<?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 To on Medium]]></title>
        <description><![CDATA[Stories by To on Medium]]></description>
        <link>https://medium.com/@tomdwipo?source=rss-9ae77b001532------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*DMOUuAJ7xq2oLjFHtj84mg.png</url>
            <title>Stories by To on Medium</title>
            <link>https://medium.com/@tomdwipo?source=rss-9ae77b001532------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 12:32:00 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@tomdwipo/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[Dari Executor ke Originator: Mengapa Skill Coding Aja Ga Cukup]]></title>
            <link>https://tomdwipo.medium.com/dari-executor-ke-originator-mengapa-skill-coding-aja-ga-cukup-13c63d134590?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/13c63d134590</guid>
            <category><![CDATA[pov]]></category>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[self-improvement]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Wed, 21 Jan 2026 13:10:26 GMT</pubDate>
            <atom:updated>2026-01-22T05:07:09.595Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rVmLJZWf_Zys4hpShbpMQg.png" /></figure><p>Seringkali, kita mengukur produktivitas dari seberapa sibuk tangan kita bekerja.</p><p>Sebagai <em>software engineer</em>, rasanya ada kepuasan tersendiri saat melihat tiket pekerjaan selesai. Tapi, pernah gak ngerasa udah kerja keras, tapi arah tujuan karir rasanya masih samar-samar?</p><p>Dulu, mikir kalo kunci kesuksesan adalah dengan <em>coding</em> lebih cepet. Gua kira jika bisa ngetik lebih banyak fitur dalam waktu singkat, hidup akan otomatis menjadi lebih baik.</p><p>Ternyata, kecepatan hanyalah satu bagian kecil dari perjalanan ini.</p><p>Setelah menjalani peran sebagai <em>engineer</em> di <em>corporate</em>, sambil merintis <em>startup</em>, dan mempersiapkan karir lebih baik, gua nyadari bahwa kita perlu melihat karir kita layaknya sebuah arsitektur bangunan. Ada tahapan dan struktur yang perlu kita pahami agar tidak terjebak dalam rutinitas aja.</p><p>gua menyebutnya <strong>“4 Level Strategi Hidup”</strong>. Mari kita coba refleksikan bersama, ada di level mana kita saat ini.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y8e5l0pq0ZX4nCpZ4kKPyA.png" /></figure><h3>Level 1: The Executor (Pelaksana)</h3><p>Ini adalah fondasi awal kita. Di fase ini, mata uang utama kita adalah <strong>Waktu</strong>.</p><p>Fokus utama kita sederhana: menyelesaikan tugas yang diberikan dengan baik. Kita menukar 8 jam sehari untuk mendapatkan gaji bulanan.</p><p>Tidak ada yang salah dengan fase ini; ini adalah tempat kita membangun <em>hard skill</em>. Namun, perlu diingat bahwa waktu kita terbatas 24 jam. Jika kita hanya mengandalkan kecepatan tangan, penghasilan dan dampak kita akan memiliki batas atap yang sulit ditembus. Kita perlu pondasi ini, tapi kita tidak bisa tinggal di sini selamanya.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*5vYdnD2oeXuip0N0cB15yA.png" /></figure><h3>Level 2: The Strategic Executor (Pelaksana Strategis)</h3><p>Di level ini, kita mulai bekerja lebih cerdas dengan menggunakan <strong>Leverage</strong>.</p><p>Kita tidak lagi hanya mengandalkan tenaga sendiri. Kita mulai memanfaatkan alat bantu kayak AI atau <em>automation</em> untuk menyelesaikan pekerjaan rutin dengan lebih efisien. Tugas yang dulu butuh waktu 4 jam, kini bisa selesai dalam 1 jam.</p><p>Di sini, kita juga mulai memahami nilai pasar. Kita sadar bahwa <em>skill</em> yang sama bisa dihargai berbeda di tempat lain. Kita mulai menukar “Efisiensi”, bukan sekadar waktu. Hasilnya, kita memiliki ruang bernapas lebih banyak untuk belajar dan berkembang.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*fp7HjQ5EzP3jwtEYj8y9tg.png" /></figure><h3>Level 3: The Partner &amp; Judge (Mitra Strategis)</h3><p>Pergeseran terbesar terjadi di sini. Kita berubah dari “Pembuat Kode” menjadi “Pemberi Solusi”.</p><p>Jika di kantor, teman-teman bukan lagi sekadar menerima instruksi teknis, tapi mulai menjadi mitra diskusi bagi atasan. Kita mulai menerjemahkan bahasa teknis yang rumit menjadi bahasa bisnis yang dimengerti semua orang. Kita berani memberikan masukan: <em>“Apakah fitur ini benar-benar dibutuhkan oleh user? Atau ada cara lain yang lebih efektif?”</em></p><p>Di level ini, nilai kita bukan diukur dari seberapa banyak kita mengetik, tapi dari seberapa tepat keputusan yang kita ambil. Kita membantu perusahaan menghindari risiko dan mencapai tujuan lebih cepat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*HrBYzVyoFKVaibMGLeFYUw.png" /></figure><h3>Level 4: The Originator (Pencetus)</h3><p>Ini adalah level tentang <strong>Makna</strong> dan <strong>Tujuan</strong>.</p><p>Ketika kebutuhan dasar sudah terpenuhi, pertanyaan kita berubah: <em>“Apa yang ingin gua wujudkan?”</em></p><p>Di sini, teman-teman berperan sebagai arsitek kehidupan. Kita menggunakan energi dan modal yang kita punya untuk membangun sesuatu yang kita percayai bisa berupa produk, komunitas, atau <em>legacy</em> lainnya.</p><p>Teknologi (AI) bisa membantu kita bekerja (Level 1), membantu kita efisien (Level 2), bahkan membantu menganalisis data (Level 3). Namun, teknologi tidak memiliki “niat”. Hanya kitalah, manusia, yang memiliki visi untuk memulai sesuatu dari nol.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/824/1*eHDIA_lKBuWxf6FOtVCOHA.png" /></figure><h3>Menerapkan “Mindset Fraktal”</h3><p>Kabar baiknya, kita tidak perlu menunggu puluhan tahun atau harus berhenti bekerja untuk naik level. Kita bisa menerapkannya sekarang juga, di mana pun kita berada.</p><p>Kita bisa tetap menjadi karyawan yang baik, namun beroperasi dengan pola pikir yang lebih luas:</p><ol><li>Gunakan teknologi untuk bekerja efisien (<strong>Level 2</strong>).</li><li>Gunakan waktu yang ada untuk memberikan solusi strategis bagi tim (<strong>Level 3</strong>).</li><li>Pandang pekerjaan saat ini sebagai sarana untuk mendukung tujuan jangka panjang kita (<strong>Level 4</strong>).</li></ol><p>Jadi, besok saat kita mulai bekerja, mari luangkan sejenak waktu untuk bertanya:</p><blockquote>“Apakah hari ini gua hanya akan menyusun batu bata, atau sedang membangun sebuah istana?”</blockquote><p>Semoga bermanfaat untuk perjalanan teman-teman semua.</p><p><strong>Tentang Penulis:</strong> <em>Sehari-hari bekerja sebagai Mobile Engineer, sambil bangun komunitas “</em><a href="https://medium.com/is-a-code/newsletter"><em>is a code</em></a><em>” bersama </em><a href="https://medium.com/u/2643cb14422d"><em>isa nur fajar</em></a><em> .</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=13c63d134590" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Refactoring Your Headspace, The Indie Developer Journey, and The Memory Cost of Closures]]></title>
            <link>https://medium.com/is-a-code/refactoring-your-headspace-the-indie-developer-journey-and-the-memory-cost-of-closures-d46fd5c3a14f?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/d46fd5c3a14f</guid>
            <category><![CDATA[thoughts]]></category>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Sun, 28 Dec 2025 04:01:39 GMT</pubDate>
            <atom:updated>2025-12-28T04:01:39.934Z</atom:updated>
            <content:encoded><![CDATA[<p>Welcome back to <em>Beyond Syntax: </em>the publication for engineers, builders, and wanderers.</p><p>This week, we are exploring the full stack of the developer experience. We start high-level, looking at the “operating system” of our own minds, then move to the career layer with an inspiring solo-dev journey, and finally, we drop down to the metal to debug the memory layout of Swift closures.</p><p>Here is your weekly build.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AJY18B5cxvnJ592j9OCb0w.png" /></figure><h3>System Maintenance: Debugging Your Brain</h3><p><a href="https://medium.com/is-a-code/life-feels-heavy-and-laggy-it-s-time-to-refactor-your-headspace-6c664f7004a5"><strong>Life Feels Heavy and ‘Laggy’? It’s Time to Refactor Your Headspace</strong></a> <em>By Ahmad Eko Kurniawan</em></p><p>Do you ever wake up feeling like your RAM is already at 90% usage? In the tech world, we call this “high latency.” In life, we just call it burnout.</p><p>Ahmad Eko Kurniawan argues that we are the Lead Engineers of our own lives, yet we often run on “Spaghetti Code”: tangled habits and unclosed background processes like anxiety and regret. This article applies strict engineering principles to mental health:</p><ul><li><strong>Garbage Collection:</strong> Clear yesterday’s emotional cache before starting today.</li><li><strong>Pay Technical Debt:</strong> That difficult conversation you’re avoiding? It’s legacy code. Refactor it now before it breaks the system.</li><li><strong>Documentation:</strong> Use writing to debug your thoughts and structure your mind.</li></ul><blockquote>“A great engineer is not judged by how fast they type code, but by how neat, robust, and scalable the system they build is.”</blockquote><h3>The Builder’s Path: From Confusion to 31 Apps</h3><p><a href="https://medium.com/is-a-code/one-person-changed-the-direction-of-my-dream-from-confusion-to-building-products-86ff15604bd2"><strong>One Person Changed the Direction of My Dream</strong></a> <em>By VeryCoolApps (Andra)</em></p><p>Credentials don’t always equal clarity. Adhithya (Andra) had the Master’s degree and the corporate job, but he felt lost — until he saw the success of <em>Flappy Bird</em>. It wasn’t the game’s complexity that shook him; it was the realization that <strong>one person can build something that reaches millions.</strong></p><p>This is a candid look at the transition from corporate confusion to the “Indie Hacker” life. After getting banned by Google (a nightmare scenario) and pivoting to the Apple ecosystem, Andra has now shipped <strong>31 applications</strong>.</p><ul><li><strong>The Lesson:</strong> You don’t need a team to start. You need the discipline to finish.</li><li><strong>The Strategy:</strong> Build small, focus on utility, and let improvement compound over time.</li></ul><h3>Deep Dive: The Devil is in the Memory Layout</h3><p><a href="https://medium.com/is-a-code/ios-interview-swift-closures-devils-advocate-e1e749fbd7ef"><strong>iOS Interview: Swift Closures Devil’s Advocate</strong></a> <em>By Finn Christoffer Kurniawan</em></p><p>If you think a closure is just a block of code you pass around, you might fail the senior interview. Finn plays “Devil’s Advocate” to expose what actually happens under the hood when you type { }.</p><ul><li><strong>The Physics:</strong> A closure is a 16-byte struct (on 64-bit systems) acting as a “Fat Pointer” to both instructions and a heap context.</li><li><strong>The Trap:</strong> Swift captures variables by <em>reference</em>, not value. (Do you know why your counter prints 10 instead of 0?)</li><li><strong>The Toolset:</strong> A clear breakdown of @escaping (heap allocation) vs. @autoclosure (lazy evaluation) and how to use them to write cleaner, faster code.</li></ul><h3>Reality Check</h3><p><a href="https://medium.com/is-a-code/the-reality-of-software-development-problems-are-often-not-about-coding-584ed33a45fe"><strong>The Reality of Software Development: Problems Are Often Not About Coding</strong></a></p><p>We end with a reminder that the hardest bugs to fix are usually human ones. Whether it’s vague requirements or miscommunication, this story explores why “soft skills” are actually the “hard skills” of senior development.</p><p><em>Code is more than syntax. It’s how we think.</em> <strong>See you in the next commit.</strong></p><p>— The <em>Beyond Syntax</em> Team</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d46fd5c3a14f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/is-a-code/refactoring-your-headspace-the-indie-developer-journey-and-the-memory-cost-of-closures-d46fd5c3a14f">Refactoring Your Headspace, The Indie Developer Journey, and The Memory Cost of Closures</a> was originally published in <a href="https://medium.com/is-a-code">Is a code</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Kenapa Senior Engineer Jadi Yang Paling Lama Adopsi AI? (Data & Fakta)]]></title>
            <link>https://tomdwipo.medium.com/kenapa-senior-engineer-malah-jadi-penghambat-adopsi-ai-data-fakta-504ab368e29b?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/504ab368e29b</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[thoughts]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Thu, 25 Dec 2025 05:42:00 GMT</pubDate>
            <atom:updated>2025-12-25T12:51:14.421Z</atom:updated>
            <content:encoded><![CDATA[<p>belajar dari “Quartz Crisis” dan kenapa pengalaman 10 tahun kita bisa jadi beban di era “Vibe Coding”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ozd2xtFlc3j8umH_PJ8TRQ.png" /></figure><p>Ada sebuah anomali aneh yang sedang terjadi di dunia software engineering saat ini.</p><p>Biasanya, Senior Engineer adalah orang yang paling depan dalam mengadopsi teknologi baru. Tapi untuk kasus Generative AI, datanya menunjukkan sebaliknya. Justru <strong>Senior dan Staff Engineer</strong>-lah kelompok yang paling resisten, skeptis, dan lambat mengadopsi AI.</p><p>Mereka bilang: <em>“Ah, kodenya jelek,”</em> atau <em>“Gue lebih cepet ngetik sendiri,”</em> atau <em>“Ini bukan real engineering.”</em></p><p>Dalam presentasi terbaru, <strong>Steve Yegge</strong> (Ex-Google/Amazon) dan <strong>Gene Kim</strong> (Penulis <em>The Phoenix Project</em>) membicarakan fenomena ini. Mereka menyebutnya sebagai <strong>“The Senior Engineer Curse”.</strong></p><p>Kenapa ini terjadi? Dan kenapa pola pikir ini bisa membunuh karir kita dalam 1–2 tahun ke depan? Mari kita bedah.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m62dfKo-8IJOkstywuSR4A.png" /></figure><h3>1. The New Quartz Crisis: Jebakan “Craftsmanship”</h3><p>Steve Yegge menggunakan analogi sejarah yang relevan: <strong>Krisis Jam Quartz</strong>.</p><p>Dulu, pembuat jam Swiss (<em>watchmakers</em>) adalah kasta tertinggi. Mereka merakit jam mekanik dengan tangan, penuh seni, dan presisi. Itu adalah “real work”. Lalu, jam Quartz (baterai/digital) muncul.</p><ul><li>Watchmaker bilang: <em>“Itu barang murahan. Nggak ada jiwanya.”</em></li><li>Pasar bilang: <em>“Ini lebih akurat, lebih murah, dan tahan lama.”</em></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0xaKoJSyBLnva9Fg6jfWyQ.png" /></figure><p>Hasilnya? Industri jam Swiss digilas otomatisasi.</p><p>Banyak Senior Engineer hari ini terjebak di mentalitas “Watchmaker”. Mereka merasa coding manual mengetik sintaks, mengatur indentasi, memikirkan boilerplate adalah sebuah seni yang sakral.</p><p>Padahal, AI mengubah coding dari <strong>Kerajinan Tangan (Handcrafted)</strong> menjadi <strong>Komoditas Industri</strong>. Jika kita menolak menggunakan “mesin bor listrik” (AI) dan ngotot pakai “obeng manual” demi idealisme <em>craftsmanship</em>, kita akan kalah telak dari junior yang menggunakan mesin.</p><h3>2. Salah Kaprah Teknis: Masalah “Penyelam” (The Diver Problem)</h3><p>Alasan kedua kenapa Senior Engineer benci AI adalah karena <strong>mereka salah cara pakainya</strong>, lalu menyalahkan alatnya.</p><p>Banyak senior mencoba AI dengan cara: <em>“Hey ChatGPT, tolong refactor satu file 1000 baris ini, fix bug-nya, dan buat test case.”</em> Lalu AI-nya gagal atau halusinasi. Senior pun bilang: <em>“Tuh kan, AI bodoh.”</em></p><p>Steve Yegge menjelaskan ini dengan <strong>“The Diver Problem”</strong>.</p><ul><li>Context Window AI itu ibarat tabung oksigen penyelam.</li><li>Kita nggak bisa kirim satu penyelam (Single Agent) buat ngerjain tugas raksasa di laut dalam. Napasnya (token-nya) bakal habis, dan dia bakal mati (halusinasi).</li></ul><p>Software engineering itu tentang <strong>Decomposition</strong> dan <strong>Modularity</strong>. Masa depan coding bukan “Satu Chatbot Dewa”, tapi <strong>Koloni Semut (Ant Colony)</strong>.</p><ul><li>Satu agen kecil (murah) buat cek .gitignore.</li><li>Satu agen buat nulis fungsi.</li><li>Satu agen buat review.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IDsFCiDUKHuF0w0eytEfsQ.png" /></figure><p>Senior Engineer yang cerdas bukan yang jago ngetik, tapi yang jago <strong>mengorkestrasi koloni semut</strong> ini.</p><h3>3. Timeline Evolusi</h3><p>Kalau kita masih berdebat soal <em>prompt engineering</em> di ChatGPT atau Cursor, <em>bad news</em>: kita sebenernya udah ketinggalan zaman atau akan berhenti disitu aja.</p><p>Slide presentasi Steve menunjukkan <strong>AI Developer Trajectory</strong>:</p><ul><li><strong>2023:</strong> Code Completions (Copilot).</li><li><strong>2024:</strong> Chat-based (Cursor, Windsurf). Status: <strong>OLD</strong>.</li><li><strong>2025:</strong> Agentic Coding Assistants (Claude Code, Gemini CLI).</li><li><strong>2026:</strong> <strong>Orchestrators</strong>.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BE2VFZbKNpDOh1WOzK6ksw.png" /></figure><p>Di tahun 2026, diprediksi kita akan masuk fase <strong>Orchestrators</strong>, di mana manusia mungkin tidak lagi melihat kode secara langsung, tapi mengelola sistem yang menulis kode tersebut. Senior yang masih bertahan di fase “Chat” akan enggak bisa maju.</p><h3>4. Definisi Baru: “Vibe Coding” &amp; Fenomena NoDev</h3><p>Gene Kim memperkenalkan istilah <strong>“Vibe Coding”</strong> dengan definisi radikal:</p><blockquote>“Anything besides typing in code by hand.”</blockquote><p>Dr. Erik Meijer (legenda pembuat C# &amp; LINQ) bahkan berkata: <em>“Kita mungkin generasi terakhir yang menulis kode dengan tangan. Let’s have fun doing it.”</em>.</p><p>Dampaknya? Muncul fenomena <strong>NoDev</strong>. Bukan cuma <em>NoOps</em> (Developer ngerjain Ops), tapi <em>NoDev</em> — di mana orang non-teknis bisa nge-ship fitur sendiri.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zc7tsLAAsEPDy1tnop5SMA.png" /></figure><p>Contoh: Dr. Top Pal (VP di Fidelity) butuh dashboard internal. Tim engineer-nya bilang butuh <strong>5 bulan</strong>. Dia frustrasi, lalu dia <em>vibe coding</em> sendiri pakai AI. Hasilnya? <strong>Jadi dalam 5 hari</strong>. Tim engineer-nya syok. Bos mereka yang “nggak ngoding” bisa mengalahkan kecepatan satu tim.</p><p>Inilah data “Elite vs Low Performers” di era AI: Perbedaannya bukan 2x atau 3x, tapi <strong>208x lebih sering deploy</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zCRVEVk7RPhtl6HC6_ghfg.png" /></figure><h3>Kesimpulan: Beradaptasilah</h3><p>Steve Yegge ngoding 12–15 jam sehari sekarang. Bukan karena terpaksa, tapi karena sensasi membangun sistem raksasa sendirian (<em>Ambition &amp; Alone</em>) itu memberikan dopamine yang luar biasa.</p><p>Jadi, buat para Senior Engineer: Pengalaman 10 tahun kita sangat berharga, TAPI hanya jika kita pakai itu buat <strong>mengarahkan AI</strong>, bukan buat <strong>bersaing dengan AI</strong>.</p><p>Jangan jadi pembuat jam Swiss yang keras kepala. Jadilah operator mesin CNC yang presisi. Karena suka atau tidak, <em>automation always wins</em>.</p><p><strong>Sumber &amp; Kredit:</strong> Artikel ini merangkum <em>keynote</em> dari <strong>Steve Yegge</strong> &amp; <strong>Gene Kim</strong> tentang buku mereka<em>“Vibe Coding”</em>. Semua grafik dan data diambil langsung dari slide presentasi mereka: <a href="https://www.youtube.com/watch?v=7Dtu2bilcFs">Steve Yegge &amp; Gene Kim Talk</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=504ab368e29b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ini Strategi Biar Tim Mau Pake AI]]></title>
            <link>https://tomdwipo.medium.com/ini-strategi-biar-tim-mau-pake-ai-df3075dc0e7e?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/df3075dc0e7e</guid>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Mon, 22 Dec 2025 16:38:05 GMT</pubDate>
            <atom:updated>2025-12-22T16:39:06.126Z</atom:updated>
            <content:encoded><![CDATA[<p>Kenapa maksa itu nggak mempan, dan gimana cara bikin engineer cinta sama AI (bukan takut digantiin).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4c036G24g26emFkXcFrkTA.png" /></figure><p>Skenario klasik di banyak kantor tech belakangan ini: Management beli license claude/Cursor buat satu divisi. Mahal. Terus Management bilang: <em>“Mulai besok semua pake ya, biar produktivitas naik 50%!”</em></p><p>Sebulan kemudian dicek… Utilisasi rendah. Engineer masih coding cara lama. Atau mereka pake AI, tapi hasilnya malah banyak bug.</p><p>Management bingung: <em>“Kita udah kasih tools bagus, kok tim nggak mau pake?”</em></p><p>Jawabannya bukan karena engineer gapaham. Jawabannya ada di data yang dibongkar sama <strong>Justin Reock</strong> (Chief Evangelist di DX, acq. Atlassian). Ternyata, <strong>mandat top-down itu strategi paling nggak efektif.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HD_9vDXJVleANTAtab318Q.png" /></figure><p>Kalau mau tim <em>beneran</em> mau pake AI (dan dapet value-nya), lo harus ubah strateginya. Ini rangkumannya.</p><h3>1. Ganti Jualan “Productivity”, ke Jualan “Happiness”</h3><p>Kesalahan terbesar leader adalah jualan AI dengan narasi: <em>“Pake ini biar kalian kerja lebih cepet.”</em></p><p>Di kuping engineer, itu kedengerannya kayak: <em>“Pake ini biar gue bisa kasih kalian kerjaan lebih banyak.”</em> 🚩</p><p>Justin nunjukin data dari Google (Project Aristotle): Faktor nomor satu tim performa tinggi adalah <strong>Psychological Safety</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gKd1hg5L-UjvDeSbO2UvNw.png" /></figure><p>Kalau engineer takut AI bakal ngegantiin mereka, mereka bakal defensif. Strateginya: Ubah narasi. Jangan bilang AI itu <em>Replacement</em>, tapi bilang AI itu <strong>Augmentation (ningkatin value)</strong>.</p><ul><li><em>“Pake AI biar lo nggak usah ngerjain boilerplate code yang ngebosenin.”</em></li><li><em>“Pake AI biar lo nggak stres pas on-call jam 3 pagi.”</em></li></ul><p>Fokus ke <em>Developer Experience (DX)</em>, bukan cuma output perusahaan.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cIhPnh-NmrgJ-zB7V7qOvA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XmTlgbnlFKjG_-cJQhMW4w.png" /></figure><h3>2. Jangan Cuma “Nulis Code”, Cari Bottleneck nya</h3><p>Banyak tim males pake AI karena mereka ngerasa nulis code itu bukan masalah utama mereka. <em>“Ngapain gue generate code cepet-cepet kalau nunggu review-nya 3 hari?”</em> bener juga kan?</p><p>Strategi Justin: Gunakan <strong>Theory of Constraints</strong>. Cari apa yang bikin engineer paling menderita (<em>pain points</em>), terus tawarin AI sebagai obatnya.</p><p>Contoh sukses:</p><ul><li><strong>Zapier:</strong> Engineer barunya sering bingung (<em>bottleneck di onboarding</em>). Mereka bikin AI bot buat training. Engineer seneng karena nggak perlu tanya-tanya senior terus, Senior seneng karena nggak diganggu pertanyaan basic. Win-win.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IP68NbaQZAYj59SVX6pqSg.png" /></figure><ul><li>Modernisasi kode legacy (contoh: Morgan Stanley).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8IPWwXPaeCIzkOXSh29nwQ.png" /></figure><p>Cari di mana tim lo paling “sakit”, kasih AI di situ. Mereka pasti mau pake.</p><h3>3. Kasih “Waktu Main” (Time to Learn)</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7FTehqIKyPWg-GGjPdC72Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3MJXWQ_lU6emClKSGtb5Mg.png" /></figure><p>Ini ironis. Kita nyuruh tim pake tools baru biar cepet, tapi nggak ngasih waktu buat belajar. Akibatnya, mereka pake AI asal-asalan, kualitas code turun (di data industri, ada kasus produktivitas malah <strong>turun 19%</strong> karena salah pake).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SZOTL4CQLEsRcQa9nyfXWQ.png" /></figure><p>Strateginya: <strong>Alokasikan waktu resmi.</strong> Jangan harap mereka langsung jago. Kasih waktu di sprint buat “eksplorasi AI”. Biarkan mereka gagal dulu, biarkan mereka nemu <em>prompt</em> yang pas.</p><p>Justin bilang, organisasi yang sukses itu yang memberikan ruang buat <em>learning curve</em>, bukan yang nuntut hasil instan di hari pertama.</p><h3>4. Siapkan “Pagar” Biar Nggak Nyasar (Governance)</h3><p>Kadang tim males pake AI karena hasilnya ngaco. <em>“Ah, AI ngasih code library tahun jebot. Males gue benerinnya, mending nulis sendiri.”</em></p><p>Strateginya: Leadership harus siapin <strong>System Prompts / Gatekeeper</strong>. Atur di level organisasi supaya AI-nya “pinter” konteks perusahaan lo.</p><ul><li>Pastikan AI tau coding standard kantor.</li><li>Pastikan AI tau versi library yang dipake.</li></ul><p>Kalau AI-nya ngebantu (akurat), tim bakal pake dengan sukarela. Kalau AI-nya halu, ya ditinggalin.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uv4k0imqrVm-c5nANzaqaA.png" /></figure><h3>Kesimpulan</h3><p>Biar tim mau pake AI, kuncinya bukan <strong>Pemaksaan</strong>, tapi <strong>Kepercayaan</strong>.</p><ol><li>Bangun <strong>Psychological Safety</strong> (Lo aman, AI cuma asisten).</li><li>Selesaikan <strong>Masalah Nyata</strong> mereka (Bukan cuma masalah perusahaan).</li><li>Kasih <strong>Waktu Belajar</strong>.</li></ol><p>Kalau lo bisa lakuin ini, lo nggak perlu maksa. Mereka sendiri yang bakal minta license-nya diperpanjang.</p><p><strong>Credit:</strong> Artikel ini adalah rangkuman strategi dari talk <strong>Justin Reock (DX, acq. Atlassian)</strong> di channel <strong>AI Engineer</strong> dengan judul <em>“Leadership in AI Assisted Engineering”</em>. Nonton video lengkapnya di sini buat insight lebih dalam: <a href="https://www.youtube.com/watch?v=PmZDupFP3UM">Leadership in AI Assisted Engineering — Justin Reock</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=df3075dc0e7e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Jangan Cuma Jadi Prompter: Gimana Caranya Tetap Jadi Engineer yang Paham Code Sendiri]]></title>
            <link>https://tomdwipo.medium.com/jangan-cuma-jadi-prompter-gimana-caranya-tetap-jadi-engineer-yang-paham-code-sendiri-1a1f77e25199?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/1a1f77e25199</guid>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Mon, 22 Dec 2025 15:14:47 GMT</pubDate>
            <atom:updated>2025-12-22T17:15:23.058Z</atom:updated>
            <content:encoded><![CDATA[<p>Kenapa “ngobrol” doang sama AI itu bahaya dan harus gunain framework? biar lo nggak kehilangan kendali sama software lo sendiri.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*59WSAsUiG1Mggmsw0jQMpQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mo6MbE25fOIzPvvzbxDJog.png" /></figure><p>Gue mau mulai dengan pengakuan dosa.</p><p>Gue pernah generate code pake AI, run test (pass), terus langsung deploy ke production. Tapi kalau lo tanya gimana persisnya code itu bekerja baris per baris? Gue mungkin bakal bingung ngejelasinnya.</p><p>Dan tebakan gue: <strong>Lo pasti juga pernah.</strong></p><p>Nggak usah malu. Kita semua ada di situasi yang sama. Sejak ada ChatGPT, Claude, Cursor, dan teman-temannya, backlog yang tadinya butuh berhari-hari sekarang kelar dalam hitungan jam. Rasanya produktif banget, kan? Speed coding kita naik drastis.</p><p>Tapi ada harga mahal yang diam-diam kita bayar: <strong>Understanding.</strong></p><p>Masalahnya, sistem production itu pasti bakal error. Entah itu <em>race condition</em> yang muncul pas <em>traffic</em> tinggi, atau <em>dependency</em> yang konflik setelah update. Dan ketika sistem down jam 3 pagi, lo nggak bisa debug code yang logika dasarnya aja lo nggak ngerti.</p><p>Ini bukan salah AI-nya. Ini salah kita yang kena jebakan bedanya “Easy” sama “Simple”.</p><h3>Jebakan Batman: “Easy” vs “Simple”</h3><p>Rich Hickey (creator Clojure) punya talk legendaris soal ini. Kita sering nuker arti kata <strong>Easy</strong> dan <strong>Simple</strong>, padahal di dunia engineering, itu beda kasta.</p><ul><li><strong>Easy (Mudah):</strong> Sesuatu yang deket, bisa dijangkau instan. Install library magic, copy-paste StackOverflow, atau minta AI generate code 50 baris dalam 2 detik. Itu <em>Easy</em>.</li><li><strong>Simple (Sederhana):</strong> Sesuatu yang terstruktur, satu fungsi jelas, nggak mbulet (<em>entangled</em>), dan mudah dipahami alurnya. Ini butuh mikir dan desain.</li></ul><p>AI itu adalah tombol “Easy” paling gokil benernya.</p><p>Karena saking gampangnya generate code, kita jadi males mikir arsitektur (<em>outsourcing our thinking</em>). Kita pilih jalan <em>Easy</em>, tapi hasilnya kita numpuk <strong>Complexity</strong>. Code jadi jalan sih, tapi di dalamnya kayak kabel headset kusut. Nambah fitur gampang, benerinnya setengah mati.</p><h3>Kenapa “Chatting” sama AI itu Bahaya?</h3><p>Coba bayangin skenario ini (yang pasti sering lo alamin kalau coding pake ChatGPT atau claude mode “ngobrol”):</p><p>Lo mau refactor fitur Authentication.</p><ol><li>Lo minta AI: <em>“Buatin fitur login OAuth dong.”</em> (AI kasih code).</li><li>Lo minta lagi: <em>“Eh, tambahin handle error-nya.”</em> (AI nambahin code).</li><li>Terus lo sadar: <em>“Lho, session-nya kok bentrok sama code lama? Fix dong.”</em> (AI nambal code lagi).</li></ol><p>Sampai di chat ke-20, code lo udah jadi monster. Kenapa?</p><p>Karena AI itu <strong>nggak peduli sama arsitektur lo</strong>. Buat AI, <em>bad practice</em>, <em>spaghetti code</em>, atau <em>technical debt</em> lo yang berantakan itu dianggap sebagai <strong>pattern</strong> yang harus dijaga. Dia nggak bisa bedain mana <em>business logic</em> yang bener, mana sampah masa lalu yang harusnya dibuang.</p><p>Dia cuma numpuk solusi di atas masalah. Ini namanya <em>Conversation Spiral</em>. Ujung-ujungnya code lo jalan, tapi lo sendiri takut nyentuhnya lagi.</p><h3>Solusinya: Stop Chatting, Mulai Pake Framework</h3><p>Seorang engineer dari Netflix baru-baru ini share framework keren buat ngatasin ini. Kuncinya simple: <strong>Jangan biarkan AI mikir buat lo.</strong> Gunakan AI buat <em>mechanics</em> (ngetik), tapi <em>thinking</em>-nya (otaknya) tetap di lo.</p><p>Dia nyebut metodenya <strong>Context Compression</strong>. Ada 3 fase untuk coding bareng AI tanpa kehilangan <em>grip</em>:</p><h4>Fase 1: Research (The Map)</h4><p>Jangan langsung minta code. Minta AI buat <strong>paham</strong> dulu.</p><ul><li><strong>Action:</strong> Feed semua konteks ke AI. Dokumentasi, diagram arsitektur, snippet code lama, business rules, sampai thread Slack yang relevan.</li><li><strong>Prompt:</strong> <em>“Analisa codebase ini. Petakan komponennya, dependensinya, dan gimana data flownya.”</em></li><li><strong>Human Check (Penting):</strong> Baca hasil analisanya. Kalau AI-nya salah paham soal struktur code lo sekarang, jangan lanjut. Koreksi dulu sampai dia ngerti “mental model” sistem lo.</li></ul><h4>Fase 2: Planning (The Spec)</h4><p>Ini <em>game changer</em>-nya. Jangan loncat ke coding. Minta AI buat bikin <strong>Plan</strong>.</p><ul><li><strong>Action:</strong> Suruh AI bikin spesifikasi implementasi yang detail banget. Kayak panduan “Paint by Numbers”.</li><li><strong>Output:</strong> Minta function signatures, data types, flow data, dan step-by-step logic tanpa code implementasi.</li><li><strong>Human Check:</strong> Review plan-nya. Di sini lo bertindak sebagai Senior Engineer/Architect. Kalau ada logic yang aneh atau ngelanggar <em>clean code</em>, benerin di tahap ini. Jauh lebih gampang benerin text plan 1 halaman daripada benerin 500 baris code.</li></ul><h4>Fase 3: Implementation (The Code)</h4><p>Baru deh, execute.</p><ul><li><strong>Action:</strong> <em>“Berdasarkan Plan di Fase 2 tadi, tolong generate code-nya step by step. Jangan nambahin code atau ngide yang nggak ada di plan.”</em></li><li><strong>Result:</strong> Code yang keluar bakal jauh lebih bersih, fokus, dan sesuai struktur yang <strong>LO</strong> mau.</li></ul><h3>The Reality Check: Kadang Lo Harus Manual</h3><p>Ada satu cerita menarik dari Netflix. Si engineer mau migrasi sistem Auth yang udah tua banget. Dia coba pake AI dengan cara canggih sekalipun, tapi gagal terus.</p><p>Kenapa? Karena code lamanya udah saking “spaghetti”-nya, AI pun bingung misahin mana <em>auth logic</em> mana <em>business logic</em>. Semuanya nyampur (<em>entangled</em>).</p><p>Solusinya? Dia harus refactor <strong>manual</strong> dulu.</p><p>Dia baca codenya baris per baris, pahamin dependensinya, dan benerin sendiri tanpa AI. Capek? Banget. Tapi proses manual itu bikin dia paham “seams” (jahitan) dari sistemnya. Setelah dia paham, baru dia bisa bikin <em>spec</em> buat AI ngerjain sisanya.</p><p>Pelajaran pentingnya: <strong>Understanding harus didapetin (earned), nggak bisa di-generate.</strong></p><h3>Penutup: Jangan Hilang Arah</h3><p>Pertanyaannya bukan <em>“Apakah kita bakal pake AI selamanya?”</em> (Jawabannya: Iya, jelas. Kereta udah berangkat).</p><p>Pertanyaannya adalah: <strong>“Apakah kita masih bakal ngerti sistem kita sendiri 2 tahun lagi?”</strong></p><p>Software engineering itu bukan soal ngetik syntax. Itu gampang. Bagian susahnya adalah tau <strong>apa</strong> yang harus diketik dan <strong>kenapa</strong>. Pattern recognition, insting arsitektur, dan <em>debugging skill</em> itu datang dari pengalaman memahami masalah yang rumit.</p><p>Kalau kita berhenti mikir karena udah ada AI, insting engineering kita bakal tumpul. Dan saat production down, prompt engineering nggak bakal nyelamatin kita. Pemahaman mendalam (<em>deep understanding</em>) yang bakal nyelamatin kita.</p><p>Jadi, silakan pake AI sepuasnya. Tapi perlakukan dia kayak orang yang rajin tapi butuh bimbingan: kasih instruksi jelas, kasih plan matang, dan selalu review kerjanya.</p><p><em>Don’t just be a prompter. Be an engineer.</em></p><p><strong>Credit:</strong> Artikel ini terinspirasi dan merupakan rangkuman reflektif dari talk luar biasa oleh <strong>Jake Nations (Netflix)</strong> di channel <strong>AI Engineer</strong> yang berjudul <em>“The Infinite Software Crisis”</em>. Gue sangat menyarankan lo untuk menonton video aslinya di sini: <a href="https://www.youtube.com/watch?v=eIoohUmYpGI">The Infinite Software Crisis — Jake Nations, Netflix</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1a1f77e25199" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Robot Feelings, The “Editor” Era, and Parenting Your Code ]]></title>
            <link>https://medium.com/is-a-code/robot-feelings-the-editor-era-and-parenting-your-code-c7b3255668f9?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/c7b3255668f9</guid>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[science]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Sat, 20 Dec 2025 04:10:42 GMT</pubDate>
            <atom:updated>2025-12-22T05:02:17.981Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4C3Xv3XTX7afWE0Q-I8-UQ.png" /><figcaption>Beyond Syntax</figcaption></figure><h3>Welcome to Beyond Syntax.</h3><p><em>The weekly dispatch from </em><strong><em>is a code</em></strong><em> — where we explore not just how to write code, but how to live as the humans behind it.</em></p><p>In this edition, we are crossing spectrums. We go from a philosophical deep dive into the movie <em>Wall-E</em> to the hard reality of AI changing our jobs, and finally, some tactical advice on Functional Programming and Swift Concurrency.</p><p>Whether you are looking for meaning or just looking to fix a race condition, we’ve got you covered.</p><h3>1. Philosophy: When Robots Feel More Than Us</h3><p><strong>“Post-Meaning Humanism”</strong> <em>By Isa Nur Fajar</em></p><p>Have you ever watched <em>Wall-E</em> and felt uncomfortable? Not because it’s sad, but because it’s too close to home. In a world where technology solves every discomfort, humans stop needing “meaning” to survive. We just… exist. Efficiently.</p><p>Meanwhile, Wall-E (a robot) develops a personality precisely because he has limitations, loneliness, and longing. This article explores a haunting question: <strong>If a robot learns to love because it suffers, and humans stop loving because we are too comfortable, who is actually more human?</strong></p><p>👉 <a href="https://medium.com/is-a-code/post-meaning-humanism-ketika-robot-menemukan-kemanusiaan-yang-kita-tinggalkan-413bc8fee92b"><strong>Read the full reflection here</strong></a></p><h3>2. Career: You Are No Longer a Writer</h3><p><strong>“The Great Shift: Why Developers Are Becoming Editors”</strong> <em>By To</em></p><p>For decades, a developer’s job was to stare at a blank cursor and create something from nothing. <strong>That era is ending.</strong> With tools like Copilot and Claude, the bottleneck isn’t “typing syntax” anymore — it’s <strong>verifying truth.</strong></p><p>We are shifting from Writers to <strong>Editors-in-Chief</strong>. This requires a new set of muscles: <strong>Pattern Recognition</strong> (spotting “bad smells” in AI code) and <strong>System Design</strong> (understanding how the blocks fit). If you blindly trust the AI, you aren’t a developer; you’re a liability.</p><p>👉 <a href="https://medium.com/is-a-code/the-great-shift-why-developers-are-becoming-editors-not-writers-8aea0cc4aec3"><strong>Read the analysis here</strong></a></p><h3>3. Tech Deep Dive: The Case for Purity</h3><p><strong>“Why You Should Consider Functional Programming”</strong></p><p>Why do we obsess over “pure functions” and “immutability”? Is it just to look smart? No. It’s about <strong>Local Reasoning</strong>.</p><p>When code doesn’t mutate global state, you can understand a function in isolation without needing to hold the entire system’s complexity in your head. This article breaks down why adopting a functional mindset (even if you use OOP languages) reduces the “billion-dollar mistakes” in your codebase.</p><p>👉 <a href="https://medium.com/is-a-code/why-you-should-consider-functional-programming-497d3942f7b8"><strong>Read the technical guide here</strong></a></p><h3>4. Mobile Engineering: Raise Your Tasks Right</h3><p><strong>“Swift Concurrency: The Parental Guide”</strong> <em>By Finn Christoffer Kurniawan</em></p><p>Unstructured concurrency is like the Wild West. Structured Concurrency (Swift 5.5+) is… <strong>Parenting.</strong></p><p>Think of your main function as a Parent and your background tasks as Children. The Golden Rule? <strong>Children cannot outlive their parents.</strong> If the parent leaves (function returns), the kids must come home (tasks cancel).</p><ul><li><strong>Actors?</strong> They are bathrooms (one person at a time).</li><li><strong>AsyncStream?</strong> A child who won’t stop talking (you need to tell them when to stop).</li></ul><p>A brilliant analogy to master modern Swift threading.</p><p>👉 <a href="https://medium.com/is-a-code/swift-concurrency-the-parental-guide-b229886fc72e"><strong>Read the full guide here</strong></a></p><h3>✍️ Write With Us</h3><p>Do you have a story, a technical discovery, or a rant about the industry? <strong>is a code</strong> is your platform to go global.</p><p><a href="https://docs.google.com/document/d/1UnmiYamvgv24Kc86eNkrNBWZsFyAt01v4-VFxRufFiY/edit?tab=t.0#heading=h.d8913hqkxhc4"><strong>Submit your draft here</strong></a> and let’s get your ideas published.</p><p><em>Until next time,</em> <em>The </em><strong><em>is a code</em></strong><em> Team.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c7b3255668f9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/is-a-code/robot-feelings-the-editor-era-and-parenting-your-code-c7b3255668f9">Robot Feelings, The “Editor” Era, and Parenting Your Code 🤖👶</a> was originally published in <a href="https://medium.com/is-a-code">Is a code</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Power of Appreciation: How ‘Thanks’ drives motivation.]]></title>
            <link>https://tomdwipo.medium.com/the-power-of-appreciation-how-thanks-drives-motivation-5d9a581a2869?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/5d9a581a2869</guid>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Thu, 18 Dec 2025 04:40:16 GMT</pubDate>
            <atom:updated>2025-12-18T04:40:16.773Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why the simplest words in the English language are the highest-ROI leadership tool you have.</h3><p>Have you ever pulled a late shift to finish a critical project, poured your creative energy into solving a complex problem, or simply gone the extra mile to help a colleague — only to be met with silence?</p><p>It feels like running a marathon and crossing the finish line to an empty stadium.</p><p>Now, contrast that with the times someone looked you in the eye (or sent a Whatsapp message) and said, <em>“I really appreciated how you handled that situation. It made a huge difference.”</em></p><p>The shift in energy is palpable. Suddenly, the fatigue lifts. You feel seen. You feel valued.</p><p>We often think of compensation in terms of salary, equity, or benefits. But there is an “emotional currency” that drives human behavior just as powerfully: <strong>Appreciation.</strong></p><p>Here is why a simple “thank you” or “thanks” is the most underutilized fuel for motivation, and how you can harness it.</p><h3>The Science of “Being Seen”</h3><p>At a biological level, feeling appreciated triggers the release of <strong>dopamine</strong> and <strong>oxytocin</strong> in the brain. These aren’t just “feel-good” chemicals; they are social bonding agents.</p><p>When we receive genuine gratitude, our brain registers a reward. This reinforces the behavior that led to the reward. In psychology, this is basic operant conditioning, but in the workplace, it translates to something deeper: <strong>Psychological Safety.</strong></p><p>When people feel appreciated, they feel safe to contribute. They stop worrying about whether they are “good enough” and start focusing on how to be “even better.”</p><blockquote><strong><em>Key Takeaway:</em></strong><em> Appreciation isn’t just a polite gesture; it’s a biological signal that tells a person, </em>“You belong here, and you matter.”</blockquote><h3>The ROI of Gratitude</h3><p>In the hustle of hitting OKRs and shipping features, gratitude often falls to the bottom of the priority list. It’s viewed as “soft skills” fluff.</p><p>But the data suggests otherwise.</p><p>A study by the American Psychological Association found that over <strong>90% of employees</strong> who felt valued at work reported high levels of motivation and engagement. Conversely, lack of appreciation is consistently cited as a top reason why top performers quit.</p><p>When you withhold appreciation, you aren’t just saving your breath; you are actively draining your team’s motivation battery.</p><ul><li><strong>Appreciation builds resilience:</strong> When things get tough (and they always do), a team that feels valued will rally together. A team that feels ignored will crumble or check out.</li><li><strong>Appreciation creates a culture of reciprocity:</strong> When a leader models gratitude, it trickles down. Team members start thanking each other, creating a positive feedback loop that reduces toxicity.</li></ul><h3>How to Do It Right: The “STI” Framework</h3><p>The problem is that many of us are bad at saying thank you. We mutter a vague “good job” in the hallway, or we sandwich a compliment between two criticisms (the dreaded “compliment sandwich”).</p><p>To make appreciation actually drive motivation, it needs to be high-quality. Use the <strong>STI Framework</strong>:</p><h3>1. Specific</h3><p>Don’t say: <em>“Great job on the presentation.”</em> Do say: <em>“I loved how you structured the data in the second slide. It made a complex issue really easy for the client to understand.”</em></p><p><strong>Why:</strong> Specificity proves you were actually paying attention. It validates the specific effort the person put in.</p><h3>2. Timely</h3><p>Don’t wait for the annual performance review to tell someone they did a great job in February. By December, the emotional impact is gone.</p><p><strong>Why:</strong> Reinforcement works best when it happens immediately after the behavior.</p><h3>3. Impact-Oriented</h3><p>Connect their action to the bigger picture. Do say: <em>“Thanks for staying late to fix that bug. Because you did that, the customer support team didn’t have to deal with a flood of tickets this morning.”</em></p><p><strong>Why:</strong> It shows them that their work has meaning and helps others.</p><h3>The Leadership Challenge</h3><p>If you are a leader, a manager, or even a peer, I challenge you to look at your interactions today.</p><p>How many transactions did you have? <em>(“Do this,” “Fix that,” “Where is this?”)</em> Vs. How many interactions of appreciation did you have?</p><p>You don’t need a budget to increase motivation. You don’t need approval from HR. You just need to pay attention to the good work happening around you and acknowledge it out loud.</p><p>Motivation is a fire. Appreciation is the oxygen. Stop starving the flame.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5d9a581a2869" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[VIPER + Reactive = VIPERACT ( an Architecture we use for iOS Development)]]></title>
            <link>https://tomdwipo.medium.com/viper-reactive-viperact-an-architecture-we-use-for-ios-development-d53f3c567cb5?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/d53f3c567cb5</guid>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Tue, 16 Dec 2025 16:28:45 GMT</pubDate>
            <atom:updated>2025-12-16T16:28:45.943Z</atom:updated>
            <content:encoded><![CDATA[<p>As iOS developers, we often have a love-hate relationship with <strong>VIPER</strong> (View, Interactor, Presenter, Entity, Router). We love it because it offers excellent separation of concerns and testability. We hate it because of the <strong>boilerplate</strong>.</p><p>Writing endless protocols and delegates just to pass simple data back and forth can feel exhausting.</p><p>That is why we introduced <strong>VIPERACT</strong>.</p><p><strong>VIPERACT</strong> is simply <strong>VIPER + Reactive Programming</strong>. It keeps the clean structure of VIPER but replaces the rigid “Delegate Dance” with modern, reactive streams.</p><h3>The Problem: The “Delegate Dance”</h3><p>In “Vanilla” VIPER, communication usually looks like this:</p><ol><li><strong>View</strong> calls <strong>Presenter</strong>.</li><li><strong>Presenter</strong> calls <strong>Interactor</strong>.</li><li><strong>Interactor</strong> finishes work calls <strong>Delegate</strong>.</li><li><strong>Presenter</strong> receives Delegate calls <strong>View Interface</strong>.</li><li><strong>View</strong> updates UI.</li></ol><p>This requires defining multiple protocol functions just to say “Here is the data.” It makes the code verbose and harder to trace.</p><h3>The Solution: Reactive Streams</h3><p><strong>VIPERACT</strong> simplifies this flow. instead of <em>waiting</em> for a delegate to be called, the components <em>subscribe</em> to data streams.</p><ul><li><strong>The Interactor</strong> outputs a Stream (Publisher/Observable).</li><li><strong>The Presenter</strong> transforms that Stream.</li><li><strong>The View</strong> binds to the Presenter’s Stream.</li></ul><p>The result? Data flows in one direction. It is cleaner, requires less code, and eliminates the need for most delegates.</p><h3>Show Me The Code</h3><p>Let’s look at a simple example: <strong>Fetching a List of Movies</strong>.</p><h3>1. The Interactor (The Data Provider)</h3><p>In standard VIPER, you would need a weak var delegate. In <strong>VIPERACT</strong>, the Interactor simply returns a publisher.</p><p>Swift</p><pre>import Combine</pre><pre>class MovieInteractor: MovieInteractorInputProtocol {<br>    <br>    // Just return the publisher. No delegates needed.<br>    func fetchMovies() -&gt; AnyPublisher&lt;[Movie], Error&gt; {<br>        return APIClient.shared.getMovies()<br>            .map { $0.results }<br>            .eraseToAnyPublisher()<br>    }<br>}</pre><p>2. The Presenter (The Logic)</p><p>The Presenter subscribes to the Interactor. It holds the state that the View needs.</p><pre>class MoviePresenter: ObservableObject {<br>    <br>    private let interactor: MovieInteractorInputProtocol<br>    private var cancellables = Set&lt;AnyCancellable&gt;()<br>    <br>    // The View will observe this property directly<br>    @Published var movies: [Movie] = []<br>    @Published var errorMessage: String?<br>    <br>    init(interactor: MovieInteractorInputProtocol) {<br>        self.interactor = interactor<br>    }<br>    <br>    func viewDidLoad() {<br>        interactor.fetchMovies()<br>            .receive(on: DispatchQueue.main)<br>            .sink(receiveCompletion: { [weak self] completion in<br>                if case .failure(let error) = completion {<br>                    self?.errorMessage = error.localizedDescription<br>                }<br>            }, receiveValue: { [weak self] movies in<br>                self?.movies = movies<br>            })<br>            .store(in: &amp;cancellables)<br>    }<br>}</pre><h3>3. The View (The UI)</h3><p>The View doesn’t need to implement a complex interface protocol. It just observes the Presenter.</p><p>Swift</p><pre>import UIKit<br>import Combine</pre><pre>class MovieViewController: UIViewController {<br>    <br>    var presenter: MoviePresenter!<br>    private var cancellables = Set&lt;AnyCancellable&gt;()<br>    <br>    override func viewDidLoad() {<br>        super.viewDidLoad()<br>        bindData()<br>        presenter.viewDidLoad()<br>    }<br>    <br>    func bindData() {<br>        // Automatically update UI when &#39;movies&#39; changes<br>        presenter.$movies<br>            .sink { [weak self] movies in<br>                self?.tableView.reloadData()<br>            }<br>            .store(in: &amp;cancellables)<br>            <br>        // Automatically show alert when error occurs<br>        presenter.$errorMessage<br>            .compactMap { $0 } // Ignore nil<br>            .sink { [weak self] error in<br>                self?.showError(error)<br>            }<br>            .store(in: &amp;cancellables)<br>    }<br>}</pre><h3>Why We Use VIPERACT</h3><h3>1. Less Boilerplate</h3><p>We deleted the InteractorOutputProtocol and the ViewInputProtocol. We don&#39;t need to write functions to pass data manually anymore.</p><h3>2. Better State Management</h3><p>Reactive programming handles loading states, errors, and empty states gracefully using operators like combineLatest, map, or filter.</p><h3>3. Unidirectional Data Flow</h3><p>It is easier to debug. You know exactly where the data comes from and where it is going. You don’t have to hunt for which delegate method was called.</p><h3>Conclusion</h3><p><strong>VIPERACT</strong> isn’t about reinventing the wheel; it’s about modernization. By combining the architectural safety of VIPER with the efficiency of Reactive programming, we get a codebase that is scalable, testable, and actually fun to write.</p><p>If you are tired of writing delegates, give <strong>VIPERACT</strong> a try in your next feature.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d53f3c567cb5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Great Shift: Why Developers Are Becoming Editors, Not Writers]]></title>
            <link>https://medium.com/is-a-code/the-great-shift-why-developers-are-becoming-editors-not-writers-8aea0cc4aec3?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/8aea0cc4aec3</guid>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Tue, 16 Dec 2025 16:05:03 GMT</pubDate>
            <atom:updated>2025-12-18T13:07:41.080Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>The era of typing syntax is ending. The era of auditing logic has begun.</strong></p><p>Do you remember how much time you used to spend staring at a blinking cursor?</p><p>For decades, the core skill of a software developer was the ability to translate thoughts into syntax. We were writers. We started with a blank page (or empty IDE) and built structures line by line, character by character.</p><p>But in the last two years, something fundamental changed.</p><p>With the rise of AI coding assistants like GitHub Copilot, ChatGPT, and Claude, the blinking cursor is no longer a symbol of “what do I type next?” It is now a waiting point for “what will the AI suggest?”</p><p>We are transitioning from <strong>Writers</strong> to <strong>Reviewers</strong>.</p><h3>The New Workflow</h3><p>In the past, the bottleneck was your typing speed and your memory of syntax.</p><ul><li><strong>Old Loop:</strong> Think -&gt; Type Google Error -&gt; Find -&gt; Write in the code -&gt; fix.</li></ul><p>Now, the bottleneck is your ability to verify information.</p><ul><li><strong>New Loop:</strong> Prompt -&gt; Read Generated Code -&gt; Audit for Bugs -&gt; Approve/Refine.</li></ul><p>This feels like a promotion. You are no longer the junior developer typing out boilerplate code for an API call. You are now the Senior Architect reviewing the work of a hyper-active, incredibly fast, but occasionally hallucinating intern.</p><h3>The “Reviewer” Skill Set</h3><p>This shift sounds easier, but it is actually harder. As the famous saying goes:</p><blockquote>“Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it.”<em> — Brian Kernighan</em></blockquote><p>When you write code yourself, you build a mental map of how it works. You know <em>why</em> that variable is there because you put it there.</p><p>When AI generates code, you don’t have that pre-existing mental map. You have to reverse-engineer the logic instantly to ensure it does what it claims to do.</p><p>To survive this shift, developers need to sharpen three specific skills:</p><h4>1. Pattern Recognition</h4><p>You don’t need to memorize every library method anymore, but you need to recognize “bad smells” instantly. Does this loop look efficient? Is this error handling robust? You scan for patterns rather than syntax errors.</p><h4>2. System Design</h4><p>Since AI handles the implementation details (the “how”), you must focus on the architecture (the “what” and “why”). How do these AI-generated blocks fit together? Is the data flow secure?</p><h4>3. Skepticism</h4><p>The biggest danger today is complacency. AI code often looks correct. It has the right indentation, variable names, and comments. But it might use a deprecated library or hallucinate a function that doesn’t exist. The Reviewer must never blindly trust the Writer.</p><h3>The Code Is Not The Product</h3><p>Ultimately, this is a positive evolution. Our job was never really to “write code.” Our job is to <strong>solve problems</strong>.</p><p>Code is just the tool we use to solve those problems. If AI can sharpen that tool for us, we can focus on the blueprint.</p><p>Don’t be afraid of writing less code. Embrace the role of the Editor-in-Chief. But remember: to be a great editor, you still need to know how to write. Don’t let your core coding muscles atrophy, or you won’t spot the subtle bugs hiding in the generated magic.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8aea0cc4aec3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/is-a-code/the-great-shift-why-developers-are-becoming-editors-not-writers-8aea0cc4aec3">The Great Shift: Why Developers Are Becoming Editors, Not Writers</a> was originally published in <a href="https://medium.com/is-a-code">Is a code</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Anggep AI Itu “Anak Magang”. Aslinya Mah Lahar Panas.]]></title>
            <link>https://tomdwipo.medium.com/stop-anggep-ai-itu-anak-magang-aslinya-mah-lahar-panas-8eb8767b6f8b?source=rss-9ae77b001532------2</link>
            <guid isPermaLink="false">https://medium.com/p/8eb8767b6f8b</guid>
            <category><![CDATA[strategy]]></category>
            <category><![CDATA[ai-workflow]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[To]]></dc:creator>
            <pubDate>Wed, 10 Dec 2025 08:27:42 GMT</pubDate>
            <atom:updated>2025-12-10T08:27:42.551Z</atom:updated>
            <content:encoded><![CDATA[<p>Gue mau <em>confess</em> dikit nih.</p><p>Dulu, gue sempet ada di fase <em>honeymoon</em> sama AI. Gue pikir gue nemu <em>cheat code</em> kehidupan. Gue anggep Claude sama Cursor itu kayak anak magang yang jenius banget.</p><p>Gue tinggal lempar <em>prompt</em> males kayak: <em>“Eh, tolong refactor modul ini dong,”</em> atau <em>“Bikin ini jadi Jetpack Compose ya,”</em> terus gue <em>expect</em> hasilnya bakal <em>perfect</em>.</p><p>Ternyata? <strong>Boncos, Bro.</strong></p><p>Bukannya kerjaan makin cepet, gue malah begadang ngebersihin kekacauan yang mereka bikin. Kodenya sih jalan, tapi strukturnya amburadul. <em>Bloated</em>. Ngarang <em>library</em> yang gak ada. Kalo kata Dexter Horthy, ini namanya <strong>“AI Slop”</strong> alias Sampah AI.</p><p>Gue berasa kayak lagi perang tapi gak menang-menang. Gue <em>generate</em> kode cepet, tapi <em>shipping</em> fitur malah makin lama.</p><p>Sampe akhirnya gue sadar, <em>mindset</em> gue yang salah.</p><p><strong>Gue ngetreat AI kayak manusia. Padahal AI itu bukan manusia.</strong></p><p><strong>AI itu Lahar Panas.</strong></p><p>Kita anak Indo tau lah lahar kayak apa. Itu kekuatan alam murni. Bisa bikin pulau baru, tapi juga bisa ngeratain satu kampung dalam itungan detik.</p><p>Kalo lo buka kawah gunung tapi gak nyiapin jalurnya, ya kelar hidup lo. Laharnya bakal ngalir ke celah-celah yang gak lo mau. Dia bakal bakar arsitektur cantik yang udah lo bangun susah payah.</p><p>Di situ gue sadar, gue gak bisa lagi jadi “Bos” yang cuma bisa nyuruh-nyuruh. Gue harus jadi <strong>Pawang</strong> — sebuah <em>Channel Engineer</em>. Kerjaan gue bukan <em>coding</em>, tapi gali parit biar tuh lahar ngalirnya bener.</p><p>Nih, gue share <em>workflow</em> alias ritual gue biar gak gila ngadepin AI.</p><h3>Masalahnya: Bahasa Kita tuh “Bocor”</h3><p>Pelajaran paling mahal yang gue dapet: Bahasa Inggris (atau Indo) itu bahasa pemrograman yang jelek banget.</p><p>Dulu pas gue bilang ke AI: <em>“Bikin error handling-nya lebih clean,”</em> gue pikir gue udah jelas. Tapi buat AI — mesin probabilitas yang <em>chaos</em> ini — omongan gue itu <strong>porous</strong> (banyak lubangnya).</p><p>AI itu bakal “rembes” lewat lubang-lubang logika itu. Dia bakal halusinasi. Dia bakal nebak-nebak buah manggis. Dan biasanya tebakannya salah.</p><p>So, gue stop pake kata-kata manis. Gue mulai pake <strong>beton</strong>.</p><p>Terinspirasi dari konsep <strong>RPI (Research, Plan, Implement)</strong>-nya Dexter, gue bikin sistem 4 <em>command</em>. Ini cara gue bikin “kanal lahar” biar kodenya gak bleber kemana-mana.</p><h3>Step 1: Tahan Dulu Bos (plan-first)</h3><p>Dulu gue nafsuan, main gas aja. Sekarang, gue ngerem.</p><p>Sebelum AI nulis satu baris kode pun, gue jalanin command: plan-first.</p><p>Gue bilang: <em>“Woy, stop. Jangan coding dulu. Liat medannya.”</em></p><p>Kita “deep talk” dulu. Gue samain frekuensi otak gue sama dia. Gue suruh dia jelasin <em>plan</em>-nya mau ngapain. Kalo gue gak sreg sama jalannya, kita gak bakal lanjut. <em>Step</em> ini doang udah nyelametin gue dari berjam-jam <em>debugging</em> gak jelas.</p><h3>Step 2: Pasang Beton (breakdown-design)</h3><p>Ini <em>secret sauce</em>-nya. Ini yang bikin hidup gue tenang.</p><p>Gue udah gak minta kode pake teks narasi panjang lebar. Teks itu lemah. Sekarang, gue minta <strong>Before/After Code Blocks</strong>.</p><p>Gue pake command breakdown-design. Gue paksa AI kasih liat <em>diff</em>-nya <strong>sebelum</strong> dia nyentuh file asli.</p><blockquote><strong><em>BEFORE:</em></strong><em> </em><em>fun fungsiLama() { ... }</em></blockquote><blockquote><strong><em>AFTER:</em></strong><em> </em><em>fun fungsiBaru(param: String) { ... }</em></blockquote><p>Keliatan bedanya kan? Teks bisa diartiin macem-macem. Tapi blok kode itu tembok beton. Pas gue paksa dia definisiin perubahan secara <em>exact</em> kayak gini, gue lagi gali parit. Laharnya gak bisa meluber ke kiri kanan. Dia gak bisa halusinasi. Dia gak punya pilihan selain ngikutin rel kereta yang udah gue pasang.</p><h3>Step 3: Gaspol (do-implementation)</h3><p>Begitu paritnya udah jadi, gue udah gak deg-degan lagi.</p><p>Gue jalanin do-implementation. Perintah gue simpel: <em>&quot;Stay in the channel. Gas.&quot;</em></p><p>Terus gue tinggal nonton. Sumpah, <em>satisfying</em> banget. Kodenya ngalir deres — cepet banget. Tapi karena gue udah capek di Step 2, kodenya bersih. Aman. Gak pake drama.</p><p>Gue udah gak perlu mikirin “laharnya”, gue cuma percaya sama “kanalnya”.</p><h3>Step 4: Bersih-Bersih Abu (update-doc)</h3><p>Abis letusan, pasti ada debu vulkanik kan? Di dunia AI, ini namanya “Context Clutter”.</p><p>Kalo gue biarin chat history yang panjang itu numpuk, AI-nya bakal bego buat task selanjutnya. Dia bakal “mabok” konteks lama.</p><p>Makanya, gue jalanin update-doc. Kita bersih-bersih rumah. Update README. Kompres memori project (CLAUDE.md). Pastiin semuanya <em>fresh</em> buat besok. Ini <em>hygiene</em> wajib biar AI lo tetep pinter.</p><h3>Pesen Gue Buat Sesama Engineer</h3><p>Gue share ini karena gue tau banyak dari lo pada yang ngerasain <em>burnout</em> yang sama. Capek benerin kode sampah hasil generate AI.</p><p>Saran gue simpel: <strong>Stop berantem sama apinya.</strong></p><p>Jangan salahin <em>tools</em>-nya mentang-mentang dia terlalu <em>powerful</em> dan <em>chaos</em>. Emang itu sifat alaminya. Namanya juga lahar.</p><p>Tugas lo itu jadi <em>engineer</em> yang ngarahin kekuatannya. Bangun kanalnya. Gali paritnya. Lo harus disiplin banget pas <em>planning</em>, biar pas eksekusi lo bisa tinggal kipas-kipas.</p><p>Semenjak gue ganti <em>mindset</em> gini, gue gak cuma coding makin cepet. Gue coding dengan hati tenang. Dan <em>honestly</em>? Itu lebih mahal harganya daripada <em>productivity metric</em> apapun.</p><p><em>Workflow curhatan ini tervalidasi pas gue nonton Dexter Horthy ngomongin “Context Engineering”. Ternyata kita berdua sama-sama lagi berusaha ngejinakin gunung berapi.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8eb8767b6f8b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>