<?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 Ricky Wong on Medium]]></title>
        <description><![CDATA[Stories by Ricky Wong on Medium]]></description>
        <link>https://medium.com/@rickywck?source=rss-22cfc42db569------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*wIqKirxu4ha5vlVIleUDuw.jpeg</url>
            <title>Stories by Ricky Wong on Medium</title>
            <link>https://medium.com/@rickywck?source=rss-22cfc42db569------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 22:53:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rickywck/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[Will increasing smarter AI Coding Agents replace software developers?]]></title>
            <link>https://medium.com/@rickywck/will-increasing-smarter-ai-coding-agents-replace-software-developers-c201bb7916eb?source=rss-22cfc42db569------2</link>
            <guid isPermaLink="false">https://medium.com/p/c201bb7916eb</guid>
            <dc:creator><![CDATA[Ricky Wong]]></dc:creator>
            <pubDate>Sun, 20 Apr 2025 20:07:19 GMT</pubDate>
            <atom:updated>2025-04-20T20:07:19.369Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VneVJNhpGK3j6igSKwcVNA.png" /></figure><p>“Vibe Coding” is a recent hot topic and in some of the discussions, it brought back to the question which was debated for years — will AI replace Software Developers? As I worked in the industry for so long around the globe with mostly Enterprise clients and some of the leading industry players, I always want to share my point of view (even though I can’t disclose my clients / employer name :)).</p><p>Simple answer is: Yes and No …. well, <strong>there are so many types of software developer so it is obviously AI can’t replace all of them, but I am pretty confident it will replace some </strong>… actually already happening.</p><p>Given my background, I would like to confine this question more to typical Enterprise environments. Usually, these companies have: (1) employed hundreds of developers themselves working in their home location, (2) employed large software vendors which have both onshore and offshore developers (very often offshore has much higher percentage), (3) some of them even have their own offshore development centre with hundreds of developers working in countries such as India, Philippines, China (less common these days) …, etc.</p><p>Except for very few, software development is not core competence of most Enterprises, so one of their primary objectives is to lower the cost of building and maintaining software.</p><p>I don’t want this to be a very long article … just cut to the chase, if you look at typical resources mix of software development teams for these Enterprises, no matter whether they have their own teams or outsource major portion of software development to other companies, in last 10+ years the common formula is to deploy more resources offshore and at the same time, these offshore teams had kept pushing down their ‘pyramid’, i.e. employ more junior (and cheaper) developers with less experienced, in order to lower cost to serve.</p><p>My opinion is, with recent development of Generative AI, there will have significant impact on how these teams will be structured — <strong>number of junior developers whose can mainly do routine coding and maintenance work will likely be reduced substantially </strong>as what they can do can be fully replaced by AI … at in some cases, AI may do the job better than these junior and less experienced developers.</p><p>Can these junior and less experiended developers be upskilled to do more value added working? I will leave this to a separated article. However, if you have experiences working in such an environment for long enough, you would probably come to similar conclusion … (<strong><em>Economist</em></strong> published an article earlier this year which said AI will divide the best from the rest … which I agree, especially in software development field).</p><p>To conclude, I personally believe there will have significant impact on those companies relying on the “traditional” software factory model for their business, e.g. most ‘IPP’ (Indian pure play) such as Infosys, TCS, …, etc., as the demand for those junior &amp; cheap software developers will be reduced significantly in future, and same will be applied for Enterprise developers (but some companies have already outsourced a lot of development to vendors, so the impact will primarily on their vendors).</p><p>p.s. I have been working in the industry for so long and was kept looking for the best forum to publish my point of view / experiences … still exploring which ones are more suitable to me. :)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c201bb7916eb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A few comments about Apple Intelligence (which most people haven’t talked about)]]></title>
            <link>https://medium.com/@rickywck/a-few-comments-about-apple-intelligence-which-most-people-havent-talked-about-99a1176eee40?source=rss-22cfc42db569------2</link>
            <guid isPermaLink="false">https://medium.com/p/99a1176eee40</guid>
            <category><![CDATA[iphone]]></category>
            <category><![CDATA[generative-ai-use-cases]]></category>
            <category><![CDATA[apple-intelligence]]></category>
            <category><![CDATA[wwdc-2024]]></category>
            <category><![CDATA[ios]]></category>
            <dc:creator><![CDATA[Ricky Wong]]></dc:creator>
            <pubDate>Wed, 12 Jun 2024 20:50:37 GMT</pubDate>
            <atom:updated>2024-06-12T20:57:28.136Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*Rm1SdB4aqllNI4NusZ4oEg.jpeg" /></figure><p>Apple has announced Apple Intelligence in WWDC 2024 this week. There are a lot of articles written as well as videos on YouTube talking about it. However most of them focus more on its ‘features’ or capabilities, but almost none of them have discussed about Apple’s thinking or strategy from a deeper angle. I would keep this short and just a highlight a few key perspectives from my opinion.</p><p><strong><em>#1 Increasing adoption of “Small Language Model”</em></strong></p><p>While Apple Intelligence also has a cloud component, the emphasis is on device language model (according to some other authors it should be based on a 3B parameters quantized model). So, while the industry is still kept releaseing larger and more powerful language model on one hand, this is a perfect example of development on another direction. The fact is, having larger and more powerful language models does not necessarily help speed up of popularization of Generative AI because of cost, latency, privacy, … issues.</p><p><strong><em>#2 Having context for LLM in a device</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EqFJ2i3Az3VoyQMGny5jcg.png" /><figcaption>Source: <a href="https://snorkel.ai/which-is-better-retrieval-augmentation-rag-or-fine-tuning-both/">https://snorkel.ai/which-is-better-retrieval-augmentation-rag-or-fine-tuning-both/</a></figcaption></figure><p>Apple Intelligence is basically a RAG (Resource Augmented Generation) application running inside Apple devices using “small” and optimized models. Instead of relying on vector database, knowledge graphs on Cloud, Apple basically tries to use data on the device as context to fit into the model, and the context information is not limited to data from a single app, but everything in the device across multiple apps (or even information maintained at OS level). While no one has done this so far, but its overall architecture and concept is very similar to RAG based applications.</p><p><strong><em>#3 Focus on overall User Experience instead</em></strong></p><p>Apple is not the 1st vendor to provide AI features on devices — Google, Samsung … even some China vendors such as Huawei have done this. However, none of them took similar approach as Apple. Instead, they focused on the capability and power of the on-device AI model, and integrate the model with individual app in order to show case how good these on-device models can be. However, Apple took a very different direction — instead of saying how good their model is (in fact, it is probably not the area they shine), they are trying to build applications empowered with AI which can improve overall user experience.</p><p><strong><em>#4 There is still a big “if” …</em></strong></p><p>Which is how non-Apple developers adopt Apple’s AI strategy. While some of the demos from WWDC is cool, or at least appealing to end users, they require users running Apple apps. If a user is using Facebook, WhatsApp, Gmail, Outlook, Microosft Office .. unless these app developers update their apps to integrate with Apple AI, Apple Intelligence will be useless even the device is upgraded to latest version of iOS. Whether these app developers will do this … well, it is hard to say. In fact, if Apple’s strategy is success, Google may do the same for their own apps as higher priority than integrating their apps with Apple AI.</p><p>In conclusion, Apple has cleverly executed its AI strategy to leverage its strength (ecosystem adoption, user experience, etc.) instead of racing with others on how powerful their LLM is. Whether Apple’s AI strategy will be successful or not, will need to wait and see, however, there is no doubt AI will be immersed in people’s experience more deeply in coming years, i.e. people will use AI without necessarily knowing they are using AI explicitly.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99a1176eee40" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Generative AI is an opportunity to software developer, not a threat]]></title>
            <link>https://medium.com/@rickywck/generative-ai-is-an-opportunity-to-software-developer-not-a-threat-53ca1f9f89fb?source=rss-22cfc42db569------2</link>
            <guid isPermaLink="false">https://medium.com/p/53ca1f9f89fb</guid>
            <dc:creator><![CDATA[Ricky Wong]]></dc:creator>
            <pubDate>Wed, 17 May 2023 20:03:20 GMT</pubDate>
            <atom:updated>2023-05-17T20:03:20.206Z</atom:updated>
            <content:encoded><![CDATA[<p>I didn’t get used to write and publish articles online but I do have tons of ideas to share, based on my 20+ years experiences working for many different types of projects and serving some very large organizations.</p><p>There is no doubt Generative AI is one of the hottest IT topics or “buzz word” these days. Almost everyone in the planet are talking about Generative AI. As someone who grew up from Software Developer since (or actually before) I started my career in IT, there is one question particularly interests me — does (or will) Generative AI replace software developers?</p><p>If we can only provide a “binary” answer, then it has to be “no”. However, if we look at the question “deeper”, I indeed see opportunity, or put it in other words, brighter future to software developers with Generative AI.</p><p>There are two ‘camps’ for software industry — package software vs custom software. Package software refers to those solutions from SAP, Workday, Salesforce, … and many other domain specific solutions such as core banking software, insurance software, etc. Custom software refers to applications which are developed from scratch.</p><p>There is a term called “Buy vs Build” — many organizations usually prefer to adopt software package solutions instead of custom build. This makes sense since the core business of many organizations is not to build software, so it is usually more economically (and quicker) to adopt package solution.</p><p>Imagine if we can build custom software 2–3 times faster (or less effort), and at the same time, the application can fit better requirements from business, and without paying software license cost … will that generate more opportunities for custom software developers?</p><p>Of course, a comprehensive evaluation of buy vs build involves a lot more factors (e.g. some companies pick package software as they want to leverage best practice functional design from those packages) which is not the primary focus of this article. However, if the effort savings in writing software with Generative AI is “real”, then it will definitely have impact on those decisions.</p><p>For those whose have tried tools like GitHub Copilot, AWS Codewhisperer, … you should know those tools can’t replace an experienced developer, it can improve developers’ productivity, but they just can’t produce a fully functional software on its own … and I would say we are still very far from that. Someone may argue I can write a fully functional Tetris game in Javascript using ChatGPT in 15 minutes … that’s simply because there is a fully functional Tetris program written in Javascript in ChatGPT’s training dataset. However, ChatGPT can’t write a banking system for you …</p><p>The key is, how much productivity those tools can really bring to software development — some industry “leaders” are talking about reduction of 50% effort, which in my opinion, could be achievable in certain cases. The second key is, whether developers (could be companies helping clients to develop software, or developers in IT departments) are willing to “commit” on such productivity improvement.</p><p>To conclude, let’s imagine Generative AI is indeed a tools to help software developers to write software significantly faster … if it is really the case, it should be an opportunity to boost demand instead of threat, right?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53ca1f9f89fb" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>