<?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 Griffine Ombogo on Medium]]></title>
        <description><![CDATA[Stories by Griffine Ombogo on Medium]]></description>
        <link>https://medium.com/@griffine_ombogo?source=rss-d0d0b163999f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*MkEwDe3S3_BfAfNLljqwWA.jpeg</url>
            <title>Stories by Griffine Ombogo on Medium</title>
            <link>https://medium.com/@griffine_ombogo?source=rss-d0d0b163999f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 19:21:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@griffine_ombogo/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[Cybersecurity Without the Enterprise Budget: How SMBs Can Build Stronger Defences Against Modern…]]></title>
            <link>https://medium.com/@griffine_ombogo/cybersecurity-without-the-enterprise-budget-how-smbs-can-build-stronger-defences-against-modern-48bbb04cb19e?source=rss-d0d0b163999f------2</link>
            <guid isPermaLink="false">https://medium.com/p/48bbb04cb19e</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[information-security]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[small-business]]></category>
            <dc:creator><![CDATA[Griffine Ombogo]]></dc:creator>
            <pubDate>Mon, 13 Jul 2026 08:22:12 GMT</pubDate>
            <atom:updated>2026-07-13T08:22:12.856Z</atom:updated>
            <content:encoded><![CDATA[<h3>Cybersecurity Without the Enterprise Budget: How SMBs Can Build Stronger Defences Against Modern Threats</h3><figure><img alt="Cybersecurity image" src="https://cdn-images-1.medium.com/max/555/1*JI-YpaE1H-xsrpT5ThdDnw.jpeg" /></figure><p>Small and medium-sized businesses (SMBs) have become some of the most attractive targets for cybercriminals. They store valuable business and customer data, often act as suppliers within larger business ecosystems, and typically lack the extensive cybersecurity resources available to large enterprises. This combination creates an ideal opportunity for attackers.</p><p>Today, SMBs face three persistent cybersecurity challenges: 1. The rapid rise of ransomware, 2. Increasingly sophisticated phishing attacks, and 3. An ongoing shortage of cybersecurity professionals. While these threats can appear overwhelming, protecting an organisation does not necessarily require enterprise-level spending.</p><p>The key lies in simplifying security operations, improving visibility across systems, and adopting solutions that existing IT teams can realistically manage.</p><h3>The Growing Threat of Commoditised Ransomware</h3><p>Ransomware has evolved dramatically over the past few years. What was once carried out primarily by highly organised cybercriminal groups has become widely accessible through <strong>Ransomware-as-a-Service (RaaS)</strong>. These subscription-based criminal services allow attackers with limited technical expertise to purchase ready-made ransomware kits and launch attacks against organisations of virtually any size.</p><p>For SMBs, this shift has significantly increased the threat landscape.</p><p>Modern ransomware groups are also becoming increasingly strategic. Rather than issuing arbitrary ransom demands, they often calculate amounts based on what each organisation is most likely able to pay. According to the <strong>World Economic Forum</strong>, approximately half of organisations worldwide now identify ransomware as their most significant cyber risk.</p><p>Responding effectively requires multiple layers of protection instead of relying on a single security product.</p><p>A resilient ransomware strategy should include:</p><ul><li>Machine learning-driven anti-ransomware protection capable of blocking known threats automatically.</li><li>AI-powered behavioural analytics that identify suspicious activities traditional signature-based detection may overlook.</li><li>Automated endpoint isolation to prevent malware from spreading throughout the network.</li><li>Centralised alert aggregation that helps security teams investigate incidents efficiently.</li><li>Regular offline backups to support rapid recovery.</li><li>Continuous employee awareness training to reduce human error.</li></ul><p>Together, these measures transform ransomware from a potentially devastating event into a manageable operational risk.</p><h3>Phishing Remains the Human Weak Point</h3><p>Despite advances in cybersecurity technology, phishing continues to be one of the most successful methods attackers use to gain initial access to organisations.</p><p>Unlike many cyber threats, phishing exploits human judgement rather than technical vulnerabilities. Today’s phishing campaigns often feature highly convincing emails, trusted sender identities, cloned websites, and increasingly, AI-generated messages that can be personalised at scale.</p><p>According to Kaspersky’s <strong>Anatomy of a Cyber World: 2026 Security Services Global Report</strong>, user execution and phishing techniques remain among the top three cyber threats globally. The findings reinforce a long-standing reality: People continue to represent one of the weakest links in organisational security.</p><p>For many SMBs, building a dedicated security awareness programme comparable to those found in large enterprises is simply unrealistic. However, an effective defence can still be achieved by combining three complementary approaches.</p><h3>1. Strengthen Business Processes</h3><p>Critical financial transactions should require multi-person approval, while access to sensitive systems and data should follow strict governance policies. These controls help limit the damage if an employee unknowingly falls victim to a phishing attempt.</p><h3>2. Invest in Continuous Employee Training</h3><p>Cybersecurity awareness should not be treated as an annual compliance exercise. Ongoing education, reinforced through regular simulations and automatic retraining for employees who demonstrate risky behaviour, helps convert mistakes into valuable learning opportunities.</p><h3>3. Deploy Protective Technology</h3><p>Modern email security solutions that scan emails, attachments and embedded links in real time provide an important safety net. Behaviour-based controls that continue monitoring activity even after a user clicks a malicious link add another critical layer of protection.</p><p>None of these measures alone is sufficient. Together, they substantially reduce both the likelihood of successful phishing attacks and the damage they can cause.</p><h3>The Cybersecurity Skills Shortage</h3><p>Technology is only part of the cybersecurity equation. Skilled people remain equally important, yet qualified cybersecurity professionals are increasingly difficult to find.</p><p>According to Kaspersky research, approximately three-quarters of organisations worldwide consider the cybersecurity skills shortage a serious concern.</p><p>This challenge affects SMBs particularly hard.</p><p>Few smaller organisations have the budget to recruit dedicated security specialists, leaving general IT personnel responsible for defending against sophisticated cyber threats despite often lacking specialised security training.</p><p>This creates a dangerous capability gap.</p><p>Advanced cybersecurity certifications may be too specialised for general IT administrators, while basic cybersecurity awareness programmes rarely provide sufficient expertise to investigate incidents, respond to attacks or manage modern Endpoint Detection and Response (EDR) and Extended Detection and Response (XDR) platforms.</p><p>As a result, attackers can exploit weaknesses that a trained security team might otherwise detect.</p><p>Rather than attempting to hire scarce cybersecurity talent, many SMBs can achieve better long-term results by investing in their existing IT staff.</p><p>Practical upskilling should focus on:</p><ul><li>Incident response fundamentals.</li><li>Secure cloud configuration.</li><li>Effective use of EDR and XDR platforms.</li><li>Security alert recognition and triage.</li><li>Basic threat investigation techniques.</li></ul><p>In addition, clearly defining cybersecurity responsibilities within IT roles ensures these skills continue to develop over time. Investing in professional development also improves employee engagement and retention, helping reduce the turnover that often worsens the skills shortage.</p><h3>Reducing Complexity Builds Cyber Resilience</h3><p>Although ransomware, phishing and workforce shortages appear to be separate challenges, they share a common underlying problem: increasing complexity.</p><p>Many SMBs are making genuine efforts to improve cybersecurity, yet the threat landscape continues evolving faster than their teams and technology can comfortably manage.</p><p>The instinctive response is often to purchase additional security products. Unfortunately, this frequently creates more problems than it solves.</p><p>Multiple disconnected tools generate larger volumes of alerts, increase integration complexity, and introduce visibility gaps that attackers can exploit.</p><p>A more sustainable strategy focuses on consolidation.</p><p>Bringing prevention, detection, response and user awareness into a unified security platform enables smaller IT teams to manage security more effectively while reducing operational overhead.</p><p>Solutions such as <strong>Kaspersky Next Optimum</strong> combine real-time protection with endpoint detection and response (EDR), extended detection and response (XDR), threat visibility, investigation capabilities and automated response functions designed specifically for organisations operating with lean security teams.</p><p>For businesses lacking sufficient in-house expertise, a managed detection and response service such as <strong>Managed Extended Detection and Response (MXDR)</strong> can provide continuous expert monitoring and incident response without requiring a dedicated internal security operations centre.</p><h3>Building Stronger Security Starts with Simplicity</h3><p>Cyber resilience is not solely determined by the size of an organisation’s security budget. It depends on how effectively security technologies, business processes and people work together.</p><p>By reducing unnecessary complexity, SMBs can detect threats earlier, respond more quickly, minimise operational disruption and shift from reacting to incidents toward proactively managing cyber risk.</p><p>With the right combination of practical security practices, ongoing staff development and appropriately scaled security solutions, small and medium-sized businesses can significantly strengthen their cybersecurity posture while building long-term resilience against an increasingly sophisticated threat landscape.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=48bbb04cb19e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Google’s FunctionGemma: Tiny AI, Real Actions, Fully Offline]]></title>
            <link>https://medium.com/@griffine_ombogo/googles-functiongemma-tiny-ai-real-actions-fully-offline-2824dee4bac9?source=rss-d0d0b163999f------2</link>
            <guid isPermaLink="false">https://medium.com/p/2824dee4bac9</guid>
            <category><![CDATA[functiongemma]]></category>
            <category><![CDATA[google-assistant]]></category>
            <category><![CDATA[virtual-assistant]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Griffine Ombogo]]></dc:creator>
            <pubDate>Mon, 22 Dec 2025 22:34:23 GMT</pubDate>
            <atom:updated>2025-12-22T22:34:23.545Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5Eht2pB6qjc66mxC" /><figcaption>Photo by <a href="https://unsplash.com/@user_pascal?utm_source=medium&amp;utm_medium=referral">User_Pascal</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>AI assistants today are powerful, but they come with trade-offs: privacy concerns, internet dependency, lag, and battery drain. Google’s <strong>FunctionGemma</strong>, released on December 18, 2025, quietly challenges that model by bringing real AI actions directly onto your device; no cloud required.</p><p>FunctionGemma is a compact, purpose-built AI model designed to run entirely on-device. It does not send your data to servers, works without internet connectivity, and responds almost instantly. This marks a meaningful shift in how AI assistants are built.</p><p><strong>What Is FunctionGemma</strong></p><p>FunctionGemma is a specialized version of Google’s Gemma 3 model with 270 million parameters: tiny model compared to cloud models with billions and trillions of parameters. Instead of answering questions, it focuses on <em>function calling</em>: turning natural language into real actions on your phone.</p><p>Key features:</p><ol><li><em>Fully offline and private</em></li><li><em>Runs locally on mobile devices</em></li><li><em>Small footprint (approx. 288–550 MB RAM)</em></li><li><em>Built for speed and efficiency</em></li></ol><p>Example, <em>“Add Griffine to my contacts with number 079194…,”</em> FunctionGemma converts this request into a structured system action instantly, without internet access.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L02XVSBtshK-wrfoJgBQKw.png" /><figcaption>i. Flow Diagram</figcaption></figure><p><strong>Why This Matters</strong></p><p>Most AI assistants like Siri and Google Assistant still rely on cloud processing, <br> you speak(input data) → data is sent to servers → response comes back.</p><p>This creates latency, raises privacy risks, drains battery, and fails when you’re offline.</p><p>FunctionGemma changes this approach:</p><ol><li><em>Responds in as little as 0.3 seconds</em></li></ol><p><em>2. Minimal battery usage</em></p><p><em>3. 100% local processing</em></p><p>Google calls this a “traffic controller” model: simple tasks are handled on-device, while complex requests can optionally route to the cloud when available.</p><p><strong>What Can It Do?</strong></p><p>Current demos show FunctionGemma handling tasks like:</p><ol><li><em>Setting alarms and reminders</em></li><li><em>Creating calendar events</em></li><li><em>Controlling smart home devices</em></li><li><em>Logging health or activity data</em></li><li><em>Navigation using local resources</em></li></ol><p>All actions happen locally, fast and private.</p><p><strong>Limitations (For Now)</strong></p><p>FunctionGemma is not perfect:</p><ol><li><em>Struggles with multi-step or vague commands</em></li><li><em>Out-of-the-box accuracy is ~58%, rising to ~85% with fine-tuning</em></li><li><em>Best results require task-specific training</em></li></ol><p><strong>The Bigger Picture</strong></p><p>FunctionGemma challenges the idea that AI must always be massive and cloud-based. It proves that small, specialized models can outperform larger ones for specific tasks while being faster, more private, and more accessible.</p><p>This is not just a smaller model. <strong>It’s a signal that the future of AI is moving from data centers to pockets.</strong></p><p>FunctionGemma is open-source and ready for developers who want to experiment with private and offline AI. To download the model and run it locally, and explore fine-tuning, use the link below.</p><p><a href="https://ollama.com/library/functiongemma">https://ollama.com/library/functiongemma</a></p><p>This is your gateway to building the next generation of on-device AI assistants.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2824dee4bac9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Next Frontier for AI: Why Data Centers in Space Are Becoming Inevitable]]></title>
            <link>https://medium.com/@griffine_ombogo/the-next-frontier-for-ai-why-data-centers-in-space-are-becoming-inevitable-a3bd809a9a42?source=rss-d0d0b163999f------2</link>
            <guid isPermaLink="false">https://medium.com/p/a3bd809a9a42</guid>
            <category><![CDATA[environmental-justice]]></category>
            <category><![CDATA[data-center]]></category>
            <category><![CDATA[articial-intelligence]]></category>
            <dc:creator><![CDATA[Griffine Ombogo]]></dc:creator>
            <pubDate>Fri, 12 Dec 2025 11:11:08 GMT</pubDate>
            <atom:updated>2025-12-12T11:11:08.434Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*hUV98GdYGG4fSNnn3MnJFA.jpeg" /></figure><p>Where does your data actually live? Every message you send, every video you stream, every search you make, none of it floats in an invisible “cloud.” It all sits inside massive physical buildings packed with servers, humming away day and night. These buildings are called data centers, and they are the backbone of the modern digital world.</p><p>Data centers store, process, and distribute information across continents in milliseconds. Their importance has grown with every technological shift: from mobile phones, to streaming platforms, to today’s AI systems. But this growth comes with a cost.</p><p><strong>The Hidden Problem: Earth Is Running Out of Power and Water</strong></p><p>Data centers demand astonishing amounts of energy and cooling. By 2023, global data centers consumed roughly 1 - 1.5% of all electricity used worldwide, a figure projected to climb sharply as AI adoption accelerates. Training advanced AI models requires thousands of high-performance chips running non-stop for days or weeks.</p><p>Goldman Sachs estimates that AI alone could drive a <strong>165% increase in data-center power demand by 2030</strong>. And power energy is only part of the issue. Many conventional data centers rely on millions of liters of water each day to stay cool straining local supplies already stressed by droughts and climate change.</p><p>Big technology firms are racing to secure clean energy: wind farms, solar fields, geothermal wells, even small modular nuclear reactors. But Earth has constraints; limited land, rising energy costs, and vulnerable water systems. The planet simply cannot sustain infinite physical expansion.</p><p>And so, some of the world’s most influential companies are now turning their eyes upward literally.</p><p><strong>Data Centers in Orbit: A Radical, Realistic Idea</strong></p><p>The concept sounds like science fiction, but it is rapidly becoming a serious engineering pursuit. If computing can’t keep scaling on Earth, why not place the hardware in orbit?</p><p>Google is already exploring this through <strong>Project Suncatcher</strong>, which aims to deploy AI-supporting compute modules aboard small satellites. These satellites would circle the planet, connected to earth’s network stations, and powered entirely by sunlight. In space, solar energy is constant: no nighttime, no cloudy weather, no atmospheric loss, making continuous power generation far more efficient.</p><p>Elon Musk has hinted that SpaceX could create orbital “data hubs” leveraging the enormous Starlink satellite network. Emerging startups such as <strong>StarCloud</strong>, working alongside NVIDIA, are designing systems to place advanced AI chips directly in space.</p><p>The theoretical benefits are immense:</p><ol><li><strong>Unlimited solar energy with zero fuel cost</strong></li></ol><p><strong>2</strong>. <strong>Vast physical space without land-use conflicts</strong></p><p><strong>3. Reduced strain on freshwater and local ecosystems</strong></p><p><strong>4. Potential for faster global data routing when paired with existing satellite networks</strong></p><p>If AI demands exponential growth, then space might be the only environment capable of supporting it sustainably.</p><p><strong>But Space Computing Comes with Serious Risks</strong></p><p>As with any frontier, the challenges are formidable:</p><ol><li><strong>Rocket launches emit significant carbon dioxide.</strong><br> A single launch can release hundreds of tons of emissions, including particulates that linger in the upper atmosphere.</li><li><strong>Space is harsh.</strong><br> Satellites experience extreme radiation, temperature swings, and micrometeoroid impacts. AI chips and storage hardware must survive conditions far beyond what data centers face on Earth.</li><li><strong>Maintenance is difficult and expensive.</strong><br> If something breaks, sending a repair mission is not as simple as dispatching a technician.</li></ol><p>Yet momentum is growing. Google has already tested its AI chips under simulated radiation conditions with successful results. As reusable rockets continue to lower launch costs, the economics of orbital computing become more plausible each year.</p><p>Space-based data centers are no longer just a futuristic idea. They represent a possible turning point: a shift from Earth-bound computing to a distributed, orbital architecture powered entirely by the sun.</p><p>Whether this becomes the next great leap in AI infrastructure or a technological dead-end, one thing is clear: the race to build the future of computing has already begun, and it’s taking place above our heads.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a3bd809a9a42" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ALIGNING AI WITH CLIMATE AND BUSINESS GOALS]]></title>
            <link>https://medium.com/@griffine_ombogo/aligning-ai-with-climate-and-business-goals-40b2fdc46773?source=rss-d0d0b163999f------2</link>
            <guid isPermaLink="false">https://medium.com/p/40b2fdc46773</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Griffine Ombogo]]></dc:creator>
            <pubDate>Sat, 06 Dec 2025 20:17:57 GMT</pubDate>
            <atom:updated>2025-12-06T20:49:17.778Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zpUg445ZAN4gC0dC" /><figcaption>Photo by <a href="https://unsplash.com/@tinkerman?utm_source=medium&amp;utm_medium=referral">Immo Wegmann</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Artificial intelligence (AI) is capturing global attention not only for its transformative potential across industries but also for its significant environmental impact. Unlike previous technological advances such as smartphones or cloud computing, AI’s rapid growth and the vast resources required to operate its data centers have made it a focal point for climate and environmental concerns. This spotlight is due to AI’s unique nature of demanding enormous amounts of energy and water, raising urgent questions about sustainability in the digital age.</p><p>One key reason AI faces sharper scrutiny is the scale of its resource consumption. In 2023, AI data centers accounted for about 4.4% of total electricity use in the United States, with projections indicating a tripling of this demand by 2028. These data centers not only require large amounts of electricity but also depend heavily on water for cooling, often in regions already facing water stress. This surge in resource use puts pressure on companies to carefully consider the environmental trade-offs involved in their AI investments. While AI technology brings substantial business value and innovation, firms must also adopt sustainable practices such as sourcing renewable energy, optimizing cooling techniques, and strategically locating facilities to minimize water use.</p><p>A notable example illustrating AI’s water footprint is the 2023 study ‘Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models’ that found ChatGPT’s, which is hosted on the state-of-the-art Microsoft’s data center, operations consume roughly 500 ml of water for every 20 to 50 prompts answered. While this amounts to 10 to 25 ml of water per prompt, the impact scales significantly across millions of users, emphasizing the broader footprint of AI’s energy and water consumption. Understanding these figures helps highlight the importance of addressing AI’s environmental costs on a systemic level.</p><p>To effectively manage AI’s carbon footprint, companies should focus on metrics such as Scope 1 emissions (direct emissions from owned sources), Scope 2 emissions (indirect emissions from purchased electricity), and Scope 3 emissions (related to supply chains). Additionally, performance indicators like Power Usage Effectiveness (PUE), Water Usage Effectiveness (WUE), and carbon intensity per kilowatt-hour can guide organizations toward more sustainable AI operations. Monitoring these measures allows firms to balance AI’s benefits against its environmental toll and identify improvement areas.</p><p>Industries that have quickly embraced AI, including healthcare, finance, automotive, and retail are leading the charge toward integrating AI while exploring greener infrastructure solutions. These sectors demonstrate that innovation and sustainability can progress hand in hand when companies prioritize environmental responsibility alongside technological advancement.</p><p>By fostering awareness of AI’s environmental impact and adopting strategic sustainability measures, businesses can ensure AI’s future growth aligns with global climate goals and resource conservation. This balanced approach is essential for harnessing AI’s power without compromising the planet.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40b2fdc46773" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>