<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Parth Amin on Medium]]></title>
        <description><![CDATA[Stories by Parth Amin on Medium]]></description>
        <link>https://medium.com/@parthsamin?source=rss-b8cd286e840f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*1QxCNQSM2k95_9dK862w_g.png</url>
            <title>Stories by Parth Amin on Medium</title>
            <link>https://medium.com/@parthsamin?source=rss-b8cd286e840f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 08:32:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@parthsamin/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The people who never see your giveaway are the ones you most want]]></title>
            <link>https://parthsamin.medium.com/the-people-who-never-see-your-giveaway-are-the-ones-you-most-want-b29c2c941a80?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b29c2c941a80</guid>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[twitch]]></category>
            <category><![CDATA[content-creation]]></category>
            <category><![CDATA[live-streaming]]></category>
            <category><![CDATA[gaming]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Sat, 17 Oct 2026 16:14:19 GMT</pubDate>
            <atom:updated>2026-10-17T16:14:19.579Z</atom:updated>
            <content:encoded><![CDATA[<p><em>I build </em><a href="https://kloot.gg/"><em>Kloot</em></a><em>, a giveaway and overlay tool that works the way this post argues for, so weigh this accordingly. Drafted with AI assistance and edited by me. This first appeared on the Kloot </em><a href="https://kloot.gg/blog"><em>blog</em></a><em>.</em></p><p>Typing !enter is a good interface. It is fast, it needs nothing installed, and for years it matched where the audience actually was: a desktop browser with the chat column open beside the video. The argument here is not that chat commands were a mistake. It is that the assumption underneath them — that watching implies chatting — quietly stopped being true, and giveaway tools did not notice.</p><h3>Who a chat command cannot reach</h3><p>Start with the phone. A viewer watching full-screen in a mobile app has no chat on screen; opening it means leaving the video or shrinking it to a third of the display. They will do that to say something they care about. They will usually not do it because they half-heard you mention a giveaway.</p><p>Then the passive surfaces: a TV app, a console browser, a stream on a second monitor while someone works. Then anyone not signed in, which includes every person who arrived in the last few minutes from a clip or a recommendation. Then, if you stream to more than one platform, everyone on whichever platform your bot is not connected to. And if you are streaming to a room — an event, a meetup, a booth — literally everybody, because there is no chat at all.</p><p>None of these people are inattentive. They are watching. They are simply not typing, and a chat command cannot tell the difference between “not watching” and “not typing”. That conflation is the whole problem in one sentence: the tool treats a silent viewer as an absent one, and silent viewers are the majority on most streams most of the time. A giveaway that only reaches the people already participating is measuring engagement it already had rather than creating any.</p><h3>The excluded group is not random</h3><p>This is the part that matters commercially. The viewers least likely to have chat open are disproportionately the new ones: they have not logged in, they do not have the layout arranged, they are not yet invested enough to participate in text. A giveaway is usually run to convert exactly those people into regulars — so the mechanism excludes the audience the exercise is for.</p><p>Loyalty-point entry makes this structural rather than incidental. If entry costs points earned by watching, a first-time viewer cannot enter at any price. The giveaway is invisible to them by design, and no amount of announcing it changes that.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ON8HSkt19BSEzAUOAhzMYg.jpeg" /></figure><h3>Why it stayed this way</h3><p>It is worth understanding why almost every giveaway tool works this way, because it is not laziness. Giveaways were built inside chat bots, and chat bots were the first thing streamers automated — the bot was already connected, already authenticated, already reading every message, so adding a command that recorded a name was nearly free. Building an entry surface that lives on the overlay instead means rendering something, hosting a page viewers land on, and handling entries yourself. That is a product, not a feature.</p><p>So the ecosystem optimised in the direction that was cheap, and by the time mobile viewing changed the assumption, every tool in the category had years of chat-shaped architecture behind it. That is a normal way for software to end up somewhere odd, and it is not a criticism of any of the products involved.</p><h3>What changes if entry is on screen</h3><p>Put the entry surface on the stream itself — a code to scan, a short code to type — and the question of what a viewer has open disappears. The phone viewer scans the thing they are already looking at. The person on a TV types six characters into their phone. The room at an event all scan the projector. Nobody needs an account on your platform, because the tool never touches your platform.</p><p>The trade is real and worth stating: you lose the things a chat bot gives you for free. There is no command, no moderation, no timers, no auto-triggered alerts. If those matter more than reach, a chat bot is the right tool and this argument does not apply to you.</p><h3>The number you can actually check</h3><p>Rather than trusting any argument about this, including this one, there is a measurement you can take on your own stream. Run one giveaway the way you normally do and note entries against concurrent viewers. Run the next one with the code on screen and note the same ratio. It is not a controlled experiment — the prize, the night and the mood all vary — but a large gap that repeats is the only evidence about your audience that is actually about your audience.</p><p>We deliberately have not put a percentage in this post. Every number we could quote would come from channels that are not yours, and the whole argument is that the answer depends on how your particular viewers watch.</p><h3>What it does to the draw itself</h3><p>One knock-on effect is worth naming. When entry does not require chat, the winner probably is not watching chat either — so announcing the name there is unreliable. That is why an on-screen entry mechanism tends to come with a claim step: the winner is told on the device they entered from, and the draw moves on automatically if they do not respond.</p><p>It is a small thing that removes the most common failure on stream, which is the silence after a name is read out to someone who stepped away. Chat-based giveaways get away with it because the entrants were, by definition, in chat when they entered — but that guarantee expires the moment the draw takes longer than someone’s attention, which is most draws.</p><p>If you want to run that two-giveaway test, Kloot’s free plan covers two giveaways a month with no card: a QR code on your overlay, the draw on stream, and the winner claiming from their phone. <a href="https://kloot.gg/features/giveaways?utm_source=medium&amp;utm_medium=community&amp;utm_campaign=scout&amp;utm_content=medium-enter-is-dead">See how entry works</a></p><p>The 30-second version:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F_1lKNU0_8rk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fshorts%2F_1lKNU0_8rk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F_1lKNU0_8rk%2Fhq2.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/dc9d90aaac794dc313508be754b18a2e/href">https://medium.com/media/dc9d90aaac794dc313508be754b18a2e/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b29c2c941a80" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Press a button. It’s on your stream.]]></title>
            <link>https://parthsamin.medium.com/press-a-button-its-on-your-stream-4e96ec1dc907?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e96ec1dc907</guid>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[live-streaming]]></category>
            <category><![CDATA[content-marketing]]></category>
            <category><![CDATA[gaming]]></category>
            <category><![CDATA[streamlit]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Thu, 01 Oct 2026 01:07:52 GMT</pubDate>
            <atom:updated>2026-10-01T01:07:52.752Z</atom:updated>
            <content:encoded><![CDATA[<h4>Death counters with 100+ looks, giveaways anyone can enter, polls, BRB screens and emoji rain — all from one deck of buttons. A streamer’s tour of Kloot.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bfGMkG88zwF0jmQ-_5x2hA.png" /><figcaption>One deck of buttons, one browser source in OBS. Everything in this post runs through it.</figcaption></figure><p>You know the moment. The boss finally drops, chat explodes, and you’re supposed to bump the death counter, fire something loud, and maybe start that giveaway you promised an hour ago. Your hands are on the controller. The widget you need lives in a dashboard, in a tab, behind an alt-tab you’re not going to do mid-fight.</p><p>So the moment passes. Not because you didn’t care — because your tools made you choose between playing and producing.</p><p>I built <strong>Kloot</strong> so you don’t have to choose. It’s one browser source in OBS and a deck of buttons that lives in your browser <em>and</em> on your phone. Press a button. It’s on your stream. That’s the whole idea, and the rest of this post is what those buttons actually do.</p><p><strong>The 30-second version:</strong></p><ul><li><strong>One overlay URL.</strong> Paste it into OBS or Streamlabs once. Every tool below renders through it. No new browser source per widget, ever.</li><li><strong>A deck you can hold.</strong> It’s a web page, so it’s also your phone. Add it to your home screen and fire everything without leaving the game.</li><li><strong>Works on every platform.</strong> Twitch, YouTube Live, Kick, TikTok Live. Viewers join giveaways and polls by QR code or a short join code, so there’s no chat bot to set up and nothing that breaks when you multistream.</li><li><strong>Free to start.</strong> No card, no trial clock.</li></ul><p>Here’s the short explainer if you’d rather watch:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F3EO_SCpkOsM%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D3EO_SCpkOsM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F3EO_SCpkOsM%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/1dbef8a366b4c23d57a7b8be7afe1210/href">https://medium.com/media/1dbef8a366b4c23d57a7b8be7afe1210/href</a></iframe><h3>1. Counters that know which boss you’re on</h3><p>Counters are the Kloot tool streamers search for most, so they’re the one I’ll show you first.</p><p>A normal counter widget gives you one number for the whole run: <em>Deaths: 223.</em> A Kloot counter splits by segment, so every boss, level or map keeps its own count and its own personal best. The overlay reads <em>“this boss 16 · PB 9”</em> instead of a number nobody can put in context.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ioaA6ixnffYcYY1lXsPTkg.jpeg" /><figcaption>Three counters on one browser source: deaths per boss with a PB line, bosses beaten as a tally, and tries — each in its own look.</figcaption></figure><p><strong>What makes it worth switching:</strong></p><ul><li><strong>Bump it from your phone.</strong> +1, −1 and reset sit on your deck. Keep the game fullscreen and tap.</li><li><strong>Let your mods count for you.</strong> On Creator, send a <strong>co-pilot link</strong> to a mod or VIP. They sign in once with Google or Twitch and get a one-counter phone deck, and every bump is named. Five co-pilots on Creator, twenty on Pro.</li><li><strong>100+ designed looks (Creator).</strong> Search by game or by what you count, click one, and the preview plays it live before you save. Grave ledgers, ember piles, ink tallies, gilded gates.</li><li><strong>Milestones that stay off the game (Creator).</strong> Every 10 deaths, or numbers you pick: the counter grows, embers burn inside the box, or a ribbon slides along the bottom edge. Once per run. Never over your gameplay.</li><li><strong>A recap card when you end the stream (Creator).</strong> Press <em>Start session</em> when you go live and <em>End session</em> when you’re done. Download a card of tonight’s counts in one of five layouts, sized for X, Discord, Instagram or Stories.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YsrIk9ZvxA1rwBJKVBpCoA.jpeg" /><figcaption>The look picker: search “elden ring” or “hydration”, filter by what you’re counting, and preview any look before it goes live.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EEuEcI3AQgmVjqyeLGckfg.jpeg" /><figcaption>The stream recap. Tonight’s numbers next to your all-time totals, ready to post before you’ve closed OBS.</figcaption></figure><p>On <strong>Pro</strong>, counters get stakes: fold kills, deaths and assists into one <strong>scoreboard</strong> box, show a charity counter as money (<em>$1 per death</em> reads <em>$47</em> on stream), or set a countdown that puts your BRB screen up by itself when it hits zero. And from Creator up, Stream Deck and Streamer.bot users can trigger counters with a plain link — nothing to install.</p><p>The full walkthrough is here:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FVXoDNRW2fwU%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DVXoDNRW2fwU&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FVXoDNRW2fwU%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/1b7a21d5c661489ddd06f5dd16591353/href">https://medium.com/media/1b7a21d5c661489ddd06f5dd16591353/href</a></iframe><p><strong>Try it:</strong> <a href="https://kloot.gg/counters?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-counters">kloot.gg/counters</a> — one live counter is free on every plan.</p><h3>2. Giveaways your whole audience can enter</h3><p>Most giveaway tools still run on !enter. That works until you’re on YouTube instead of Twitch, the bot loses its token, or you’re live on two platforms and only one has a bot.</p><p>Kloot doesn’t read chat at all. A QR code and a six-character join code sit on the giveaway card. Viewers scan it or type the code at <strong>kloot.gg/join</strong>, enter a name for the wheel, and they’re in. No account, no extension, no command to learn — and it’s identical on every platform.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IE6LQPWlk47a93iUahKv2A.jpeg" /><figcaption>The wheel spins live on your overlay. The prize card, join code and QR sit right beside it.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ON8HSkt19BSEzAUOAhzMYg.jpeg" /><figcaption>What your viewer sees: one field, one button, done. Then they watch the stream to see if they won.</figcaption></figure><p><strong>The bits that save you from the awkward parts:</strong></p><ul><li><strong>The winner claims from their own phone.</strong> A countdown appears on the winner’s screen — two minutes by default. No tap in time and Kloot rerolls on its own, server-side, even if you’ve closed the panel. No dead air while you shout a name into the void.</li><li><strong>Up to five prizes in one giveaway.</strong> One spin each, nobody wins twice, and the overlay shows “prize 2 of 3” so people stay to the end.</li><li><strong>Prize codes delivered for you.</strong> On Creator and Pro, paste your Steam keys or gift-card codes before you go live and each winner gets theirs by email the moment the wheel lands.</li><li><strong>Rules written for you.</strong> Eligibility, the age limit, how the winner is picked and the platform’s required disclaimer are generated from your setup and shown on the entry page. On every plan.</li><li><strong>A fairness receipt.</strong> Each draw commits to its random seed before entries close and publishes it afterwards. When someone types “rigged”, you have a link instead of an argument.</li></ul><p>Free includes <strong>two giveaways a month</strong>. Creator gives you ten, Pro removes the cap.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FqGZLk7kEEjw%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DqGZLk7kEEjw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FqGZLk7kEEjw%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/534e321c2b868739aea0a4fb967735d1/href">https://medium.com/media/534e321c2b868739aea0a4fb967735d1/href</a></iframe><p><strong>Run your first one:</strong> <a href="https://kloot.gg/how-to/run-a-giveaway-on-stream?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-giveaways">How to run a giveaway on stream</a></p><h3>3. Polls — no Affiliate badge required</h3><p>Twitch’s built-in polls need Affiliate. YouTube’s sit in the chat panel. A Kloot poll sits <em>on the stream</em>, where everyone’s already looking, with the same QR and join code as giveaways.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WCZlhkujmwdRFEXE00S-Jg.jpeg" /><figcaption>Polls live in the same panel as everything else. “Which map do we run next?” is up, its join code is on screen, and the deck is one tap away.</figcaption></figure><p>Fire it from a deck key, watch the bars move in real time, close it when the room has decided. Free gets <strong>two polls a month</strong> with up to four options. Creator unlocks the other four types:</p><ul><li><strong>Multi-select</strong> — “pick two games for tonight.”</li><li><strong>This or that</strong> — two images, one split bar. Thumbnails, builds, cover art.</li><li><strong>Quiz</strong> — one right answer, points for speed, a live leaderboard.</li><li><strong>Word cloud</strong> — “one word for tonight’s stream.” It fills as they type.</li></ul><p>And the trick I like most: turn on <strong>vote-to-enter</strong> and every vote is also a giveaway entry. The poll closes, the wheel spins on the voters, and you’ve asked your audience to do one thing instead of two.</p><p><strong>Set one up:</strong> <a href="https://kloot.gg/how-to/add-a-poll-to-your-stream?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-polls">How to add a live poll to your stream</a></p><h3>4. Screens with a clock that actually runs</h3><p>Starting Soon, Be Right Back and ending screens, flipped on from your deck — no OBS scene juggling. There’s a library of <strong>64 templates across 15 moods</strong>, from pixel-type retro to neon synthwave to cozy study, and every tile is a live preview of the real thing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*40O7WxsZhzJPMWv6QaeXSQ.jpeg" /><figcaption>The Screens library. Every template comes as a Starting Soon, a BRB, an ending screen and a short moment card.</figcaption></figure><p>The countdowns run for real. On Creator they keep counting past zero in red, so viewers can see you’re running late (honest, and weirdly charming), the backgrounds can move, and trivia, your schedule and your socials rotate on the screen while you’re away.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0qFAa6FsdDQhlICLo6FH2A.jpeg" /><figcaption>On Pro, a counter can hand off to a screen: count down from 5, and at zero your BRB goes up by itself with its timer already running.</figcaption></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F3vfXnJ19b44%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D3vfXnJ19b44&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F3vfXnJ19b44%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/a5069478c2242ad544500300be381821/href">https://medium.com/media/a5069478c2242ad544500300be381821/href</a></iframe><p><strong>Browse them:</strong> <a href="https://kloot.gg/screens/library?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-screens">kloot.gg/screens/library</a> — 19 templates go live on every plan, including Free.</p><h3>5. Emoji rain, confetti and the inside joke</h3><p>The fastest way to see what Kloot feels like. Six one-tap presets — confetti, fireworks, fire storm, heart rain, star burst, shockwave — plus emoji rain and big text pops with any emoji you pick.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F3jVaqcr0MnGy3FE0jFZzA.jpeg" /><figcaption>A custom burst from the Emojis tool. Every stream has its duck. Now yours has a button.</figcaption></figure><p>Then build your own: your channel’s inside joke, as a burst of your emojis and words, pinned to a deck key. Set a cooldown if your mods get trigger-happy. <strong>Every emoji effect is free, forever.</strong></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F0GW5jZH1G3A%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D0GW5jZH1G3A&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F0GW5jZH1G3A%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/87f98e0035e737b81e44b2cdc92f9923/href">https://medium.com/media/87f98e0035e737b81e44b2cdc92f9923/href</a></iframe><h3>6. The rest of the deck</h3><p>Everything above shares a deck with the tools you’d expect from a full production setup:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PkFXN_gqIhe8qKzJwgXiqA.jpeg" /><figcaption>The deck on a phone: a grid of keys per page, more pages when you need them, and every tool along the bottom.</figcaption></figure><ul><li><strong>Lower thirds</strong> — name bars for guests and segments, in five layouts.</li><li><strong>Ticker</strong> — a scrolling bar for your schedule, socials and shout-outs.</li><li><strong>Alerts</strong> — follow, sub and raid cards, styled to you, fired when <em>you</em> choose.</li><li><strong>Sponsors</strong> — an always-on banner plus a one-tap promo-code callout that lands exactly when you read the ad (Pro).</li><li><strong>Borders and Cover</strong> — frame your webcam, or drop an opaque panel over a DM or a password in one tap.</li><li><strong>Media</strong> — your own WEBM stingers, sounds and frames on deck keys.</li></ul><p>One thing you won’t find: auto-firing alerts. Kloot is manual on purpose. The confetti lands when the run finishes, not three seconds into your punchline. You’re the producer; the deck just makes producing one tap.</p><h3>What it costs</h3><ul><li><strong>Free</strong> — $0, no card. Every emoji effect, two screens with a countdown, one live counter, two giveaways and two polls a month, and one of each other tool. A small Kloot badge on the overlay.</li><li><strong>Event Pass</strong> — <strong>$9, one time, three days of everything in Pro.</strong> This is how you try the paid features on a real stream. No subscription.</li><li><strong>Creator</strong> — $99 a year (that’s $8.25/mo, billed annually) or $15 monthly. 100+ counter looks, co-pilots and recap cards, all five poll types, ten giveaways a month with prize delivery, the full Screens library, and no badge.</li><li><strong>Pro</strong> — $149 a year ($12.42/mo, billed annually) or $25 monthly. Unlimited giveaways and polls, sponsor overlays, counter stakes and scoreboards.</li></ul><p>Full details on <a href="https://kloot.gg/pricing?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-pricing">the pricing page</a>.</p><h3>Try it in the next 30 seconds</h3><p>You don’t need an account to see this. Open <a href="https://kloot.gg/demo?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-demo-cta"><strong>the live demo</strong></a>, press a button on the deck on the left, and watch it land on the stream preview on the right. It’s the same overlay components your viewers would see.</p><p>If it clicks, <a href="https://kloot.gg/?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-signup-cta">build your deck for free</a>. Setup is one copy-paste into OBS:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FMSaB9XWqi30%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DMSaB9XWqi30&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FMSaB9XWqi30%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/743d2cf2708bf95858a32cd5db30a2f7/href">https://medium.com/media/743d2cf2708bf95858a32cd5db30a2f7/href</a></iframe><p>I’m one person building this, and the roadmap is shaped almost entirely by streamers telling me what they need. If something’s missing, tell me — I’m <strong>@klootgg</strong> everywhere, and I read everything.</p><p>Go make some moments.</p><p><em>Kloot is built by Nisq Technologies Inc. in Toronto · </em><a href="https://kloot.gg/?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-footer"><em>kloot.gg</em></a><em> · guides at </em><a href="https://kloot.gg/how-to?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=streamer-tour&amp;utm_content=md-footer-howto"><em>kloot.gg/how-to</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e96ec1dc907" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Workflows Beat Agents: A Field Guide to Laying Out Agentic AI]]></title>
            <link>https://parthsamin.medium.com/workflows-beat-agents-a-field-guide-to-laying-out-agentic-ai-f7a19b0159af?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/f7a19b0159af</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Fri, 29 May 2026 19:19:46 GMT</pubDate>
            <atom:updated>2026-06-02T02:04:09.611Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why most “agentic AI” projects are misclassified workflows — and the three reframes that separate demos from systems.</h3><figure><img alt="Climbing toward autonomy" src="https://cdn-images-1.medium.com/max/1024/1*KOqknHsoGfF3Cp79R6gMwg.jpeg" /><figcaption>Climbing toward autonomy: <em>Autonomy isn’t a switch you flip. It’s a ladder you earn, one rung of evidence at a time.</em></figcaption></figure><p>Every enterprise I talk to wants agentic AI. Almost none of them are asking the right question about it.</p><p>The question they ask is <em>“where can we put an agent?”</em> The question that actually predicts success is <em>“what would have to be true for us to let it act on its own?”</em> — and most teams don’t answer that until something breaks in production.</p><p>I’ve spent the last couple of years laying out agentic solutions in environments where being wrong is expensive — regulated, audited, high-stakes. Along the way I’ve watched a lot of impressive demos die on the way to production, and a few quiet, un-flashy systems make it all the way to real value. The difference is almost never the model. It’s the operating model.</p><p>Here are the three reframes that, in my experience, separate the two — illustrated with two deliberately opposite cases I’ll carry through the whole piece: a <strong>health plan automating prior-authorization decisions</strong> (low volume, high stakes, heavily regulated) and a <strong>retailer building a post-purchase customer service agent</strong> (high volume, lower per-case stakes, real money moving). If you can hold both in your head, you’ve covered most of the judgment calls you’ll ever face.</p><h3>Reframe #1: Workflows beat agents whenever the path is knowable</h3><p>The word “agentic” has quietly become a synonym for “advanced,” and that’s a problem. An agent — an LLM running in a loop, choosing its own steps and tools toward a goal — is the most powerful tool in the box and the most expensive, least predictable, and hardest to govern. You should reach for it last, not first.</p><p>There’s a spectrum here, and being able to place a problem on it is half the expertise:</p><ul><li>Deterministic rules and RPA for stable, zero-ambiguity steps.</li><li>Classical ML for narrow predictions.</li><li>A single LLM call for drafting or extraction.</li><li>An <strong>LLM workflow</strong> — a chain of model and tool calls along paths <em>you</em> defined in code.</li><li>An <strong>agent</strong> — where the <em>model</em> decides the path.</li><li>Multi-agent, where several agents collaborate.</li></ul><figure><img alt="The agent spectrum" src="https://cdn-images-1.medium.com/max/1024/1*Is6wn2rjjx51g5rOraYFrA.jpeg" /><figcaption>The agent spectrum: <em>Most “agentic” asks are actually workflow problems wearing a costume. Reclassifying down is often the highest-value move in the whole engagement.</em></figcaption></figure><p>The expert instinct is counterintuitive: <strong>the highest-value thing you can do is often talk a client out of building an agent.</strong> When a team says “we want an agentic prior-auth system,” the truth is usually that intake and triage are a deterministic workflow, the clinical reasoning is genuinely agent-worthy, and the <em>decision to sign</em> should never be automated at all. The same request decomposes into three different levels — and only the middle one is actually an agent.</p><p>Saying that out loud builds more trust than any demo. It signals you’re optimizing for their outcome, not for shipping the buzzword. It’s also how you earn the bigger phase of work.</p><h3>Reframe #2: Autonomy isn’t a switch. It’s a ladder.</h3><p>This is the one I’d tattoo on every product team if I could.</p><p>Beginners ask, <em>“Is the agent autonomous?”</em> — as if it’s a system-level yes or no. It isn’t. Autonomy is a property of <strong>each action, for each type of case</strong>. One deployed system sits on several rungs at once: full autopilot for “where’s my order,” tightly capped action for small refunds, and pure human-approval for high-value disputes — simultaneously.</p><p>So instead of a binary, picture a ladder — and you only climb to the next rung when the evidence says you’ve earned it:</p><ol><li><strong>Shadow</strong> — the agent runs on real inputs, but its output is hidden. You’re comparing it to humans, building trust data at zero risk.</li><li><strong>Suggest (copilot)</strong> — it drafts; a human reviews and approves everything. Every human edit is a correction signal that makes the system smarter.</li><li><strong>Assist</strong> — it auto-handles the clear, easy cases and escalates the rest. This rung lives or dies on whether the agent reliably <em>knows what it doesn’t know.</em></li><li><strong>Act-with-limits</strong> — it takes real action inside hard bounds (dollar caps, allow-lists), and a human spot-checks a sample after the fact instead of approving each one.</li><li><strong>Autopilot</strong> — full autonomy on a defined, well-evaluated, reversible scope — monitored, with a kill switch. “Autopilot” never means “unwatched.”</li></ol><figure><img alt="The five rungs of autonomy" src="https://cdn-images-1.medium.com/max/848/1*b0mBIXFPvUm4TDeJE5RJHA.jpeg" /><figcaption>The five rungs of autonomy: <em>You don’t deploy autonomy. You earn it — rung by rung, gated by evidence.</em></figcaption></figure><p>The healthcare and retail cases climb the ladder very differently, and that’s the point. In retail, “where’s my order” can reach full autopilot, and small refunds can reach act-with-limits behind a dollar cap. In prior authorization, drafting a recommendation can climb to assist or act-with-limits — but <strong>issuing a denial is never autonomous, at any rung, ever.</strong> That ceiling isn’t an engineering decision; it’s a regulatory and human one. The ladder makes that limit explicit and visible instead of buried in a config file.</p><p>There’s a safety mechanism that makes the whole thing defensible: agents don’t only climb, they get <strong>demoted</strong>. If live accuracy drops, if a critical incident fires, if the input distribution drifts into territory you never evaluated — the system automatically drops to a lower, human-supervised rung and alerts someone. Climbing is only responsible if falling is automatic.</p><h3>Reframe #3: The thing nobody budgets for is the thing that decides everything</h3><p>Here’s where I see the most expensive mistakes. Teams pour their effort into the build and treat measurement and economics as an afterthought. It’s backwards. <strong>Evaluation is the engine, not the inspection at the end of the line.</strong></p><p>You cannot gate, ship, trust, or price what you can’t measure. And the reason the Autonomy Ladder works at all is that <em>the evaluation system produces the evidence that the ladder consumes as its gates.</em> Promote a rung only when accuracy clears a pre-agreed bar on a representative dataset, over a big enough sample, signed off by a named owner. No measurement, no promotion. The two are one system.</p><p>A few things I wish more people understood about the economics:</p><p><strong>Quote cost-per-<em>successful</em>-task, never cost-per-task.</strong> A cheap agent that succeeds 60% of the time and dumps the other 40% (plus its own wasted spend) back on humans is <em>expensive</em>. Divide cost by success rate and the comparison flips — the pricier, more accurate model is frequently the cheaper one.</p><p><strong>Know which cost regime you’re in.</strong> This is where the two cases diverge beautifully. In retail, the per-task token cost is a real, visible slice of the few dollars you save per contact — so cost-engineering (routing easy steps to cheap models, caching, capping loops) genuinely matters. In prior auth, the token cost is a rounding error next to expensive clinician time — so success rate and reducing human-in-the-loop effort dominate the math, and you should happily spend on the better model and the better evaluation. Knowing which regime you’re in tells you <em>where to spend your engineering effort.</em> Most teams optimize the wrong variable.</p><p><strong>The human-in-the-loop ratio is the dollars, and it falls as you climb.</strong> At the copilot rung, humans touch 100% of cases. At assist, only the escalations. At act-with-limits, just a spot-check sample. Plot that curve and you’ve drawn your value-realization story — and your funding argument — in a single chart.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OJUCi_XdjcpIpKZigamFUg.jpeg" /><figcaption>The flywheel that funds the next use case: <em>Evaluation produces evidence → evidence is the gate to climb → climbing drops human effort → better economics fund the next agent. Build the loop once; every use case after gets cheaper.</em></figcaption></figure><p>This is the closed loop, and it’s the real prize. Evaluation generates the evidence. The evidence is the ladder’s gate. Climbing the ladder drops the human-in-the-loop cost. The improved economics fund the next use case — whose human corrections flow right back into the evaluation set. Build that loop once and every subsequent agent you ship is cheaper, safer, and faster. Skip it, and you’re stuck running impressive demos forever.</p><h3>So how do you actually start?</h3><p>If you take nothing else from this: before you write a line of code, answer four questions.</p><ol><li>What <em>job</em> is being done badly, and is the bottleneck genuinely ambiguity and judgment over language? (If not, you don’t need an agent.)</li><li>Can the path be predetermined? (If yes, build a workflow, not an agent.)</li><li>How will you measure “good,” and what’s the baseline <em>today</em>? (Measure before you launch — you can’t prove a lift you never recorded.)</li><li>For every action it can take: what’s the blast radius if it’s wrong, and can a mistake be undone? (This sets the rung.)</li></ol><p>Answer those honestly and you’ve already avoided the failure modes that kill most projects. The rest — the architecture patterns, the risk taxonomy, the portfolio strategy, the full roadmap — is craft you can learn and systematize.</p><p>Which is exactly why I built it into something you can actually use rather than just read.</p><blockquote><a href="https://parthsamin.github.io/agentic-ai-playbook">https://parthsamin.github.io/agentic-ai-playbook</a></blockquote><blockquote><em>I’ve turned this whole methodology into an interactive guide — the full eight-phase operating model, the Autonomy Matrix, the evaluation and economics templates, and a set of tools you can run against your own use cases. It goes far deeper than a blog post can.</em></blockquote><blockquote><a href="https://parthsamin.github.io/agentic-ai-playbook"><strong><em>Explore the interactive guide →</em></strong></a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7-ebTwRwkc3DesIELP9qMA.png" /><figcaption>Why Agentic AI Projects Fail in Production?</figcaption></figure><p>I’m Parth — a PM and serial builder who’s spent the last few years getting agentic AI from “cool demo” to “actually running.” I share what I learn along the way. If any of this helped, come say <a href="https://www.linkedin.com/in/aminparth">hi on LinkedIn</a> — I’d genuinely love to hear what you’re building.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f7a19b0159af" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Definitive Guide to Managing Dev & Prod Environments in AWS Serverless]]></title>
            <link>https://parthsamin.medium.com/the-definitive-guide-to-managing-dev-prod-environments-in-aws-serverless-4af342f50bdb?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4af342f50bdb</guid>
            <category><![CDATA[aws-lambda]]></category>
            <category><![CDATA[serverless-architecture]]></category>
            <category><![CDATA[api-gateway]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[environment]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Tue, 12 Nov 2024 11:02:02 GMT</pubDate>
            <atom:updated>2024-11-12T11:02:02.398Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction</h3><p>When building serverless applications on AWS, managing different environments (development, staging, and production) is crucial for maintaining a robust development lifecycle. This comprehensive guide will walk you through setting up and managing development and production environments using <strong>AWS Lambda</strong> and <strong>API Gateway</strong>.</p><h3>Prerequisites</h3><ul><li>AWS Account with appropriate permissions</li><li>Basic understanding of AWS Lambda and API Gateway</li><li>AWS CLI installed and configured</li><li>Basic Node.js knowledge</li></ul><h3>Step 1: Initial Lambda Function Setup</h3><h3>1.1 Creating the Lambda Function</h3><ol><li>Navigate to Lambda Console</li><li>Click “Create function”</li><li>Choose “Author from scratch”</li><li>Configure basic settings:</li></ol><ul><li>Function name: myServiceFunction</li><li>Runtime: Node.js 20.x</li><li>Architecture: x86_64 or arm64</li><li>Select the IAM role (if created earlier) or keep the default execution role to ‘Create a new role with basic Lambda permissions’</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/819/1*n53CsYde0CjmFt64mO-0Gg.png" /><figcaption>1.1 Creating the Lambda function</figcaption></figure><h3>1.2 Implementing Environment-Aware Code</h3><pre>exports.handler = async (event) =&gt; {<br>    try {<br>        // Validate stage variables<br>        if (!event.stageVariables || !event.stageVariables.aws_env) {<br>            throw new Error(&#39;Stage variable aws_env is not configured&#39;);<br>        }<br>        <br>        const aws_env = event.stageVariables.aws_env;<br>        <br>        // Configure environment-specific resources<br>        let t_user_details;<br>        if (aws_env === &quot;dev&quot;) {<br>            t_user_details = process.env.TABLE_USER_DETAILS_DEV;<br>        } else if (aws_env === &quot;prod&quot;) {<br>            t_user_details = process.env.TABLE_USER_DETAILS_PROD;<br>        } else if (aws_env === &quot;staging&quot;) {<br>            // Similarly add details for staging environment too<br>        } else {<br>            throw new Error(`Invalid environment: ${aws_env}`);<br>        }<br>        <br>        // Add logging for debugging<br>        console.log(&#39;Environment:&#39;, aws_env);<br>        console.log(&#39;Table:&#39;, t_user_details);<br>        <br>        // Your function logic here<br>        const response = {<br>            statusCode: 200,<br>            body: JSON.stringify({<br>                message: `Running in ${aws_env} environment`,<br>                tableUsed: t_user_details<br>            }),<br>        };<br>        return response;<br>        <br>    } catch (error) {<br>        console.error(&#39;Error:&#39;, error);<br>        return {<br>            statusCode: 500,<br>            body: JSON.stringify({<br>                message: &#39;Internal server error&#39;,<br>                errorDetails: process.env.aws_env === &#39;dev&#39; ? error.message : undefined<br>            })<br>        };<br>    }<br>};</pre><h3>1.3 Setting Up Environment Variables</h3><ol><li>Navigate to Configuration → Environment variables</li><li>Add the following variables:</li></ol><ul><li>TABLE_USER_DETAILS_DEV = user-details-dev</li><li>TABLE_USER_DETAILS_PROD = user-details-prod</li></ul><p>Note: Environment variables are:</p><ul><li>Encrypted at rest using AWS KMS</li><li>Available during function execution</li><li>Specific to each version when published</li></ul><p>The above variables are just examples of how different table names could be called based on different environments.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/816/1*4HTPLI4EwMQbXwGOwn1c4w.png" /><figcaption>1.3 Setting Up Environment Variables</figcaption></figure><h3>Step 2: Version Management and Aliases</h3><h3>2.1 Understanding Lambda Versions</h3><p>Lambda versions are immutable snapshots of your function, including:</p><ul><li>Function code</li><li>Configuration (memory, timeout, etc.)</li><li>Environment variables</li><li>Lambda layers</li><li>Function description</li></ul><p>Key points about versioning:</p><ul><li>$LATEST is always the latest code</li><li>Published versions get numeric IDs (1, 2, 3, etc.)</li><li>Each version has a unique ARN</li><li>Versions are immutable — cannot be changed once published</li></ul><h3>2.2 Publishing Versions</h3><ol><li>Test your function thoroughly</li><li>Navigate to Version → Publish new version</li><li>Add a description of changes, as shown below, and click “Publish”</li></ol><pre>Version 1:<br>- Initial production release<br>- Basic CRUD operations<br>- Environment variable support</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/828/1*XZ3TZwudR7DgejOnZLxnJA.png" /><figcaption>2.2 Publishing Versions</figcaption></figure><h3>2.3 Creating and Managing Aliases</h3><p>Create aliases for each environment:</p><ol><li>Go to “Aliases” tab</li><li>Click “Create alias”</li><li>Set up the following:</li></ol><p>For Development:</p><pre>Name: dev<br>Description: Development environment<br>Version: $LATEST</pre><p>For Staging:</p><pre>Name: staging<br>Description: Staging environment<br>Version: (specific version number)</pre><p>For Production:</p><pre>Name: prod<br>Description: Production environment<br>Version: (specific version number)<br>Additional configuration: <br>- Enable weighted alias for blue-green deployment (optional)<br>- Configure routing additional version (optional)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZoLmLF8PwXlFpoJ_vRN6Cg.png" /><figcaption>2.3 Creating and Managing Aliases</figcaption></figure><h3>2.4 Best Practices for Version Management</h3><p><strong>Version Promotion Strategy:</strong></p><ul><li>Test thoroughly in dev ($LATEST)</li><li>Publish new version</li><li>Point staging alias to new version</li><li>Test in staging</li><li>Update prod alias after validation</li></ul><p><strong>Rollback Strategy:</strong></p><ul><li>Keep previous version ARN documented</li><li>Update alias to point to previous version if needed</li><li>Maintain deployment history</li></ul><p><strong>Monitoring Considerations:</strong></p><ul><li>Set up separate CloudWatch alarms for each alias</li><li>Monitor error rates and duration metrics</li><li>Configure X-Ray tracing for detailed analysis</li></ul><h3>Step 3: API Gateway Configuration</h3><h3>3.1 Creating the API Resource</h3><ul><li>Navigate to API Gateway</li><li>Create a new REST API</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/820/1*uaaHBJ6lK6lLUomW5Qx2Fg.png" /></figure><ul><li>Create a new resource and method</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/817/1*FQNAiJ08lI0yNLxGdSnZxA.png" /></figure><ul><li>Configure the Lambda integration</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/817/1*9QN58NuHsKoggqY5xJsdYw.png" /><figcaption>Choose the Lambda function for integration created above</figcaption></figure><h3>3.2 Configuring Lambda Integration</h3><ul><li>Edit the Integration Request</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j_kw0_N2XHJHPEoT99oAuA.png" /></figure><ul><li>In the Lambda Function field, modify the ARN to include stage variables and add <strong>:${stageVariables.aws_env}</strong>at the end</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/826/1*clPQgZZ_POdT7KssFE4VTQ.png" /></figure><h3>3.3 Setting Up IAM Permissions</h3><ul><li>As soon as you add the above, you will see the below permission command. <strong>Copy </strong>this full command and take it into an editor (Notepad, VS Code, etc.)</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/776/1*-rkH9y3dGjs_SOauODflgA.png" /></figure><ul><li>Modify this full command so as to change ${stageVariables.aws_env} to dev, staging, and prod based upon your environments and aliases created above.</li><li>Run each of this AWS CLI commands in CloudShell to grant necessary permissions:</li></ul><pre># For development environment (Example Only)<br>aws lambda add-permission \<br>    --function-name &quot;arn:aws:lambda:us-east-1:92178:function:myLamdaFunction:dev&quot; \<br>    --source-arn &quot;arn:aws:execute-api:us-east-1:92178:api-id/*/HTTP-METHOD/resource-path&quot; \<br>    --principal apigateway.amazonaws.com \<br>    --statement-id unique-statement-id \<br>    --action lambda:InvokeFunction</pre><h3>3.4 Deploying the API</h3><ol><li>Select your API resource</li><li>Click “Actions” → “Deploy API”</li><li>Select the target stage or set up a new one, preferably “development”.</li><li>Repeat for each stage</li></ol><h3>3.5 Creating API Stages</h3><ol><li>Go to Stages in API Gateway</li><li>Create three stages: dev, staging, and prod</li><li>For each stage, <strong>add a stage variable</strong>:</li></ol><ul><li>Name: aws_env</li><li>Value: dev/staging/prod (respective to the stage)</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jDBgTPchuo61gtCeANRmJQ.png" /></figure><h3>Step 4: Testing the Setup</h3><h3>4.1 Testing Different Environments</h3><p>Test your API endpoints using the different stage URLs:</p><pre>If created properly, it should look something like below:</pre><pre>Development:<br>https://api-id.execute-api.region.amazonaws.com/dev/resource-path</pre><pre>Staging:<br>https://api-id.execute-api.region.amazonaws.com/staging/resource-path</pre><pre>Production:<br>https://api-id.execute-api.region.amazonaws.com/prod/resource-path</pre><h3>Coming Next: AWS Cognito Integration</h3><p>In the next part of this series, we’ll cover how to:</p><ul><li>Set up separate Cognito user pools for dev and prod environments</li><li>Integrate Cognito with our existing API Gateway setup</li><li>Manage user authentication across different environments</li></ul><h3>Best Practices and Considerations</h3><ol><li><strong>Version Control</strong>: Always version your Lambda functions before deploying to production</li><li><strong>Environment Variables</strong>: Keep environment-specific configurations in environment variables</li><li><strong>Testing</strong>: Thoroughly test changes in dev and staging before promoting to production</li><li><strong>Monitoring</strong>: Set up separate CloudWatch alarms and metrics for each environment</li><li><strong>Access Control</strong>: Implement strict IAM policies to control who can deploy to different environments</li></ol><h3>Troubleshooting Common Issues</h3><p><strong>Permission Issues</strong></p><ul><li>Verify IAM roles and policies</li><li>Check Lambda execution role permissions</li><li>Ensure API Gateway has permission to invoke Lambda</li></ul><p><strong>Stage Variable Problems</strong></p><ul><li>Confirm stage variables are properly set</li><li>Verify Lambda alias exists</li><li>Check API Gateway logs for any integration errors</li></ul><p><strong>Deployment Issues</strong></p><ul><li>Clear API Gateway cache if changes aren’t reflecting</li><li>Verify correct Lambda version is associated with aliases</li><li>Check CloudWatch logs for detailed error messages</li></ul><h3>Conclusion</h3><p>This setup provides a robust foundation for managing multiple environments in your AWS serverless architecture. By following these steps, you can maintain separate development, staging, and production environments while ensuring proper isolation and configuration management.</p><p>Remember to always follow AWS best practices for security and resource management, and regularly review and update your configurations as your application evolves.</p><h3>Resources</h3><ul><li><a href="https://docs.aws.amazon.com/lambda/">AWS Lambda Documentation</a></li><li><a href="https://docs.aws.amazon.com/apigateway/">API Gateway Documentation</a></li><li><a href="https://docs.aws.amazon.com/cli/">AWS CLI Reference</a></li></ul><p>Hey there! I’m Parth Amin, an AWS Certified Solutions Architect Associate with a passion for cloud architecture and AI innovation. During the week, I help drive tech modernization initiatives as a Senior Consultant at Cognizant, where I guide enterprises through their cloud transformation journeys and manage large-scale migrations.</p><p>On weekends, you’ll find me building <a href="https://thisorthis.ai/">thisorthis.ai </a>— a platform I created to help people compare responses from multiple AI models side by side. If you’ve ever wondered which AI model works best for your specific needs, this tool lets you test them simultaneously, saving you valuable time and effort.</p><p>If you found this guide helpful, I’d love to connect with you on <a href="https://www.linkedin.com/in/aminparth/">LinkedIn</a>! Please feel free to share this guide with your network — knowledge grows by sharing, and I’m always excited to help fellow developers build better cloud solutions.</p><p><em>Want to discuss AWS architectures, AI implementations, or tech modernization? Let’s connect and explore ideas together!</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4af342f50bdb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Comparing Multiple AI Outputs Is Revolutionizing Decision-Making Across Industries]]></title>
            <link>https://parthsamin.medium.com/how-comparing-multiple-ai-outputs-is-revolutionizing-decision-making-across-industries-bf91895a886a?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf91895a886a</guid>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Tue, 08 Oct 2024 18:00:25 GMT</pubDate>
            <atom:updated>2024-10-14T02:15:06.274Z</atom:updated>
            <content:encoded><![CDATA[<p>In today’s rapidly evolving digital landscape, artificial intelligence (AI) has become an indispensable tool for professionals across all industries. However, as AI capabilities expand, a new challenge emerges: <strong>how do we ensure we’re getting the best possible AI-generated output for our specific needs?</strong> Enter <a href="https://thisorthis.ai/?ref=medium"><strong>thisorthis.ai</strong></a>, a groundbreaking platform that’s transforming the way we interact with AI by allowing users to compare outputs from multiple AI models simultaneously. In this blog post, we’ll explore how this innovative approach is revolutionizing decision-making across various sectors and why AI diversity is becoming crucial in our AI-driven world.</p><h3>The Problem with Single-Model AI Reliance</h3><p>Before we dive into the solution, let’s consider the problem. Most professionals today rely on a single AI model for their needs, whether it’s for content creation, data analysis, or problem-solving. While these models are powerful, they each have their own biases, strengths, and weaknesses. Relying on a single model can lead to:</p><p><strong>- Limited perspectives on complex problems<br>- Potential reinforcement of biases<br>- Missed opportunities for creative solutions<br>- Lack of confidence in AI-generated outputs</strong></p><p>This is where the power of AI diversity comes into play.</p><h3>Introducing thisorthis.ai: Harnessing the Power of AI Diversity</h3><p><a href="https://thisorthis.ai/?ref=medium">thisorthis.ai</a> is not just another AI platform; it’s a paradigm shift in how we interact with AI. By allowing users to compare outputs from multiple AI models side-by-side, it offers a unique solution to the limitations of single-model reliance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8krbJLeR47CTaEN9tK3mNg.png" /><figcaption>thisorthis.ai</figcaption></figure><p>Key features of thisorthis.ai:<br><strong>- Simultaneous comparison of up to 6 AI models<br>- Continuation of conversations with follow-up prompt<br>- Analyze and compare images using advanced AI vision capabilities<br>- Shareable URL for each thread, fostering collaboration and transparency</strong></p><p>But how does this translate to real-world value? Let’s explore some practical applications across different industries.</p><h3>Real-World Applications: AI Diversity in Action</h3><h4>1. Content Creation and Marketing</h4><p>Content creators and marketers are constantly seeking fresh, engaging ideas that resonate with their audience. With thisorthis.ai, they can generate and compare multiple content ideas, headlines, or even full articles from different AI models.</p><p>Example multi-prompt thread:</p><p>1. “<em>Generate 5 headline ideas for a blog post about sustainable fashion</em>”<br>2. “<em>Expand on headline #3 with an introductory paragraph</em>”<br>3. “<em>Suggest 3 key points to cover in the body of the article</em>”</p><p>Check out the below comparison at <a href="https://thisorthis.ai/s/JUbsEuPn">https://thisorthis.ai/s/JUbsEuPn</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p9oEeS0skZDIa3hh74yP_Q.png" /><figcaption>Content Creation multi-prompt AI comparison at thisorthis.ai</figcaption></figure><p>By comparing outputs, content creators can cherry-pick the best ideas from each model, resulting in more creative and effective content.</p><h4>2. Scientific Research and Academia</h4><p>Researchers can use thisorthis.ai to generate hypotheses, analyze data, and even draft sections of papers. The ability to compare outputs from multiple models can lead to novel insights and research directions.</p><p>Example multi-prompt thread:</p><p>1. “<em>Summarize recent findings on the relationship between gut microbiome and mental health</em>”<br>2. “<em>Generate 3 potential hypotheses for further research based on these findings</em>”<br>3. “<em>Outline a methodology to test hypothesis #2</em>”</p><p>Check out the below comparison at <a href="https://thisorthis.ai/s/x9NhXeAK">https://thisorthis.ai/s/x9NhXeAK</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aAWQqh4hL18AGsLxhXzHtw.png" /><figcaption>Scientific Research multi-prompt AI comparison at thisorthis.ai</figcaption></figure><p>This approach can accelerate the research process and uncover connections that might have been missed when relying on a single AI model or traditional research methods.</p><h4>3. Business Strategy and Entrepreneurship</h4><p>Entrepreneurs and business strategists can leverage thisorthis.ai to analyze market trends, generate business ideas, and develop strategic plans.</p><p>Example multi-prompt thread:</p><p>1. “<em>Analyze the current trends in the renewable energy sector</em>”<br>2. “<em>Suggest 5 innovative business ideas in this sector</em>”<br>3. “<em>Create a SWOT analysis for business idea #4</em>”</p><p>Check out the below comparison at <a href="https://thisorthis.ai/s/zjXIHaUP">https://thisorthis.ai/s/zjXIHaUP</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yOXf3XDRXp3LNIleF4Ts3A.png" /><figcaption>Business Strategy and Entrepreneurship multi-prompt AI comparison at thisorthis.ai</figcaption></figure><p>By comparing outputs from different models, entrepreneurs can gain a more comprehensive view of their market and potential opportunities, leading to more informed decision-making.</p><h4>4. Legal and Financial Services</h4><p>Professionals in legal and financial fields can use thisorthis.ai to research case law, analyze financial data, and generate reports.</p><p>Example multi-prompt thread:</p><p>1. “<em>Summarize recent Supreme Court decisions related to data privacy</em>”<br>2. “<em>Analyze the potential implications of these decisions for tech companies</em>”<br>3. “<em>Draft a memo outlining key compliance steps for a social media startup</em>”</p><p>Check out the below comparison at <a href="https://thisorthis.ai/s/VE8laN43">https://thisorthis.ai/s/VE8laN43</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*j6OgDiwDtTfW5hryRNnIvA.png" /><figcaption>Legal and Financial Services multi-prompt AI comparison at thisorthis.ai</figcaption></figure><p>This approach can help legal and financial professionals consider multiple perspectives on complex issues, leading to more robust advice and strategies.</p><h4>5. Product Development and UX Design</h4><p>Product managers and UX designers can use thisorthis.ai to generate and compare product ideas, feature suggestions, and user interface concepts.</p><p>Example multi-prompt thread:</p><p>1. “Generate 5 innovative features for a meditation app targeting young professionals”<br>2. “Create a user story for feature #3”<br>3. “Suggest 3 different UI designs to implement this feature”</p><p>Check out the below comparison at <a href="https://thisorthis.ai/s/fvHvkn1m">https://thisorthis.ai/s/fvHvkn1m</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HDnVxGHls6lGiSHHCLllxw.png" /><figcaption>Product Development and UX Design multi-prompt AI comparison at thisorthis.ai</figcaption></figure><p>By comparing outputs from different AI models, product teams can explore a wider range of possibilities and create more user-centric designs.</p><h3>The Future of AI-Assisted Decision Making</h3><p>As we’ve seen, the ability to compare outputs from multiple AI models is more than just a neat trick — it’s a powerful tool that can transform how we make decisions across various industries. <strong><em>But what does this mean for the future of work?</em></strong></p><p>1. <strong>Enhanced AI Literacy</strong>: As professionals become more accustomed to comparing AI outputs, they’ll develop a better understanding of AI strengths and limitations.</p><p>2. <strong>More Nuanced Decision-Making</strong>: The ability to consider multiple AI-generated perspectives will lead to more thoughtful, nuanced decisions.</p><p>3. <strong>Increased Creativity</strong>: Exposure to diverse AI outputs can spark new ideas and approaches that might not have been considered otherwise.</p><p>4. <strong>Greater Confidence in AI Utilization</strong>: By actively choosing the best output for their needs, users will feel more in control and confident in their use of AI.</p><p>5. <strong>Collaborative AI Interaction</strong>: The ability to share threads via URLs will foster more collaborative approaches to AI-assisted work.</p><h3>Tailored AI Guidance for Every Profession</h3><p><a href="https://thisorthis.ai/?ref=medium">thisorthis.ai</a> goes beyond just comparing AI outputs — it guides you in harnessing AI’s full potential for your specific profession. The platform’s innovative <a href="https://thisorthis.ai/ai-explorer?ref=medium">AI Explorer</a> section is a game-changer for professionals seeking to optimize their AI usage.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G7AyGeZNYTrHMQ237gVZOg.png" /><figcaption>AI Explorer at thisorthis.ai</figcaption></figure><p>This feature allows you to explore AI use cases and prompts tailored to various professions, from marketing and law to education and software development. By offering a curated selection of commonly used prompts for each field, AI Explorer helps you hit the ground running.</p><p>You can instantly compare how top AI models respond to these profession-specific prompts, gaining insights into which AI excels in different areas of your work. Whether you’re new to AI or looking to refine your prompts, AI Explorer is your compass in the vast sea of AI possibilities, helping you revolutionize your workflow with precision and efficiency.</p><h3>Getting Started with thisorthis.ai</h3><p>Ready to experience the power of AI diversity for yourself? Here’s how to get started:</p><p>1. Visit <a href="https://thisorthis.ai/?ref=medium">https://thisorthis.ai</a> and create an account.<br>2. Start a new thread by entering your first prompt (optionally add up to 5 images for analysis)<br>3. Compare the outputs from different AI models.<br>4. Continue the conversation with follow-up prompts.<br>5. Share your thread with colleagues using the unique URL.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_AB37yQrjyPcOTcqJVp2gQ.png" /></figure><p><em>Pro Tip</em>: Start with broad prompts and then use follow-up prompts to refine and specify. This allows you to see how different models handle both general and specific queries.</p><h3>Conclusion: Embracing AI Diversity</h3><p>In a world where AI is becoming increasingly prevalent, the ability to harness the power of AI diversity is a game-changer. <a href="https://thisorthis.ai/?ref=medium">thisorthis.ai</a> is at the forefront of this revolution, providing a platform that not only generates AI outputs but empowers users to make informed choices about which outputs best serve their needs.</p><p><strong>By comparing multiple AI perspectives, we open ourselves up to a world of possibilities — more creative solutions, more comprehensive analyses, and more confident decision-making</strong>. Whether you’re a content creator, researcher, entrepreneur, legal professional, or product developer, thisorthis.ai offers a new way to interact with AI that can transform your work.</p><p>As we move forward in this AI-driven era, tools like thisorthis.ai will play a crucial role in shaping how we leverage artificial intelligence. It’s not just about having AI assistance — it’s about having the right AI assistance for each unique situation.</p><p>Are you ready to experience the power of AI diversity? Visit <a href="https://thisorthis.ai/?ref=medium">thisorthis.ai</a> today and start revolutionizing your decision-making process.</p><blockquote>Remember, in the world of AI, it’s not about finding a one-size-fits-all solution — it’s about choosing the right output for your unique needs. With thisorthis.ai, that choice is at your fingertips.</blockquote><p>P.S. As the founder of thisorthis.ai, I’m deeply passionate about helping professionals harness the full potential of AI. I’m also the author of <a href="https://promptology.thisorthis.ai/">Promptology</a>, a weekly newsletter dedicated to the art and science of prompt engineering. If you found this article helpful or have any suggestions or feedback, I’d love to hear from you! Feel free to connect with me on <a href="https://www.linkedin.com/in/aminparth/">LinkedIn</a>. Let’s continue the conversation about the future of AI and how we can shape it together.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf91895a886a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Top 5 Use Cases for AI Model Comparison]]></title>
            <link>https://parthsamin.medium.com/top-5-use-cases-for-ai-model-comparison-71792551aba0?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/71792551aba0</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Tue, 25 Jun 2024 15:44:43 GMT</pubDate>
            <atom:updated>2024-06-25T16:07:30.065Z</atom:updated>
            <content:encoded><![CDATA[<p>In the rapidly evolving world of artificial intelligence, having options is crucial. But with the multitude of AI models available, how do you know which one is best for your specific needs? This is where AI model comparison comes into play. Let’s explore the top 5 use cases where comparing AI models can make a significant difference.</p><h3>1. Crafting Compelling Marketing Copy</h3><p>Marketers, rejoice! AI model comparison can be your secret weapon in creating attention-grabbing ad copy. Different models may excel at various aspects of copywriting — from crafting punchy headlines to developing persuasive body text. By comparing outputs from multiple models, you can cherry-pick the best elements to create a marketing message that truly resonates with your audience.</p><p>For instance, you might find that one model excels at creating catchy slogans, while another is better at writing detailed product descriptions. Combining these strengths can result in more effective, conversion-driving copy.</p><p>Try this prompt on thisorthis.ai: “Write a compelling 50-word product description for a new smart water bottle that tracks hydration levels and syncs with fitness apps.” — <a href="https://thisorthis.ai/s/Lgk563Iv">https://thisorthis.ai/s/Lgk563Iv</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cQYZTqpwisHcs1zipAm6DQ.png" /></figure><h3>2. Enhancing Customer Service Responses</h3><p>In the realm of customer service, speed and accuracy are paramount. AI model comparison allows companies to evaluate different AI-generated responses to common customer inquiries. This process helps identify which model provides the most helpful, empathetic, and brand-aligned responses.</p><p>By comparing models, you can ensure that your automated customer service maintains a consistent tone of voice while providing accurate information. This leads to improved customer satisfaction and reduced workload for human agents.</p><p>Try this prompt on thisorthis.ai: “Craft a polite and helpful response to a customer who received a damaged product and is requesting a refund or replacement.” — <a href="https://thisorthis.ai/s/bR4pNy7g">https://thisorthis.ai/s/bR4pNy7g</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Yn-1gntMTCh4YcJUwuAdKQ.png" /></figure><h3>3. Accelerating Software Development</h3><p>Developers, imagine having multiple AI assistants at your fingertips, each offering unique perspectives on your coding challenges. By comparing outputs from different AI models, you can:</p><p>- Debug more efficiently by cross-referencing suggested solutions<br>- Explore diverse approaches to algorithm implementation<br>- Generate more robust and optimized code snippets</p><p>This approach not only speeds up development but also encourages innovative problem-solving by exposing developers to a variety of coding styles and techniques.</p><p>Try this prompt on thisorthis.ai: “Write an optimized algorithm to find the shortest path in a weighted, directed graph using Dijkstra’s algorithm, handling up to 100,000 nodes and 1,000,000 edges.” — <a href="https://thisorthis.ai/s/kobirzUY">https://thisorthis.ai/s/kobirzUY</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Icky41OOXUfvVeOH0UECNA.png" /></figure><h3>4. Revolutionizing Healthcare Diagnostics</h3><p>In healthcare, accuracy can be a matter of life and death. AI model comparison in medical diagnosis can provide doctors with a powerful second opinion — or rather, multiple opinions. By comparing diagnostic suggestions from various AI models, healthcare professionals can:</p><p>- Identify potential oversights in initial diagnoses<br>- Explore less common but possible conditions<br>- Develop more comprehensive treatment plans</p><p>This multi-model approach could significantly reduce misdiagnoses and improve patient outcomes.</p><p>Try this prompt on thisorthis.ai: “Based on the following symptoms: persistent cough, fatigue, and unexplained weight loss, suggest three possible diagnoses and recommend appropriate next steps for diagnosis.” — <a href="https://thisorthis.ai/s/H5ZiMk6B">https://thisorthis.ai/s/H5ZiMk6B</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KiSAM_C8BJ1PPELLCt07PQ.png" /></figure><h3>5. Supercharging Creative Writing</h3><p>Writers and content creators can leverage AI model comparison to overcome writer’s block and spark creativity. Whether you’re crafting a novel, a screenplay, or a blog post, comparing outputs from different AI models can:</p><p>- Generate diverse plot ideas and character backstories<br>- Offer varied approaches to scene descriptions<br>- Suggest multiple directions for story development</p><p>This process can serve as a brainstorming tool, helping writers explore new creative avenues they might not have considered otherwise.</p><p>Try this prompt on thisorthis.ai: “Create an intriguing opening paragraph for a mystery novel set in a small coastal town, introducing the main character and hinting at a central conflict.” — <a href="https://thisorthis.ai/s/wi12a62Z">https://thisorthis.ai/s/wi12a62Z</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bYG1GlUSg45G1Iw7PIgvVw.png" /></figure><p><strong>Conclusion</strong>:</p><p>The power of AI model comparison lies in its ability to harness the unique strengths of multiple AI models, leading to more informed decisions and better outcomes. Whether you’re in marketing, customer service, software development, healthcare, or creative industries, comparing AI models can provide valuable insights and enhance your work.</p><p>As we continue to push the boundaries of what’s possible with AI, the ability to effectively compare and leverage different models will become increasingly crucial. So why settle for a single AI perspective when you can harness the collective intelligence of multiple models?</p><p>Ready to experience the power of AI model comparison for yourself? Visit <a href="https://thisorthis.ai/">thisorthis.ai</a> to start comparing AI models and unlocking new possibilities in your field!</p><p>A few more example prompt response comparisons are mentioned below:</p><ol><li>𝐂𝐫𝐞𝐚𝐭𝐢𝐯𝐞 𝐖𝐫𝐢𝐭𝐢𝐧𝐠 — “Write a short story about a time traveler who accidentally ends up in the year 3000, discovering a world ruled by artificial intelligence and robotic beings.” — <a href="https://thisorthis.ai/s/UUpXF63d">https://thisorthis.ai/s/UUpXF63d</a></li><li>𝐆𝐞𝐧𝐞𝐫𝐚𝐥 𝐊𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 — “Explain the theory of relativity in simple terms that a high school student could understand. Include both special and general relativity in your explanation.” — <a href="https://thisorthis.ai/s/JGcoHQGL">https://thisorthis.ai/s/JGcoHQGL</a></li><li>𝐏𝐫𝐨𝐟𝐞𝐬𝐬𝐢𝐨𝐧𝐚𝐥 𝐄𝐦𝐚𝐢𝐥 — “Compose a professional email to a client who is upset about a recent delay in the delivery of their order. Apologize for the inconvenience and offer a solution to rectify the situation.” — <a href="https://thisorthis.ai/s/0aJJWhdh">https://thisorthis.ai/s/0aJJWhdh</a></li><li>𝐇𝐢𝐬𝐭𝐨𝐫𝐢𝐜𝐚𝐥 𝐀𝐧𝐚𝐥𝐲𝐬𝐢𝐬 — “Discuss the key factors that led to the fall of the Roman Empire. Provide at least three major reasons and elaborate on their impact.” — <a href="https://thisorthis.ai/s/avaeMBbe">https://thisorthis.ai/s/avaeMBbe</a></li></ol><p>If you like this post, just want to connect or have any feedback on the same, please comment here or reach out directly via <a href="http://www.linkedin.com/in/aminparth">LinkedIn</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=71792551aba0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Design Thinking: The importance of asking the right questions]]></title>
            <link>https://medium.com/mytake/design-thinking-the-importance-of-asking-the-right-questions-b4deefaf5a27?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b4deefaf5a27</guid>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[design-process]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Tue, 26 Jul 2022 15:41:26 GMT</pubDate>
            <atom:updated>2023-02-03T16:17:01.099Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>“If I would have asked people what they wanted, they would have said faster horses. 🐎 🐎 🐎 ” — Henry Ford 🚘 .</blockquote><p>The process of design thinking starts with empathizing with the users and probing them to understand their pain points. This helps to accurately <em>identify the pain points</em> and <em>define the problem</em>.</p><p>There could be various channels through which one communicates with the user but it’s of utmost importance to ask them the right questions in a way that you could deep dive into their thought processes and environment.</p><p><strong>Product Managers should avoid asking questions that end in either a Yes or No.</strong> Instead, the focus should be to ask the right questions that give a deeper insight into the customer journey and how it could possibly help to identify the actual problem the customer is facing. Many times, a customer is not even aware of the problem he/she is facing, and asking direct binary questions is not going to help. <strong>PMs should also avoid stating probable solutions as part of their questions.</strong></p><p>For example, if you are a PM at Amazon working on Alexa<br>❌ The type of questions not to ask the customer:<br>◾ Have you used Alexa to do … ?<br>◾ Is using X feature on Alexa difficult for you?<br>◾ Would using [..a great solution..] be helpful to you?</p><p>✅ Instead, use the below type of questions to ask the customer:<br>◾ Please walk me through your experience when you use Alexa for …<br>◾ I noticed you said something about using feature X. Could you please elaborate?<br>◾ Would you like to change anything about your overall experience using feature X?</p><p>Design Thinking is a non-linear process that involves the following steps:<br>Empathize → Define → Ideate → Prototype → Test</p><p>You need to get the problem statement correct if you want to design truly innovative or actual solutions that the customer needs even if you know that the customer doesn’t want them. Its the duty of a product manager to ask the right questions, focus on the customer journey, and define the real problem.</p><blockquote>👉🏻 Again to sum this up, “It’s really hard to design products by focus groups. A lot of times, people don’t know what they want until you show it to them.” — Steve Jobs</blockquote><p><a href="https://www.linkedin.com/feed/hashtag/?keywords=designthinking&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#designthinking</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=productmanagement&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#productmanagement</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=productmanager&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#productmanager</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=empathy&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#empathy</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=customerresearch&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#customerresearch</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=designthinking101&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#designthinking101</a> <a href="https://www.linkedin.com/feed/hashtag/?keywords=productmanagement101&amp;highlightedUpdateUrns=urn%3Ali%3Aactivity%3A6957529119517483009">#productmanagement101</a></p><p>… Originally published on <a href="https://www.linkedin.com/posts/aminparth_designthinking-productmanagement-productmanager-activity-6957529119517483009-kLEH/">LinkedIn</a></p><blockquote>Check out valuable insights and fundamentals on Product Management on my YouTube channel — <a href="http://youtube.com/@_parthamin"><strong>Parth Amin’s YouTube Channel</strong></a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b4deefaf5a27" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mytake/design-thinking-the-importance-of-asking-the-right-questions-b4deefaf5a27">Design Thinking: The importance of asking the right questions</a> was originally published in <a href="https://medium.com/mytake">MyTake</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Improving digital payments for India]]></title>
            <link>https://medium.com/mytake/improving-digital-payments-for-india-53b291624f2a?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/53b291624f2a</guid>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[payments]]></category>
            <category><![CDATA[ideas]]></category>
            <category><![CDATA[india]]></category>
            <category><![CDATA[buy-now-pay-later]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Wed, 28 Jul 2021 11:40:19 GMT</pubDate>
            <atom:updated>2023-08-23T18:53:10.272Z</atom:updated>
            <content:encoded><![CDATA[<p>In India, e-Commerce works differently in various parts of the country especially in homes where payment decisions are made by only one person in the family or is someone who has the sole accessibility to make digital payments. I would like to highlight a particular problem and then suggest a solution that can be implemented by e-commerce companies or provided by digital payment solutions providers.</p><p><strong><em>Scenario 1</em></strong><em>: </em>Mom calling his son to place an order for an item she has recently added to her cart and is trying to checkout (COD is unavailable OR Prepaid has more discount)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jiBkmDpT7xzSbWxq.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4s7mN-NpTZzIFqQLegvx-w.png" /></figure><p><strong><em>Scenario 2</em></strong><em>: </em>Wife asking for OTP from husband to place an order with his credit/debit card details</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nLDYr1LWQNNrmnJcG34Ysw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f9GXzp5Pto072vL04GlYeA.png" /></figure><h3><strong>THE PROBLEM</strong></h3><p>The majority of the households in India, even today, have only a single person who uses internet banking, credit/debit cards, or UPI mode of payment for doing an online transaction. This results in heavy dependence by the household members to rely upon the only guy who knows or has the access to these digital payment mechanisms. In India, not everyone has access to debit or credit cards and even if they have many do not know to use them or are afraid to use them because of fraudulent transactions. So they have to rely on someone else to make payments for them.</p><p>More and more people are learning to browse and order on e-commerce apps with the ever-improving UI/UX of apps, learning through social media, and finding it easy to navigate them with vernacular options.</p><blockquote><em>However, at the time of checkout, dependence on someone else makes the checkout process look tedious, and often the customers have to let go of good deals because of delay in payments. They have to mainly wait for the person to return home to place an order or ask for an OTP to complete the payment transaction in case of having card details or just add to the cart and wait for him to place the order on your behalf through the e-commerce website or app.</em></blockquote><p>This breakdown during the checkout process often results in the customer never placing an order or the customer loses out on a good deal. Hence, the bounce rate from the checkout screen for such specific scenarios should be a lot higher which the e-commerce companies need to address.</p><h3>Affected Users</h3><ul><li>Parents who do not have a credit/debit card</li><li>Spouses who use a common credit/debit card for making online purchases (This is the case with many people who want to keep a single card for payments to easily track down their monthly spending)</li><li>People who just want to place the order on their friend’s behalf</li></ul><h3><strong>THE SOLUTION</strong></h3><p>The above problem can be solved simply by creating a payment link for each order that gets placed. It is more of like an extension of “ <strong>Buy Now, Pay Later</strong> “ but instead of paying it yourself, the other person makes the payment on your behalf for the order you just placed. Here’s how it can work:</p><ol><li>The customer (Mom) goes to the checkout page and selects the payment option say LinkPay or MiddlePay</li><li>As soon as the option is selected, Mom has to insert the name and mobile number of the person (her Son in this case) whom she wants to make the payment for her order</li><li>As soon as Mom clicks on “Place Order”, the order gets placed with an on-hold or payment due status and a payment link is sent to her son’s mobile number and email if provided</li><li>The order remains on hold for a few hours similar to how a ticket gets blocked for a few minutes until you make the payment. If the payment is not made, the order gets canceled automatically and the inventory kept on hold, if any, is set free in the vendor’s system</li><li>Now when the son gets the message and opens the link, he is able to see the cart details, the price, who has placed the order, which company is his mother purchasing from, and the hours he has to complete the payment on behalf of his mother</li><li>So the son can easily make the payment on his time and can also verify the purchase of the order with his mom in the time being</li><li>As soon as the son completes the payment, the status of the order changes to Payment Complete in the vendor system and it goes into processing</li></ol><h4>IDEA IN ACTION</h4><p>Let’s see how the solution discussed above can look in reality.</p><h4>Selection of payment mode and placing an order by the customer</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GLZUh-L7dn6GdMHwiAAUlA.jpeg" /></figure><h4>Making of the payment by friend/relative on behalf of the customer</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_NFhYtGyQhJzpqrRvGfQ6A.jpeg" /></figure><h4>DEMERITS / ASSUMPTIONS</h4><p>While the idea and solution may sound good enough, there are always concerns regarding its feasibility and implementation. A few of them which I would like to highlight are:</p><ol><li>Possibility of phishing and scamming people</li><li>Loss of potential sales if items go out of stock when they are on hold (payment is pending) for a longer duration</li><li>Additional costs to e-commerce companies for integrating the solution in their system (It can be avoided if payment aggregators like Cashfree, Razorpay, Stripe, etc. start providing the solution)</li></ol><p><em>I would love to hear some feedback. It will help me do better in the upcoming posts. Also, please feel free to comment with your views on this topic or DM me on </em><a href="http://www.linkedin.com/in/aminparth"><em>LinkedIn</em></a></p><blockquote>Check out valuable insights and fundamentals on Product Management on my YouTube channel — <a href="http://youtube.com/@_parthamin">Parth Amin’s YouTube Channel</a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53b291624f2a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mytake/improving-digital-payments-for-india-53b291624f2a">Improving digital payments for India</a> was originally published in <a href="https://medium.com/mytake">MyTake</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What is Gamification?]]></title>
            <link>https://parthsamin.medium.com/what-is-gamification-fa02f72c7c6b?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/fa02f72c7c6b</guid>
            <category><![CDATA[product]]></category>
            <category><![CDATA[games]]></category>
            <category><![CDATA[strategy]]></category>
            <category><![CDATA[gamification]]></category>
            <category><![CDATA[product-management]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Sat, 12 Sep 2020 16:50:53 GMT</pubDate>
            <atom:updated>2023-02-03T16:18:25.115Z</atom:updated>
            <content:encoded><![CDATA[<p>This term has recently gained momentum in the technology world and almost every now and then you will see job openings for Product Manager — Gamification or Gamification Expert vacancy and so on… But what actually are these companies looking for by hiring such people? To better understand this, let’s figure out the term first.</p><p>As a child, I grew up playing video games like Mario, Contra, Aladdin, etc. and have almost played all kinds of sports from cricket, basketball, kabaddi, tennis to squash. And then there also came some unique games, that I played during the weekends at my uncles’ home located in a village, like MaalDadi, Indoo, Marbles and many more. As I grew up and the technology evolved, I began to play strategy games like Roller Coaster Tycoon (though it only clicked to me that it was a strategy game at one of my marketing classes during my MBA) and many other games.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/414/1*hGPUr6edn_sA_bS1IP3t_A@2x.jpeg" /></figure><p>While playing all of these games or sports, one thing that I realize now is that there were certain aspects that led me to play and I am listing those out:</p><ul><li><strong>Fun</strong> (As the word says, you are game for fun — Its more about accomplishment)</li><li><strong>Friends</strong> (The word social wasn’t derived back then in detail as it is today. It was all about playing and enjoying together — All FOMO, Social inclination rest here. It’s opposite for single-player games or products in our case where the community is not required)</li><li><strong>Win / Loss</strong> (If you win, you are elevated! If you lose, you are elevated to try again, learn, improve and get back. You always start from basic and then level up)</li><li><strong>My Time</strong> (One wanted a reason to take a break from studies either during the weekends, holidays, vacations, or coaching classes for sports)</li><li><strong>Explore / Adventure</strong> (What’s coming next? If its a game, what will next level bring to the table — all about uncertainty)</li></ul><p>As you see, these elements were the primary motivators in playing games or sports. Many scholars have over the years worked in depth in elaborating these concepts and have come up with their own gamification models and frameworks. They feel that any industry can benefit if they involve gamification into their products or services so that the users are primarily motivated to consume their offerings and it in a way becomes a never-ending process. Lots of companies are shifting focus to gamification and all have different interests, while some want to increase the stickiness quotient, others are only limiting it to points and leaderboards.</p><p>One has to realize that gamification starts with a change from the core level of the product and an entire thought process has to be evolved around it. You start by measuring the core value, the true meaning of your product, and how it takes shape into an entire product.</p><p>In the last year, I have studied various gamification concepts and I would like to define Gamification on my own now. According to me,</p><blockquote>“Gamification is the process that makes a user self-motivated to use or explore your product or service offerings by methodically and secretly taking him from being a novice to an expert.”</blockquote><p>To make more sense in the definition above, just think what if Facebook launched today with all the features it has to offer right now, what will the new user think of it? It built step by step making the users play along in their product journey.</p><p>Companies are now on-boarding gamification experts to drive their future growth and make their products more meaningful to their customers while beating out the competition who will sadly fall behind. Gamification allows companies to focus on and refine their overall vision and strategy and brings together all the teams in the companies who have to work in synergy, and hence the need for Product Managers who can understand gamification and co-ordinate across all teams and departments. It is the way forward for all companies and soon you will see gamification experts across all industries. Many startups and established companies, like Swiggy, Meesho,etc., have already started hiring such breed of product people.</p><blockquote>Check out valuable insights, tips, strategies and fundamentals on Product Management on my YouTube channel — <a href="http://youtube.com/@_parthamin">Parth Amin’s YouTube Channel</a></blockquote><p>Originally published on 11th September 2020 via <a href="https://www.linkedin.com/pulse/what-gamification-parth-amin">LinkedIn</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fa02f72c7c6b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Heatmap with Python and GoogleMaps Tutorial]]></title>
            <link>https://parthsamin.medium.com/heatmap-with-python-and-googlemaps-tutorial-b3d623b77c95?source=rss-b8cd286e840f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b3d623b77c95</guid>
            <category><![CDATA[google-maps]]></category>
            <category><![CDATA[heatmap]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-visualization]]></category>
            <dc:creator><![CDATA[Parth Amin]]></dc:creator>
            <pubDate>Tue, 04 Aug 2020 19:06:38 GMT</pubDate>
            <atom:updated>2023-02-03T16:18:38.050Z</atom:updated>
            <content:encoded><![CDATA[<p>As a product manager, I always look to upgrade my skills and learn new things. But only learning doesn’t help unless you actually implement it and make its use.</p><p>One such thing happened with me in June 2020. I was pursuing Python for Data Science and Machine Learning Bootcamp from <a href="https://www.udemy.com/course/python-for-data-science-and-machine-learning-bootcamp/">Udemy</a> and we were into our weekly leadership meeting when Jitendra Rawal, Head of Operations at Frendy, started a discussion around measuring the density of orders being delivered in Covid-19 situation across the city so as to optimize operations and deliveries. Since we store the location of our delivery fleet when they mark the order as delivered from the app, we already have the lat-long co-ordinated of all the orders. So, the only task we had was to plot the heatmap on Google maps and we can easily have the data visualization.</p><p>We have a dedicated AI/ML team in the company, still, I thought that why not make use of this to test myself and put effective learning into practice. I told the team that I will be able to solve this for them and get back within 24 hours. And so I did it and here’s how you could do it too.</p><p>I used a Jupyter Notebook environment for running Python scripts which can be easily set up using <a href="https://www.anaconda.com/products/individual">Anaconda</a>. You will require to use Pandas library.</p><pre>import pandas as pd</pre><p>For beginners, just pressing Shift + Enter will execute your code line in the notebook.</p><p>Next, you will need to install the Google Maps library as the heat map will be plotted over the maps.</p><pre>import gmaps<br>import gmaps.datasets</pre><p>Once done, you will need to configure the Google Maps with an API Key for a Google project with <a href="https://developers.google.com/maps/documentation/javascript/get-api-key">Maps Javascript API</a> enabled. The API has a free-usage limit and you will need to enable billing to have its access. Don’t worry as Google Cloud Console offers $300 for first-time users for free and only using this API for learning purposes will never deplete the free quota. I still use it for free! Please make sure that you DO NOT share your API key in public.</p><pre>gmaps.configure(api_key=&#39;YOUR_GOOGLE_MAPS_API_KEY&#39;)</pre><p>So now it’s time to import the locations. First of all, create a .CSV file storing all the latitudes and longitudes. The title of columns A and B should be <em>latitude </em>and <em>longitude </em>respectively. When the file is ready, you will need to host it on cloud for direct read/download access. For convenience, I uploaded it on Amazon S3 and took its public link for use. You can use any of the hosting providers for media storage.</p><p>Now, you need to find and EDIT the file called “<strong>datasets.py</strong>” in gmaps package. It should be located mostly at an address like this -</p><p><em>C:\Users\Lenovo\anaconda3\pkgs\gmaps-0.9.0-py_0\site-packages\gmaps\datasets</em></p><p>Once found, please add the following code in the METADATA object of the file datasets.py</p><pre>&quot;PROJECT_NAME&quot;: {<br>        &quot;url&quot;: &quot;YOUR_PUBLIC_URL_OF_.CSV_FILE&quot;,<br>        &quot;description&quot;: &quot;ANY_DESCRIPTION_YOU_WANT&quot;,<br>        &quot;source&quot;: &quot;YOUR_WEBSITE&quot;,<br>        &quot;headers&quot;: [&quot;latitude&quot;, &quot;longitude&quot;],<br>        &quot;types&quot;: [float, float]<br>    },</pre><p>As you can see, PROJECT_NAME will be next called in through the GoogleMaps datasets in Python to load the CSV file. Once done, save this file and jump back to your jupyter notebook for running python scripts.</p><pre>locations = gmaps.datasets.load_dataset_as_df(&#39;PROJECT_NAME&#39;)<br>fig = gmaps.figure(map_type=&#39;HYBRID&#39;)<br>heatmap_layer = gmaps.heatmap_layer(locations)<br>fig.add_layer(heatmap_layer)<br>fig</pre><p>If everything goes well, you should be able to see a heatmap below. For the test purpose, I used a sample size of our actual data for rendering this, it shows the heat map plotted for the deliveries done by Frendy in the city of Ahmedabad for a few days.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vWHSKtPuQ6FPDWX66T6xfQ.png" /><figcaption>Heatmap plotted over Google Maps for visualizing orders density for e-commerce app Frendy</figcaption></figure><p>That is it. You can now toggle between Map and Satellite view or can also play around with some colours used for plotting by running the following code:</p><pre>heatmap_layer.max_intensity = 100<br>heatmap_layer.point_radius = 5</pre><pre>heatmap_layer.gradient = [<br>    &#39;silver&#39;,<br>    &#39;green&#39;,<br>    &#39;yellow&#39;<br>]</pre><p>The colour gradient of the heatmap is specified as an array of CSS colour strings. All <a href="http://www.w3.org/TR/css3-color/#html4">CSS3 colours</a> — including RGBA — are supported except for <a href="http://www.w3.org/TR/css3-color/#svg-color">extended named colours</a> and HSL(A) values.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*T7U5lXC6cfk5ttnDiiR3Hw.jpeg" /><figcaption>Customized Heatmap</figcaption></figure><p>You can come up with various variations. For the same data, I changed the gradient colours to also get this heatmap. It totally depends upon how you would like to visualize it.</p><p>For learning more on this, you can check out the below links that I found useful:</p><ul><li><a href="https://developers.google.com/maps/documentation/javascript/heatmaplayer">Heatmap Layer | Maps JavaScript API | Google Developers</a></li><li><a href="https://jupyter-gmaps.readthedocs.io/en/latest/tutorial.html">gmaps</a></li></ul><p>I hope you liked this simple tutorial on creating your own custom heatmaps with GoogleMaps and Python. Do reach me out on <a href="http://www.linkedin.com/in/aminparth">LinkedIn </a>to say ‘Hi’ if this works for you too, I would be glad to know that.</p><blockquote>Check out valuable insights, tips, strategies and fundamentals on Product Management on my YouTube channel — <a href="http://youtube.com/@_parthamin">Parth Amin’s YouTube Channel</a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b3d623b77c95" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>