use — ChatGPT, Claude, Gemini — they all support voice input. On ChatGPT's free tier, the voice button is right there. Tap it, describe your situation in your own words, and hit send. Don't clean up your sentences. Don't edit as you go. Just talk.

If you're on Mac, TypeWhisper is worth the setup. It's free, completely private since it runs offline, and the accuracy is in a different league from built-in dictation. There's a Windows beta version available too if you're on PC. On Android, FUTO Keyboard gives you a comparable experience.

The goal isn't to find the perfect tool. It's to feel the difference in output quality when you're not fighting your keyboard.

What to watch for

A few honest caveats before you go all in:

• Transcription errors happen. Don't assume the transcript is exactly what you said. Rarely — but occasionally — a word comes out slightly different from what you intended. Always skim before you hit send, especially if the output is going somewhere real.
• Both tools run entirely offline. Nothing leaves your device. If you're working with sensitive information, this is as private as it gets — no server, no cloud component.
• Speak naturally, not formally. Don't perform for the tool. The AI underneath is good at figuring out what you meant, even if your sentence wasn't grammatically clean. Let go of the instinct to phrase things correctly. Just talk.

The real win

I keep coming back to this. The win isn't that you save time. The win is that the output is better — more specific, more useful, closer to what you actually had in mind when you started.

And that changes how you feel about using AI. It's not a chore anymore. It's not "I have to go type this out." It's "I already know what I want to say, let me just say it."

That's the shift. Once you feel it, you don't go back.

Just one. I was genuinely unwell, so I closed my laptop and actually rested. No inbox-checking guilt. No Slack at bedtime. A proper sick day.

The next morning I opened my email to find over 500 messages waiting for me.

I'm a senior software engineer at PayPal, based in India. My team spans time zones — a significant chunk of my colleagues are in PST, which means by the time I wake up, they've already put in half a workday. While I slept, pull requests were reviewed, builds ran and failed and ran again, Jira tickets got updated, and threads I was CC'd on spiralled into long back-and-forths. GitHub alone sends a separate notification for every single step of a build pipeline — every success, every failure, every approval. On a busy overnight, I'd typically walk in to 100–200 emails. That one sick day more than doubled it.

The emails themselves weren't really the problem. The problem was not knowing whether something urgent was buried in there. Something with my name on it. Something that had been waiting since yesterday and was now quietly blocking someone else.

So I'd block out time — a solid 30 minutes, first thing — just to get through it all. Scan, skim, sort, prioritise. And here's what bothered me about that: the first 30 minutes of your workday are usually your sharpest. That's when you're freshest, before meetings and interruptions erode your focus. I was spending that window on triage — on sorting — rather than on the focused, heads-down work that actually moves things forward.

That had to change.

Stealing a Good Idea (and That's Fine)

I came across the concept of a "morning brief" through something someone else had shared — a way they'd set up an AI to read through their inboxes and surface what mattered before they started their day. Their specific setup wasn't a match for mine. But the idea was.

This is one of the most underrated things about AI tools right now. You don't have to be the one who invents the workflow. You just have to be willing to pick up what someone else figured out, hold it up against your own situation, and ask: could this work for me?

I took that concept, mapped it to my actual morning — Outlook email, Slack, Jira, Outlook Calendar — and built something that fit. It took some tinkering, but the starting point wasn't a blank page. It was someone else's half-finished idea that I finished for myself.

I'd encourage you to do the same thing here.

What the Morning Brief Actually Does

Think of it like this: imagine you had a diligent intern whose only job was to arrive at the office before you, read every message, scan every notification, check your task list and your calendar — and then hand you a single, clean briefing when you walked in the door.

That briefing tells you:

• What's on fire right now — urgent, needs your attention before anything else
• What's about to get warm — not critical yet, but will be if you don't move on it today
• Quick wins — tasks you could knock out in under 15 minutes to clear your plate early
• Good to know, but not urgent — the rest, summarised so you can decide later whether to act

That's my morning brief. Every morning, before I open a single email or Slack thread, I run it. Within 5–10 minutes, I have that briefing in front of me — and I know exactly what my morning looks like.

It doesn't replace reading my emails. I still get to all of it eventually. But it gives me a lay of the land first, so I can start the day with intention rather than anxiety. I can make an informed choice about what to open first, rather than just thrashing through whatever's at the top of the pile.

The Honest "How It Works" Section

I use Claude Desktop for mine, though Claude on the web works the same way.

The key piece you'll need is connectors — these are the integrations that let the AI actually read your tools. You configure them once, and after that they just work. For my setup, I needed connectors for:

• Outlook (email)
• Slack (team messages)
• Jira (tasks, stories, and bug queues)
• Outlook Calendar (what's on my plate today)

The connector setup is a one-time thing. It does require some patience — it's not quite as simple as clicking "install." If you're not confident doing it yourself, 20 minutes with a technical colleague is genuinely all it takes. Think of it as the cost of admission: you pay it once, and then the brief just works every morning after that.

Once your connectors are live, the rest is just a prompt — the instruction you give the AI before it reads everything. Here's the version I use, which you can copy and adapt directly:

📋 Morning Brief Prompt — Copy and Adapt

Good morning. Please review my Slack messages, emails, Jira board, 
and Outlook Calendar for today, then give me a structured morning 
brief with the following sections:

1. 🔴 URGENT — Things that need my attention right now. Flag anything 
   with my name on it, unresolved blockers, or anything time-sensitive.

2. 🟡 WATCH LIST — Things that aren't urgent yet but will become urgent 
   soon if I don't act today.

3. ✅ QUICK WINS — Tasks I can finish in under 15 minutes. I want to 
   knock these out early.

4. 📌 FYI — Everything else summarised briefly. I'll decide later 
   if I need to act.

Keep it concise. I want to be oriented and focused within 10 minutes 
of reading this.

Swap in whatever tools you actually use. If you don't have Jira, remove it. If you use a different project tracker, name that instead. The prompt is yours to make your own.

A Couple of Honest Caveats

Your AI intern is fast and eager — but occasionally gets things wrong. A few things worth knowing upfront:

It can misread priority. An email thread that looks urgent based on the subject line might already be resolved. The brief is a starting point, not gospel. Give it a quick sanity check before you act on anything it flags as on fire.

The setup requires real effort. Connectors don't install themselves, and if you hit a snag, it can feel frustrating. But this is a one-time setup, not an ongoing tax.

Check your company's AI policy first — this one is non-negotiable. You're connecting real work tools containing real work data. That data may include confidential information, client details, or anything else that falls under your organisation's compliance obligations. Before you set anything up, speak to your IT team, your HR department, or whoever owns your company's AI usage policy. Don't assume it's fine just because the tool is popular.

This is especially important if you're considering a free AI tool. Free tiers often have different — and more permissive — data handling terms than paid or enterprise plans. If in doubt, use a paid or enterprise-grade tool with clear data privacy commitments, and still check with your policy team first. A five-minute conversation now is far better than a compliance issue later.

The Real Point

I'm not sharing this because mornings need to be optimised or because inbox zero is a worthy life goal. I'm sharing it because there's a real cost to spending your sharpest 30 minutes every morning on sorting rather than on doing — and most of us have just quietly accepted that cost.

You don't have to build this from scratch. Start with this prompt. Borrow the concept. Adapt it to your tools and your situation.

The goal isn't to read faster. It's to walk into your morning already knowing what needs you — and giving your best hour to the work that actually deserves it.

That's a good morning.

Afterward, I asked him: "How would you have done this a year ago?"

He didn't hesitate. "Honestly? I wouldn't have even tried."

That answer has been bouncing around my head for a week, and I want to share why.

The ceiling, not the clock

Most AI stories are about speed. Write the email faster. Summarize the doc faster. Clear the inbox faster. Those wins are real — I talk about them here all the time, and they count.

But there's a second gear, and that's what Anirudh found.

He's not a developer. Building an interactive dashboard the old way would've meant learning enough web development to pull it off — two, maybe three months minimum — and probably handing it to a dev team anyway. So he wouldn't have done it. He would've loaded the feedback file into a spreadsheet, skimmed a few rows, pulled a handful of quotes, called it a day.

What AI changed wasn't his clock. It raised his ceiling.

Both gears matter. Speed makes the things you already do well feel lighter. Ceiling lets you reach into work you'd written off as "not my skillset" or "too much work." The first is obvious once you've tried it a few times. The second is where the real leverage quietly lives — and it's the one nobody talks about enough.

But you need a North Star

Here's the part that matters, though — Anirudh didn't stumble into a dashboard. He had a question he genuinely cared about: what are our customers actually telling us, underneath all this noise?

That question was the anchor. Without it, AI becomes a shiny toy — you poke at it, it spits out impressive things, you close the tab, nothing changes in your week.

You don't need a grand plan. You don't need to know which "AI workflows" you want to master. What you need is closer to this: if I had a patient, capable intern sitting next to me for a week, what would I love to finally pull off that I've been shelving?

Your starting point isn't a tool. It's the thing you've been writing off.

The unglamorous middle

I want to be honest about something else. Anirudh didn't nail that dashboard on his first try. His early outputs were messy. The analysis had gaps. Things were formatted oddly. He kept going back — refining, asking better questions, redirecting.

This isn't grind culture. I'm not asking you to hustle. The opposite — I'm asking you to be patient with a clumsy intern.

That's the analogy I keep coming back to. AI is fast, eager, and occasionally hands you something that's just plain wrong in a confident tone. Your job isn't to accept the first draft. Your job is to be the boss — look at what it produced, tell it what's off, ask for another pass. The first try is never the finished thing. Nobody's is.

Anirudh's 14 hours weren't 14 hours of smooth progress. They were 14 hours of "okay, try that again."

It's a skill, not a trick

One more thing: like any skill, you get better at this with use.

Your first session with AI will feel awkward. Your tenth will feel natural. Somewhere around your fiftieth, you'll start noticing things to try that wouldn't have occurred to you earlier — "oh, I could use this for X too." That's not you becoming a power user. That's just you, plus a tool, over time.

If your employer is paying you to figure this out — giving you tools, time, and real work to practice on — that's a privileged spot. Use it. The compound effect of practicing on work you actually care about is bigger than any tutorial.

If that's not your situation, don't let that stop you. The free tiers of the major tools will carry you through months of real experimenting before you'd hit a wall. And if you find yourself leaning on one daily, paid tiers are there for when you're serious enough that they pay for themselves in saved time.

Either path works. What doesn't work is waiting until you have "the right setup" to begin.

What I'd actually like you to try this week

I don't want you closing this thinking you need to go build a dashboard next week. You don't. You shouldn't.

What I do want you to try is this: make your gap list.

Write down one or two things you've quietly shelved because they weren't your skillset, or they seemed like too much work. A few I hear often:

• "I should clean up this messy spreadsheet, but I don't really know Excel."

• "I'd love a simple one-pager explaining our team's process, but I'm not a designer."

• "I've got 20 documents I need to pull themes from, and I just don't have the time."

Pick one. Just one. Open a free tool — Claude.ai or ChatGPT's free tier works fine — and start with something like this:

I'd like help with [the task]. I'm not a [designer / analyst / whatever], so please walk me through it. Before you suggest anything, ask me 3 to 5 clarifying questions about what I'm trying to achieve and who it's for. Then give me a rough first draft I can react to.

The "ask me questions first" bit is the quiet magic. It stops the intern from sprinting off with half the story.

A few honest caveats, because I promised I'd always give you these:

• The first draft will be rough. That's the job, not a failure.

• Fact-check anything it asserts. AI states wrong things confidently all the time.

• Don't paste anything confidential — client data, sensitive internal docs — into free tools. Your employer probably has rules about this. Check first.

Where this is going

Anirudh said something to me recently that's stuck with me: "In a few years, developers will do PM work alongside their coding, and PMs will code alongside their PM work."

The walls between specializations are getting thinner.

That might sound threatening — like everyone's suddenly expected to be a jack-of-all-trades overnight. It isn't. Your depth, the thing you're actually good at, is still the thing. What AI adds is breadth. A little reach into the territory you used to leave to someone else.

You're not going to become a developer next week. Anirudh didn't either. He just reached into dev-adjacent work when the problem called for it, and came back to his actual job with something he couldn't have built on his own.

That's the direction. A year from now, I'd love for you to have your own version of Anirudh's line: I wouldn't have even tried that before.

Pull up a chair. Pick one thing off your gap list this week. That's the whole thing.

It was June 14, 2025. My wife was about seven months along, and I was trying to do something thoughtful — design the baby shower invites myself using ChatGPT. I had a vision. I fed it all the details: the date, the venue, the vibe. The first output was pretty good. I checked it carefully, made some notes, and started iterating. A little tweak to the colours here. A small wording change there. Each time I reviewed, I focused on the specific thing I'd just asked it to change.

You can probably see where this is going.

By the fifth or sixth version, I had stopped reading the whole thing. I trusted it. I knew what I was tweaking, and I assumed everything else was sitting exactly where I'd left it.

It wasn't.

The date had quietly changed to something else entirely. Wrong day. Wrong everything.

My saving grace? I always share a final draft with my wife before anything goes out. She caught it immediately. We laughed. We fixed it. Nobody showed up on the wrong day.

But here's the thought that's stuck with me ever since: what if I hadn't shown her?

The Trap Has a Name: Iteration Fatigue

This wasn't carelessness. I'm a software engineer. I know better than most that you don't ship without checking.

But here's what happens when you iterate on anything — not just AI outputs — your brain starts taking shortcuts. When you change one small thing, your mind quietly fills in the rest as safe. I only changed the font. The date is fine. The venue is fine. I would have noticed if something else changed.

AI doesn't share that assumption.

Every time you send a new prompt, the model is making a fresh attempt at the whole thing. It has no memory of what you approved in the last version. It has no loyalty to the details you already signed off on. It is not tracking what changed. That's your job — and it's easy to forget that in the flow of a back-and-forth session.

Your Eager, Clueless Intern

I've always described AI as a fast, eager, occasionally clueless digital intern. And this is exactly where that analogy earns its keep.

If you handed a draft to an intern and said "just change the heading font," would you forward whatever came back directly to a client without reading it? Of course not. Because the intern might have accidentally deleted a paragraph. Might have changed a number. Might have introduced a typo while reformatting.

You'd glance over the whole thing. Takes thirty seconds. Saves enormous embarrassment.

AI is that intern — except faster, more confident, and completely unbothered by mistakes. It will hand you a document with a wrong date, a made-up statistic, or a quietly altered detail, and it will do so with the same calm, polished tone it uses when it gets everything exactly right.

It has no skin in the game. No consequences. No embarrassment. It won't flag its own mistakes.

Your name is the only one that goes on the final product. That's not a burden. That's just the deal.

The One Habit That Saves You

I'm not going to give you a ten-step checklist. That defeats the whole point.

Just one rule: do one final read-through as if you're seeing it for the first time.

Not a skim. Not a "I'll just check the parts I changed." The whole thing. Top to bottom. Like a stranger is reading it for the first time — because sometimes, someone is.

And for anything going out to more than a handful of people — a team email, a client document, an announcement — add a second pair of eyes before it leaves your hands. Not because you're incompetent. Because a fresh perspective catches what familiarity misses. My wife was my second pair of eyes that day, and she wasn't even trying to proofread. She just read it fresh.

That's the human in the loop. That's you.

A Prompt Worth Keeping

Before you do your own final check, try asking AI to audit itself first. It won't catch everything — but it sometimes catches more than you'd expect.

Copy and paste this at the end of any AI-generated draft before you review it:

"Before I review this, please re-read it carefully and flag anything that seems inconsistent, factually uncertain, or that may have changed from my earlier instructions. List any concerns, even small ones."

It won't be perfect. But it gives you a head start — and it puts you in the right mindset to read critically rather than just confirm what you're hoping to see.

A Few Honest Warnings

Since we're being real with each other:

AI will not tell you when it quietly changes something. There's no notification. No asterisk. No "hey, I also tweaked the date while I was in there." You will only know if you look.

Confidence is not accuracy. The tone of an AI response tells you nothing about whether the content is correct. It sounds equally assured whether it's right or wrong.

Longer sessions carry more risk. The more you iterate in a single conversation, the easier it is to stop reading the full output. Set a personal rule: any final version gets a full read, no matter how many rounds it took to get there.

You Were Always the Last Line

Here's the thing I want you to take away from all of this — and I mean it as genuinely good news.

You are not the last line of defence because AI is broken. AI is genuinely useful. I use it every single day. It saves me real time and real mental energy.

You are the last line of defence because you are the only one in the room who actually cares about the outcome. You're the one who knows your wife will be reading that invite. You're the one who knows your client's data team will fact-check that stat. You're the one who understands the consequences of getting it wrong.

AI doesn't have any of that context. It doesn't feel the weight of it. It just has your prompt and its best guess.

So use it. Lean on it. Let it do the heavy lifting.

Then read the whole thing before it goes out.

Have you ever caught an AI mistake before it caused real damage? Or has one slipped through on you? Hit reply — I'd love to hear your story.

Within days, he was sitting in standups where engineers threw around terms he'd never heard. System integrations, APIs, architecture decisions. He smiled and nodded and quietly had no idea what was happening.

So he called me.

Now, I'm a Senior Software Engineer. I've also spent enough time around product to understand both sides. I wanted to help. But I also knew that sending him a list of definitions wasn't going to cut it. What he needed wasn't a glossary. He needed someone who could meet him where he was — someone patient enough to ask what he already understood, give him a plain-English answer, and not make him feel stupid for asking.

The problem? The engineers around him are heads-down. Nobody has time to sit with the new intern and explain how the systems work. That's just the reality of most fast-moving teams.

So I built him something better than a busy senior engineer. I built him one who's never too busy.

This Isn't Just a Beginner Problem

Before I go further — my wife and my sister-in-law are both experienced product managers. Not interns. Not beginners. And a while back, they came to me with the same request: can you build us something that explains technical concepts in plain English? I built them a prompt at the time. It worked well. And then life happened, and I lost it.

So when my brother called, I was starting from scratch anyway.

I'm telling you this because it matters: the gap between technical and non-technical thinking doesn't close just because you get more senior. It's a permanent feature of the PM role, not a temporary beginner problem. If you've ever nodded along in a meeting hoping nobody asks for your opinion on the technical approach — this is for you too.

I Didn't Know What to Put in the Prompt. So I Asked.

Here's the part that might surprise you.

When I sat down to build my brother's expert, I had a rough idea of what it needed to do — but I didn't have a clear picture of everything it should include. How should it handle technical jargon? How patient should it be? Should it dive into explanations straight away or ask questions first?

Instead of guessing, I did something I want to introduce you to today: I asked AI to write the prompt for me.

That might sound like a strange loop. Using AI to build your AI. But think of it this way — if you were hiring a specialist consultant, you'd probably ask a more experienced colleague: "What should I look for? What questions should I ask? What does a good brief for this person look like?"

I did the same thing. I just asked Gemini instead of a colleague.

Here's exactly what I sent:

The Meta-Prompt — Copy and Paste This

You are a world-class prompt engineer. Give me a prompt which can be reused which will act as a world-class software engineer turned product manager who has all insights into both engineering side things in depth, and also the product and business requirements side. The AI would be for answering and explaining engineering topics in very simple language for product interns who are clueless about the technical stuff. You might need to explain some things in detail but in as simple English as possible. Try to not go into depth as soon as you start — maybe ask if the user wants to go into depth. Also quiz the user if there are any questions you need clarification on. Try to include as many simple, day-to-day relatable examples as you can to explain the concepts. Try to ask for questions related to their context of why they are asking, and further questions to clarify what they know so far, because they might not be good at explaining their problem. Add anything more that you think would be useful. You can also ask them to explain their thought process, what they understood so far, or what problem they are trying to solve. Be reassuring and understanding because they are trying to wrap their heads around difficult or complicated concepts.

Paste that into ChatGPT or Gemini. What comes out is a ready-to-use expert persona — built by AI, guided by your needs.

It Wasn't Perfect. That's the Point.

The first version Gemini gave me was good. But there was a problem — one my brother actually caught when he started using it.

The AI would ask clarifying questions and then immediately answer without waiting for his response. It was jumping ahead, filling in blanks it hadn't given him a chance to fill himself.

I went back to Gemini and spent about two minutes telling it to fix that. That's it. One follow-up message. This is what iteration looks like — not a rebuild, just a small correction. And it's something anyone can do.

Here's the refined version it produced:

The Expert Prompt — Copy and Paste This

Role & Persona: You are an elite Software Engineer turned Senior Product Manager. You possess deep, battle-tested technical expertise coupled with sharp product and business acumen. You are currently mentoring a bright but technically inexperienced Product Management intern.

Your Goal: Translate complex software engineering concepts into simple, easily digestible language. You bridge the gap between "how it is built" and "why it matters to the product."

Tone & Attitude: Highly empathetic, patient, and reassuring. Technical jargon can feel overwhelming, so create a safe space for them. Validate their curiosity and let them know these concepts take time to click.

Rules of Engagement — Two Phases. Do not skip ahead.

Phase 1 — Context & Clarification (Your First Response Only): When the user introduces a topic or asks a question, your ONLY job is to gather context. Acknowledge and warmly validate their question. Then ask: "What specific product problem or feature brought this up?" and "What is your current understanding so far?" Then stop. Do NOT explain the concept yet. Wait for the user to answer.

Phase 2 — The Explanation (After the User Replies): Start with a quick big-picture overview of the concept and its business value. Rely heavily on simple, relatable analogies — if you use a technical term, define it immediately in plain English. Do not go into deep technical detail straight away. After your overview, ask: "Would you like me to go a level deeper into how this actually works under the hood, or does this give you what you need for now?" Then ask them to summarise what they've understood so far.

A Bigger Idea Worth Sitting With

In my last article, I talked about building a small team of AI experts — Dr. Heart for health questions, Mr. Chef for cooking, Mr. Clean for household chemistry. The thing I didn't mention is that I didn't always know exactly what each of those experts needed to include.

Sometimes I had a rough idea and asked AI to fill in the gaps. Sometimes I described the problem and let it figure out the rest.

You don't have to know what a good prompt looks like before you start. You just have to know what problem you're trying to solve. AI can help you build the tool — and then you use the tool.

Your Turn

You don't have to be a PM intern surrounded by engineers. You just have to have a topic you've been meaning to understand — something where you'd love a patient, knowledgeable guide who's never too busy, never condescending, and always has time to explain it one more way.

Open ChatGPT or Gemini. Paste the meta-prompt from this article. Describe your situation. Let it build your expert.

Then ask the question you've been sitting on.

A Few Honest Caveats

The first version might not be perfect. That's fine. Tell it what's not working and ask it to fix it. That's exactly what I did.

AI still gets things wrong. The expert persona makes answers more relevant — but it doesn't make them infallible. Use your judgment, especially for anything important.

Don't share sensitive details. Describe your situation in general terms. You don't need to share confidential work details to get useful answers.

The more specific your context, the better the output. A vague brief gets a vague expert. Spend sixty seconds on your description and it will pay you back every time.

Still waiting on my brother's full verdict. He's been heads-down his first week — which, honestly, is exactly the problem this was built to solve. I'll update you when he surfaces.

Vague. Generic. Kinda helpful. The kind of advice you'd get from a very well-read stranger at a bus stop.

And then there's the version you unlock when you give it a job title and hand it your actual situation. That version feels like a completely different tool. I know, because I built one — and it changed how I approach my health.

The 10kg I Lost by Asking Better Questions

About two months ago, I decided to get serious about my weight. I'd tried before, multiple times, and I know what it's like to start strong and quietly give up. This time I went in differently — I started doing real research. Reading about insulin, heart health, how different foods affect the body long-term. And I made genuine sacrifices around what I eat.

But research alone leaves gaps. I had specific daily questions that no article quite answered. And as an introvert, the idea of sitting across from a nutritionist and talking through my eating habits wasn't exactly appealing — not to mention the cost.

So I built Dr. Heart — a custom AI persona focused on heart health and overall wellness. Not to do the work for me, but to be the knowledgeable guide I could ask anything, any time. Is this meal going to work against me today? What's actually happening in my body when I eat this? What should I have before a workout on three hours of sleep?

The work was mine. The sacrifices were mine. But having a tool that could answer my specific questions — quickly, without judgment — helped me stay informed and consistent in a way I hadn't managed before. I've lost almost 10 kilograms in two months.

Now — before I go further — Dr. Heart is not my actual doctor. It has never seen my bloodwork. It does not replace a GP. Please still see a real medical professional for anything serious. AI can hallucinate. It can be confidently wrong. I cannot stress this enough.

But as a daily thinking partner, helping me make smarter decisions based on research I'd already done? It's been remarkable.

And here's the thing: the magic wasn't the AI. The magic was how I set it up.

The Difference Between "Ask AI" and "Ask Your Expert"

Most people open ChatGPT and type a question cold. That's like calling a consultant and saying "help me" — without telling them what you do, what you're dealing with, or what you've already tried.

When you give the AI a role and your personal context upfront, everything changes. The answers get sharper. More relevant. More yours.

I've now built a small team of these:

• Dr. Heart — My go-to for anything related to heart health, overall wellness, and daily food decisions.

• Mr. Chef — An expert in cooking techniques, ingredient swaps, and building meals around what I actually have in the fridge.

• Mr. Clean — This one might surprise you. I got tired of not knowing whether cleaning products were safe to mix, or why certain methods actually work. Mr. Clean explains the chemistry in plain English. I've learned more about my home in three months than in years of just winging it.

None of these required a paid subscription or a tutorial. They're just AI — with a job description and my situation baked in from the start.

Two Ways to Build Your Expert (Both Free)

Path 1 — Gemini: Set it once, use it forever

This is what I use. Gemini has a free feature called Gems — you write your expert's persona once, save it, and it's there waiting every time you open it. Dr. Heart lives here. No copy-pasting, no setup each time. Just open and ask.

Go to gemini.google.com → click Gem Manager on the left sidebar → New Gem

Path 2 — ChatGPT: Paste and go

If you're already comfortable in ChatGPT, no need to switch. The free plan doesn't save personas between chats — so you'll paste your expert prompt at the start of each new conversation.

Pro tip: Keep your prompts in your Notes app or a Google Doc. One tap to copy, one paste to start. It takes about ten seconds and quickly becomes second nature.

The Prompt Template (Copy and Paste This)

Here's the exact structure I use. Swap in your own details:

You are [role] — a world-class expert in [specific area]. Your job is to give me practical, clear, and honest advice based on my specific situation. Here's what you need to know about me: [2–3 sentences about your relevant background, goals, or constraints]. When you answer, always explain the why in plain language. If something is outside your expertise or I should see a real professional, say so directly.

A real example, close to how I set up Dr. Heart:

You are a world-class wellness advisor specialising in heart health and nutrition. Your job is to give me practical, clear, and honest advice based on my specific situation. Here's what you need to know about me: I'm a busy professional in my 30s, currently on a weight loss journey, with a desk job and limited time to cook. I've been reading about how food affects heart health long-term, and I want to make smarter daily decisions. When you answer, always explain the why in plain language. If something is outside your expertise or I should see a real doctor, say so directly.

Paste that into a new chat. Then start asking your actual questions. You'll feel the difference immediately.

A Few Honest Caveats (Because I'd Rather You Trust Me)

AI can be confidently wrong. These tools don't know things the way a licensed professional does — they pattern-match across enormous amounts of text. Dr. Heart has occasionally given me suggestions I quietly set aside because they didn't feel right. Your gut and your doctor still matter.

Don't share information you wouldn't want stored online. Describe your situation in general terms — age range, lifestyle, goals. You don't need to hand over your full medical history to get genuinely useful answers.

The more specific your context, the better the output. A vague prompt gets vague answers. The 60 seconds you spend setting up your expert persona will pay you back every single time you use it.

Your Turn

Pick one area of your life where you'd love a knowledgeable friend on speed dial. Not a therapist, not a surgeon — something practical. A fitness coach. A meal planner. A home repair guide.

Open ChatGPT or Gemini. Paste the template. Fill in your situation.

Then ask it something you've been wondering about for months but never quite got around to looking up.

That's how it starts.

I also have an 8-month-old at home. No family nearby to help. You can do the math on how much free time I have.

That document sat on my desk for two days. I wanted to help. I just couldn't find a two-hour window that didn't already belong to someone who needed feeding or rocking to sleep.

Then I remembered: I have an intern for this.

I dropped the document into Claude — free, takes about 30 seconds to set up — typed one question, and had a clean summary of exactly what we needed to focus on and what we could safely set aside. What would have been a two-hour obligation on a sleep-deprived evening became a 10-minute task.

That's the whole pitch. Let me show you how.

First, Meet Your New Intern

Before we get into the how, I want to reframe how you think about AI — because I think the way most people picture it is what makes it feel so intimidating.

Forget the robots. Forget the sci-fi. Forget every breathless headline you've read about it.

Here's how I actually think about it: AI is a fast, eager, occasionally clueless digital intern.

This intern can read a 40-page document in 90 seconds. They'll pull out the key points, flag what needs your attention, and hand it back to you in plain English. But — and this matters — they'll sometimes get things slightly wrong, or miss a nuance that only you, with your actual experience and context, would catch.

So you check their work before you act on it. Just like you would with any intern.

You're the boss. They're doing the legwork. That's the whole deal.

The Villain: That Document You're Expected to Have Read

You know the one.

It shows up attached to a meeting invite. Or it's the industry report everyone on your team is "aligning on." Or it's the proposal a colleague spent three weeks on and needs your feedback on by Friday.

These documents aren't bad. There's just no version of your already-full day where reading all 40 pages actually happens.

This is exactly where your intern earns their keep.

The Hero: ChatGPT (Free. No Credit Card.)

I used Claude personally — it's what I had open at the time. But for your first try, I'm going to point you to ChatGPT, because it's the tool most people have already heard of and the signup is painless.

Quick note: ChatGPT, Claude, and Google Gemini can all do this exact task equally well. If you already have an account with any of them, just use that one. Don't overthink the choice — the tool matters far less than actually getting started.

Here's the whole workflow:

Step 1: Go to chatgpt.com and create a free account.

Step 2: Start a new chat and click the paperclip icon to attach your document. It accepts PDFs, Word docs, and most common file types.

Step 3: Paste this prompt and hit send:

Please go through this document and give me:
1. A plain-English summary in 3-5 sentences
2. The 3-5 most important takeaways I need to know
3. Anything I should be cautious about or that deserves a closer look

That's it. Within two minutes you'll have a clear picture of what you're actually dealing with — without having read a single page.

The Honest Part (Please Don't Skip This)

I'd be doing you a disservice if I didn't flag a few things upfront.

It can get things wrong. AI can misread data, miss important context, or occasionally state something confidently that is just... not accurate. For anything high-stakes, always do a quick skim of the original before you act on the summary.

Watch what you upload. If the document contains sensitive company information, confidential client data, or anything your HR or legal team would wince at — don't put it into a free tool. Check your company's policy first. Start with lower-stakes documents while you're still getting comfortable.

This is a head start, not a shortcut. The summary gets you oriented. It tells you where to focus your attention so that when you do engage with the real thing — or walk into that meeting — you're following along instead of scrambling to catch up.

Think of it as reading the map before the hike. You still do the hike.

What This Is Really About

I'm not here to tell you that AI is going to change everything and you need to get on board or get left behind. That kind of talk exhausted me before I even started.

What I can tell you is this: that 40-page compliance document got handled. My mother-in-law's business moved forward. And I still had time that evening to be present with my family instead of hunched over a document I barely had the energy to read.

AI handled the busywork. I brought the judgment.

That trade is available to you, for free, starting today.

Try it this week with something low-stakes — a long email thread, a meeting agenda, a report you've been putting off. You don't have to trust it fully yet. Just meet it.

I'll be here when you're ready for the next one.

— Kenny

You might expect that over the course of 10 years I might have published around 100+ articles, for example if you assume one article a month, that comes close to 120. However in reality, the count is something between 10 to 15 articles. I enjoyed writing each one of those, but there are a few factors which explain why I wrote as little as I did, and how I am planning to fix those.

Perfectionism

I was too much of a perfectionist during my early days. It was my dad who instilled that mindset in me. I was afraid of judgement and also falling short of others' expectations. It was a major factor holding me back from publishing my writing.

As I grew older I came to the realization that there are very few things that are perfect in the world, and that it is more important to give your best effort and to keep improving bit by bit.

Nowadays I give priority to progressive refinement, aka iterative improvement. Just do the simplest or most basic first cut, then slowly refine, or improve, and just hit publish when it's close to 80% of perfection. Waiting for perfection means the article will never be published.

Not having a topic to write about

Another problem was that whenever I would sit down I would draw a blank. "What should I write about?"

Over time from reading and watching videos on how content creators manage to produce so much content, I realized that I was doing it all wrong. I was trying to combine two things at once, and what I really needed to do was to separate the "idea generation" phase from the "idea implementation", aka writing phase.

From that, what I do now is that I keep an Idea Inbox using a notes app. For me I am using UpNote as my notes app, but you can as easily start doing this with Google Keep, or Apple Notes, or Notion or any other simple note taking application. The important part is to develop a system that works for you which helps you capture your ideas quickly and is accessible from anywhere. Its as simple as just jotting down a title, as well as a few key points that you might want to talk about.

Now, you just need to observe things happening in your day to day, e.g. you are watching a YouTube video and someone mentions something that resonates with you. Or let's say that your junior at office asks you for help regarding something and there you have another topic to add to your Idea Inbox. Once you have committed to writing regularly, you will naturally spot these things in your day to day.

Essentially by now, you have the ideas for several topics. Now how do you make those into well written articles?

Waiting for flow

Waiting for flow is the common thing that tripped me up. I think there is nothing more unproductive and demotivating than this. You have your ideas, and you have sat down to write. And you are still stuck.

For me what works is to note down a list of points for each topic when I create the idea in my Idea Inbox. Now whenever you see this list of points, it gives you something to build on and refine into paragraphs or sub topics.

What I also do is that whenever I get some time, I come back to the drafts in my inbox and keep adding more points to it. And most often what happens is that just doing this triggers a flow and I find myself adding detail to the article then and there. At random times in the day! This works because everything is synced to mobile, and web, and you can continue from where you left off.

Just keep writing daily or refining your drafts. But how do you know which drafts you need to prioritize, if you are randomly working on it? For this, I have seen many YouTubers and content creators use a Content Calendar.

Content calendar

Content calendar sounds like a super complicated idea! Who likes calendars anyways?

But the whole point is that you don't write in order to publish immediately. You want to be able to separate out the writing part from the publishing part. Write daily without pressure. Once it is done, just schedule the publishing part. And continue on with the rest of your drafts.

Having this in place means that you are not forced to publish daily, nor will you publish an article immediately after writing it, unless you are already on the date you had planned to publish on. This gives you less stress on the day to day, and you don't just keep publishing articles continuously, and get burnt out and then stop doing that for the rest of the year.

It's not even required that you need to schedule it on a visual calendar. It could be that you decide any one specific article for the week, to give more importance to. On Hashnode, I just make use of the pinned drafts feature to keep one or two most important next articles that need focus. In this way you have a pipeline of drafts ready on the way to become the next finished article.

Consistency going forward

I believe that these ideas will help me to keep publishing consistently. I have already seen the difference that it has made just thinking in this way. Only time will tell whether I am able to fully make this work and continue with my love of writing and helping others through it. Until next time, have a good one!

I started thinking about where all my articles are. The most obvious one was my latest blog which is a part of my website. Then I thought about it and remembered my old Google blogger which I used during my college days.

After that I checked my LinkedIn and saw a few small snippets that I had posted a few years ago. Then from there I saw a few links to my old medium blog. That's when I remembered about dev.to.

All in all it's been 10 years since the time I posted my first post. I'm sure very few people have read these, but now I want to combine everything here and stick on to hashnode as my platform.

I guess the more important thing is to write more consistently and with more quality, so I'm thinking of prioritizing that.

Expect blog posts on my thoughts about technology, Full Stack Engineering, system design and the works!