<?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 Demostenes Brito on Medium]]></title>
        <description><![CDATA[Stories by Demostenes Brito on Medium]]></description>
        <link>https://medium.com/@demostenesbr?source=rss-581f007d953------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Qe6UAAEEAgFDA1RnGKIQ0w.jpeg</url>
            <title>Stories by Demostenes Brito on Medium</title>
            <link>https://medium.com/@demostenesbr?source=rss-581f007d953------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:56:14 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@demostenesbr/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[This is the best time to start a side project — but not in the way you imagine.]]></title>
            <link>https://medium.com/@demostenesbr/this-is-the-best-time-to-start-a-side-project-but-not-in-the-way-you-imagine-14f774d183de?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/14f774d183de</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[side-career]]></category>
            <category><![CDATA[enterpreneurship]]></category>
            <category><![CDATA[side-project]]></category>
            <category><![CDATA[innovation]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Thu, 27 Aug 2026 15:56:03 GMT</pubDate>
            <atom:updated>2026-08-27T15:57:22.533Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fiSXSsZMggKUQ8wWk6FIog.jpeg" /><figcaption>Image: Side Projects. This is best time to start — but not in the way you imagine.</figcaption></figure><p>“The starting point of all achievement is desire” — Napoleon Hill</p><p>If you’ve been keeping a list of project ideas for months — or years — 2026 might be the year the cost of bringing them to life finally drops low enough for you to stop putting them off. This isn’t about empty optimism, but rather a concrete shift in the landscape: the factors that used to hold back side projects until recently (time, technical expertise, capital) are no longer the primary bottlenecks.</p><p>What most articles on this topic fail to show is what to do with this open window of opportunity beyond just “build an AI app.” That is exactly what I want to break down here.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vPQZP5qX2w7t16TMsLnDcQ.jpeg" /><figcaption>Image: A scale comparing a large project with several small projects.</figcaption></figure><h4>What has actually changed</h4><p>Until a few years ago, launching a side project meant learning (or paying someone to implement) an entire stack of technologies before having anything to show for it. Today, “vibe coding” tools — where you describe what you want in natural language and let AI generate the implementation — have drastically shortened that gap. Editors like Cursor, application generators like Lovable and Replit Agent, and managed databases like Supabase form a combination that has become the standard for solo builders, offering estimated cost reductions of 80–90% compared to the traditional model of hiring a team.</p><p>This doesn’t eliminate the effort required to validate and sell, but it does remove a significant portion of the cost of simply giving it a try. And it is precisely this shift that makes the following question more important than “which app should I build?”: on how many different fronts can I apply the same core skill?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m04Q2rinsyXhuH1YfKmD5A.jpeg" /><figcaption>Image: Comparison between the traditional, centralized model and a new work approach involving different fronts.</figcaption></figure><h4>The framework: one core focus, multiple outputs</h4><p>The most superficial way to think about side projects is as a list of random ideas. The approach that truly pays off is to think in terms of a core skill — a central axis — with multiple outputs across different formats and monetization models. In practice, this means taking a competency you have already mastered — systems architecture, AI automation, data analysis, or whatever it may be — and applying it across at least two of these categories simultaneously:</p><ul><li><strong>Product:</strong> a tool, micro-SaaS, or AI wrapper that solves a specific, narrow problem.</li><li><strong>Content:</strong> documenting the process of building that product — architectural decisions, mistakes, lessons learned — through articles, a newsletter, or technical videos.</li><li><strong>Service:</strong> applying the same expertise as a one-off consultancy for those who lack the time or know-how to do it themselves (companies, firms, or professionals outside the tech bubble).</li><li><strong>Data/Insights:</strong> transforming what you learned during the build process into reports, dashboards, or analyses that others would pay for to avoid having to compile the data themselves.</li></ul><p>The key point is that these four fronts don’t compete for your time from scratch — they leverage the same foundational work. Building the product generates the content material. The content generates leads for the service. The service generates data for the next insight. Running “multiple projects” under this model is less costly than it seems, because each front subsidizes the next.</p><h4>Where are the specific opportunities in areas adjacent to technology?</h4><p>Here is the part often missing from this type of conversation: the most interesting opportunities for 2026 don’t lie merely in launching yet another micro-SaaS — they lie in the gaps that the rapid adoption of AI itself is creating in areas surrounding the technology.</p><ul><li><strong>Security review for “AI-generated” code. </strong>Recent research shows that a significant fraction of AI-generated code contains OWASP Top 10 vulnerabilities, and the number of vulnerabilities attributed to AI tools has been rising month by month in 2026. This poses a risk for those building too quickly — but it also presents a concrete opportunity for a side project: a rapid auditing service for solo founders who have moved from prototype to production without ever conducting a security review.</li><li><strong>Implementation consulting for non-tech companies.</strong> Accounting firms, medical clinics, local shops — businesses feeling the pressure to “use AI” but lacking in-house expertise to translate that into something functional. This is a massive, underserved adjacent market, especially for those with experience creating technical content for clients outside the development bubble.</li><li><strong>Technical content as a proprietary asset.</strong> With Gartner projecting that 80% of engineering professionals will need to upskill in AI tools by 2027, there is real demand for people who can clearly explain this landscape — to internal teams, to HR departments redesigning training paths, and to leadership teams deciding where to invest.</li><li><strong>Niche curation and data.</strong> Much of the noise in 2026 comes from people compiling lists of “the 10 best AI tools” without any filtering. Those who actually build and document what worked (and what didn’t) gain a credibility advantage that can be turned into products: reports, paid comparative studies, and industry analyses.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CBjng94M3mAXiocVbxyuww.jpeg" /><figcaption>Imagem: Descrição sobre oportunidades adjacentes à tecnologia.</figcaption></figure><h4>The Portfolio as a New Credential</h4><p>This drop in costs coincides with a shift in how the market evaluates technology professionals. Recruiters are increasingly seeking evidence of actual work rather than just degrees and certifications — and well-documented personal projects have become a direct way to demonstrate the ability to architect complete systems, not just write code. World Economic Forum estimates indicate that the demand for AI professionals is expected to outpace the supply of qualified talent by 30% to 40% by 2027.</p><p>In practice, this means that having two or three active initiatives — even small ones, or those not yet generating revenue — conveys something a traditional résumé cannot: that you know how to validate, build, sell, and iterate from end to end, across multiple contexts.</p><h4>The real bottleneck is no longer building.</h4><p>Building is no longer the bottleneck. The bottleneck is distribution — getting the first few dozen or hundred people to actually use what you’ve built or sign up for the service you’re offering. This shifts where it’s worth investing your time: fewer hours polishing the next feature, more hours validating whether there is real demand before writing a single line of code, and greater focus on distribution channels — which, if you already produce technical content, gives you an advantage that most people who only know how to program lack.</p><h4>A practical agenda for the rest of 2026 (and the leap into 2027)</h4><ul><li><strong>Choose a core competency, not just an isolated idea</strong>. Ask yourself across how many of the four fronts — product, content, service, or data — this skill can already be turned into something tangible.</li><li><strong>Validate before you build</strong>. A weekend spent researching demand is worth more than a month of coding without validation.</li><li><strong>Document the process as you build</strong>. This isn’t a separate step; it’s the content stream happening in parallel — for free.</li><li><strong>Look for gaps surrounding the technology</strong>, not just within it. Security, implementation for non-technical users, and internal training are markets with less competition than “yet another app.”</li><li><strong>Schedule time for a security</strong> review the moment any project moves from prototype to having real users — before it becomes its own case study on vulnerabilities.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gQj6hCKU9Qbfx3GZFgWRkQ.jpeg" /><figcaption>Image: Focus on the Process: Validate, Build, Secure, and Launch on a Recurring Basis. 2027 starts now.</figcaption></figure><p>Looking ahead to 2027, this dynamic is expected to intensify: more tools collapsing the distance between idea and product, more employers looking at portfolios rather than résumés, and an ever-widening gap between those who merely generate code with AI and those who can validate, sell, and sustain what they build. Starting more than one project now isn’t a lack of focus — it is the most rational way to position yourself for this next cycle.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=14f774d183de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Monolith to AI Platform. The anatomy of a modern monorepo.]]></title>
            <link>https://medium.com/@demostenesbr/from-monolith-to-ai-platform-the-anatomy-of-a-modern-monorepo-4cae24cbc567?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/4cae24cbc567</guid>
            <category><![CDATA[monolithic-architecture]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Fri, 24 Jul 2026 09:01:03 GMT</pubDate>
            <atom:updated>2026-07-24T09:01:03.225Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FGmDEBhP-ooLHieJ3DA4Cw.jpeg" /><figcaption>Image with ilustration including an office with a notebook, mousepad, books and more office objects</figcaption></figure><blockquote>How a project’s folder structure reveals architectural decisions, engineering maturity, and product vision.</blockquote><h3>Introduction</h3><p>Whenever I open the directory tree of a new project, I already know a great deal before reading a single line of code. The way a team organizes `apps/`, `packages/`, and `infrastructure/` tells a story: about how the product has grown, how many teams are working on it, and — increasingly — how Artificial Intelligence has shifted from being an “extra” to becoming a first-class citizen within the architecture.</p><p>In this article, I use the structure of a fictional reference project — the PediaCare Platform — to discuss a topic of interest to any developer or software engineer in 2026: how to organize systems that combine traditional product features, generative AI, multiple teams, and numerous external integrations — without falling into the trap of a rigid monolith or the chaos of poorly planned microservices.</p><pre>pediacare-platform/<br>apps/<br>├── website/                  # Next.js public site<br>├── admin/                    # Next.js admin portal<br>├── api/                      # NestJS REST API<br>├── ai-service/               # AI agents and RAG<br>├── worker/                   # Queues and scheduled jobs<br>└── notification-service/     # Email, SMS, push<br>packages/<br>├── ui/<br>├── auth/<br>├── api-client/<br>├── ai-core/<br>├── prompts/<br>├── validation/<br>├── shared/<br>├── types/<br>├── config/<br>└── logger/<br>ai/<br>├── agents/<br>├── accountant-agent/<br>├── tax-agent/<br>├── payroll-agent/<br>├── legal-agent/<br>├── content-agent/<br>├── seo-agent/<br>├── marketing-agent/<br>├── customer-support-agent/<br>├── reporting-agent/<br>└── orchestrator/<br>database/<br>├── prisma/<br>└── seed/<br>infrastructure/<br>├── docker/<br>├── kubernetes/<br>├── terraform/<br>├── monitoring/<br>└── github-actions/<br>docs/<br>├── architecture/<br>├── api/<br>├── prompt-engineering/<br>├── deployment/<br>└── adr/</pre><h3>1. Monorepo: one repository, multiple products</h3><p>The first key consideration is the choice of a monorepo. Instead of scattering the website, admin panel, API, AI service, worker, and notification service across separate repositories, everything coexists under one roof.</p><p>This decision isn’t just a trend — it solves real-world problems:</p><ul><li><strong>Unified versioning:</strong> a central `package.json` or `pnpm-workspace` avoids the classic “it works on my machine, but the shared package version is outdated in the other repo” scenario.</li><li><strong>Atomic refactoring:</strong> changing a type in `packages/types` and updating all consumers (API, admin, AI service) happens within a single Pull Request, reviewed and tested holistically.</li><li><strong>Standardized governance:</strong> `packages/config`, `packages/logger`, and `packages/validation` ensure that every application follows the same rules for linting, structured logging, and data validation — without relying on the goodwill of individual teams.</li><li><strong>The trade-offs are well-known:</strong> heavier builds, the need for tools like Turborepo or Nx for incremental caching, and a more sophisticated CI/CD pipeline. However, for products that heavily reuse business logic — such as a pediatric health platform requiring consistency across the patient app, admin portal, and backend automations — the gains in cohesion usually make it worthwhile.</li></ul><p>The cost is well known: a heavier build, the need for tools like Turborepo or Nx for incremental caching, and a more sophisticated CI/CD pipeline. However, for products that rely heavily on shared business logic — such as a pediatric health platform requiring consistency across a patient app, an administrative portal, and backend automations — the gain in cohesion usually makes it worthwhile.</p><h3>2. Apps: clear boundaries, distinct responsibilities</h3><p>Within `apps/`, each application plays a well-defined role:</p><ul><li>`website` and `admin` — both built with Next.js — separate the public-facing interface from internal operations. This prevents administrative configuration screens from leaking into the public site’s bundle and allows for completely different caching, SEO, and authentication policies tailored to each context.</li><li>`api` — built with NestJS — centralizes business logic and the REST layer, acting as the application’s traditional “brain” with clearly defined controllers, services, and modules.</li><li>`worker` isolates all asynchronous tasks: queues, scheduled jobs, and data reprocessing. This isolation ensures that spikes in batch processing do not degrade the performance of the API serving real-time user requests.</li><li>`notification-service` independently handles email, SMS, and push notifications — a domain with its own retry rules, external provider rate limits, and templates — keeping this logic separate from the API’s core functionality.</li><li>`ai-service` is the most recent highlight of this architecture and deserves a section of its own.</li></ul><pre>apps/<br>├── website/                  # Next.js public site<br>├── admin/                    # Next.js admin portal<br>├── api/                      # NestJS REST API<br>├── ai-service/               # AI agents and RAG<br>├── worker/                   # Queues and scheduled jobs<br>└── notification-service/     # Email, SMS, push</pre><p>In practice, this separation by bounded context is a microservices architecture “disguised” as a monorepo: the services are deployed independently, but shared code resides in `packages/`, avoiding duplication.</p><h3>3. AI as a first-class layer</h3><p>The most interesting aspect of the structure is that AI isn’t just scattered haphazardly throughout the API — it has a dedicated home:</p><ul><li>`apps/ai-service`: hosts AI agents and RAG (Retrieval-Augmented Generation) logic. This is where calls to language models, agent orchestration, and context retrieval from vector databases take place.</li><li>`packages/ai-core`: abstracts communication with LLM providers (model swapping, retries, streaming, function calling) so that ai-service and other applications aren’t tightly coupled to a single vendor.</li><li>`packages/prompts`: treats prompts as versioned artifacts — rather than loose strings scattered throughout the code. This allows the team to review, A/B test, and roll back prompts just as they would with database migrations.</li><li>`docs/prompt-engineering`: living documentation covering prompt engineering, agent design decisions, and evaluation results (evals).</li></ul><pre>ai/<br>├── agents/<br>├── accountant-agent/<br>├── tax-agent/<br>├── payroll-agent/<br>├── legal-agent/<br>├── content-agent/<br>├── seo-agent/<br>├── marketing-agent/<br>├── customer-support-agent/<br>├── reporting-agent/<br>└── orchestrator/</pre><p>This design solves a common problem in projects that “tack” AI onto a legacy system: a lack of traceability. When prompts, RAG logic, and model calls reside in dedicated packages and services, it becomes possible to audit, version, and test the AI ​​with the same rigor applied to the rest of the software — something essential in a sensitive domain like pediatric healthcare.</p><h3>4. CI/CD and Infrastructure as Code</h3><p>The `infrastructure/` folder demonstrates a platform designed for mature production operations:</p><p>`docker/`: containerizes each application independently, ensuring parity between local, staging, and production environments.</p><p>`kubernetes/`: orchestrates the deployment of multiple services, allowing the `ai-service` and `worker` to scale independently of the API — this is crucial because AI workloads and batch processing have CPU/memory consumption patterns that differ significantly from those of a traditional REST API.</p><p>`terraform/`: infrastructure as code to provision databases, queues, networks, and other cloud resources, with a versioned history that is reviewable via Pull Requests.</p><p>`monitoring/`: observability — metrics, alerts, and dashboards — treated as part of the source code rather than as manual configurations in a third-party dashboard.</p><p>`github-actions/`: CI/CD pipelines that independently test, build, and publish each application in the monorepo, typically using path filters to run the pipeline only for the specific application that has actually changed.</p><pre>infrastructure/<br>├── docker/<br>├── kubernetes/<br>├── terraform/<br>├── monitoring/<br>└── github-actions/<br>docs/<br>├── architecture/<br>├── api/<br>├── prompt-engineering/<br>├── deployment/<br>└── adr/</pre><p>This combination enables something essential for AI platforms: frequent and secure deployments. Since prompts and models evolve rapidly, the CI/CD pipeline must support AI-specific automated tests — such as agent output validation and prompt regression testing — with the same rigor applied to unit testing in traditional code.</p><p>The `docs/adr/` (Architecture Decision Records) folder completes this maturity cycle: every major architectural decision — such as “why we chose a monorepo” or “why we isolated the worker” — is documented, including context and considered alternatives, rather than existing solely in the decision-maker’s memory.</p><h3>5.Monolith vs. Microservices: What This Structure Teaches Us</h3><p>A well-organized monorepo like this one is, in practice, a mature middle ground between two extremes:</p><p><strong>Traditional monolith</strong></p><ul><li>✅ Operational simplicity, a single deployment, simple local transactions.</li><li>❌ Tight coupling: changing the notification logic may require redeploying the entire system.</li><li>❌ Teams stepping on each other’s toes in the same codebase; limited team scalability.</li><li>❌ It is difficult to scale specific parts (e.g., AI) without scaling everything together.</li></ul><p>“Pure” microservices (multiple repositories)</p><ul><li>✅ Independent deployment and scaling per service.</li><li>❌ Infrastructure overhead, code duplication, and difficulty maintaining consistent standards across repositories.</li><li>❌ Slower onboarding: understanding the entire system requires navigating multiple repositories with different histories and conventions.</li></ul><p>Slower onboarding: understanding the entire system requires navigating multiple repositories with different histories and conventions.</p><ul><li>✅ Independent deployment per application (via Docker/Kubernetes), but code and standards shared via packages/.</li><li>✅ Extensive and safe refactoring, with CI validating the entire impact within a single Pull Request.</li><li>✅ Selective scaling: the AI ​​service and worker can scale up resources without impacting the website or admin.</li><li>⚠️ It requires tooling discipline (incremental builds, CI path filters) to avoid losing agility as the repository grows.</li></ul><p>In practice, the “monolith vs. microservices” war is less relevant than the right question: where do the actual boundaries of change and scale lie within my domain? This structure addresses that by separating concerns based on the application (deployment boundary) and the package (code reuse boundary).</p><h3>6. Connecting with the world beyond the code</h3><p>No platform exists in a vacuum — and this is where many technical teams still leave value on the table. Integrating the engineering ecosystem with design and management tools isn’t just “window dressing”; it is part of the product architecture:</p><ul><li><strong>Figma → packages/ui:</strong> syncing design tokens (colors, typography, spacing) directly from Figma to the monorepo’s design system reduces the gap between what the product team designs and what gets implemented, while also speeding up the handoff between design and front-end.</li><li><strong>Canva:</strong> useful for rapidly generating communication materials (banners, posts, patient onboarding materials) without relying on the design team for every one-off asset — especially relevant for small squads that also handle product marketing and communications.</li><li><strong>Trello (or Jira/Linear):</strong> when connected via webhooks to the GitHub Actions pipeline, it allows task statuses to update automatically as a Pull Request progresses — reducing manual work and keeping non-technical stakeholders informed about actual development progress.</li><li><strong>Observability and support integrations (Slack, PagerDuty, Zendesk, etc.):</strong> connecting infrastructure and monitoring to communication channels turns technical alerts into rapid action, rather than logs gathering dust on a dashboard no one looks at.</li></ul><p>The lesson here is that a mature project’s `docs/architecture` folder shouldn’t just describe internal services, but also this set of external integrations — because these are what connect engineering work to the rest of the organization: design, product, marketing, and support.</p><h3>Conclusion</h3><p>A repository’s structure is not merely a matter of folder organization — it is a reflection of a product’s technical strategy. In the case of the PediaCare Platform, we see a project that:</p><ol><li>Uses a monorepo to maintain cohesion without sacrificing independent deployment per application;</li><li>Treats AI as a first-class component, featuring versioned prompts and isolated RAG logic;</li><li>Invests in infrastructure as code and granular CI/CD to support frequent, secure deployments;</li><li>Documents architectural decisions to preserve context over time;</li><li>Recognizes that external integrations — such as Figma, Canva, Trello, and the like — are part of the product architecture, not just an operational detail.</li></ol><p>For those transitioning from monolithic projects to more distributed architectures — or attempting to bring order to a growing set of microservices — this structure serves as a valuable reminder: the best architecture isn’t the trendiest one, but rather the one that clearly reflects the actual boundaries of change, scale, and responsibility within your domain.</p><blockquote>Did you find the analysis useful? If you are structuring a project that combines a traditional product with AI, it is worth reviewing how your prompts, agents, and data pipelines are organized — often, tomorrow’s technical debt is already being created today in the way we organize our folders.</blockquote><p>This article proposes a web repository adapted for Artificial Intelligence capabilities — presented as an option rather than a mandatory standard. The GitHub repository is publicly viewable, allowing for a hands-on look via the link below: <a href="https://github.com/demostenesbr/pediacare-platform">Pediacare Platform</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4cae24cbc567" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Personal SEO, GEO, AEO, and AIO: the new infrastructure for your technical reputation]]></title>
            <link>https://medium.com/@demostenesbr/personal-seo-geo-aeo-and-aio-the-new-infrastructure-for-your-technical-reputation-d5a2b386d047?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5a2b386d047</guid>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[careers]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Fri, 19 Jun 2026 19:17:11 GMT</pubDate>
            <atom:updated>2026-07-23T22:00:46.159Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FYs0VBAIFfCRGH5C7sGCJw.jpeg" /><figcaption>Public image on Hercilio Luz bridge — Florianoplis — SC — Brazil</figcaption></figure><blockquote>Resumes are being read by algorithms before reaching humans. Software engineers who ignore this layer are losing visibility to technically inferior but better-ranked professionals.</blockquote><p>There’s a silent irony in the tech market: the most skilled professionals at building search engines and AI (Artificial Intelligence) mechanisms, when poorly organized, are the worst at making sure those search engines find them. A senior engineer with 12 years of experience in distributed systems might be invisible to a recruiter who typed the words “senior backend engineer kafka aws” into a LinkedIn search box or another professional recruitment platform.</p><p>This isn’t bad luck. It’s architecture. And like any architecture, it can be intentionally designed so that recruiters and AI-orchestrated robots find the right professional.</p><p>In this article, I’ll map out four layers of what I call the professional visibility stack — Personal SEO (Search Engine Optimization), GEO (Generative Engine Optimization), AEO (Answer Engine Optimization), and AIO (AI Overview Optimization) — and how each interacts with the traditional resume strategy, especially for those working in technology and software engineering.</p><h3>Defining the Stack</h3><p>Before discussing strategy, we need precise definitions. These terms are often used interchangeably, which obscures their functional differences.</p><h4>Layer 01 — Personal SEO</h4><p>Optimizing your digital presence for traditional search engines (Google, Bing, or Yahoo) needs to be done efficiently. Focus on page indexing, domain authority, and the relevance of keywords associated with your name and professional expertise.</p><h4>Layer 02 — GEO</h4><p>Generative Engine Optimization. Structuring content to be cited and referenced by Large Language Models (LLMs) when answering questions about your field. What differs from classic SEO: it’s not about page ranking, but about being considered an authoritative source by generative models.</p><h4>Layer 03 — AEO</h4><p>Answer Engine Optimization. Optimization for systems that deliver direct answers — featured snippets, voice search, related questions. The goal is to be the answer, not just appear in the results.</p><h4>Layer 4 — AIO</h4><p>AI Overview Optimization. Specifically for the summary panels that Google and other search engines now automatically generate at the top of results, consolidating multiple sources into an AI-generated summary.</p><h4>⚡ Critical distinction</h4><blockquote>Classic SEO solves the problem of being found by humans via algorithm. GEO/AEO/AIO solve the problem of being cited by AI when answering human questions. The difference in flow is fundamental: in the first, you appear in a list. In the second, you become part of the answer.</blockquote><h3>Google Yourself: The Ultimate Guide to Personal SEO</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FYbA1bN6PlWM%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DYbA1bN6PlWM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FYbA1bN6PlWM%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/03b5590d0ad33a3f70e2ee9f2a4e5e09/href">https://medium.com/media/03b5590d0ad33a3f70e2ee9f2a4e5e09/href</a></iframe><p>The short video, published in 2024, explains how Personal SEO is an effective combination of technology, branding, and communication working together in a landscape where the internet is our new business card. And what it says about you matters; it’s time to take control of your personal digital brand.</p><h3>The curriculum as a legacy artifact</h3><p>The traditional CV (Curriculum Vitae) was designed for a world where humans read linear documents and made decisions based on formatting and chronological narrative. That world still exists, but now there’s a layer before it: ATS (Applicant Tracking Systems), AI parsers, and queries in tools like LinkedIn, GitHub, and even ChatGPT.</p><p>When a technical recruiter at a fintech company types “backend engineer with experience in event-driven architecture and Python” into any search engine — from LinkedIn to a corporate LLM — what determines whether you show up isn’t the quality of your experience. It’s whether you used those words, in that context, in places that those systems index.</p><blockquote>Your resume is the smallest of your documents. The largest is the sum of everything you’ve ever published on the internet.</blockquote><h3>Where the CV Technically Fails</h3><p>For a software engineer, the problem has specific dimensions:</p><p><strong>Fragmented terminology</strong>. The field uses technical synonyms inconsistently. “Microservices,” “microservices,” “distributed architecture,” “distributed systems” — a professional may have real experience but use only one term when search engines expect variations. Simple Applicant Tracking Systems (ATS) do not make semantic inferences.</p><p>Low context density. CVs often list technologies without context: “Python, AWS, Kafka.” Modern matching systems need context to infer depth: at what scale? For what problem? With what result?</p><p>Absence of external signaling. A CV is a document without backlinks, without mentions, without external validation. For ranking algorithms — whether search or LLMs — authority is social and referential, not self-declared.</p><h3>SEO: Your personalized legacy</h3><p>LinkedIn is necessary, but insufficient. The logic of Personal SEO for a software developer or software engineer works in layers of increasing authority:</p><h3>Level 1 — Own Property</h3><p>A personal domain (yourdomainname.dev or similar) is the only asset you completely control. It’s where you define the canonical Uniform Resource Locator (URL), structured data (JSON-LD with features like Person, knowsAbout, and hasOccupation), and highly specific technical content. Google indexes technical articles with real depth well — not shallow posts, but content that answers questions that few people answer well in the current landscape.</p><h3>Level 2 — Platforms with high DA (Domain Authority)</h3><p>Medium, Dev.to, Substack, and GitHub Pages have high Domain Authority. An article of yours on these platforms about a specific technical topic can rank above official documentation for long-tail queries. The strategy isn’t volume — it’s specificity. A post about “how to implement a circuit breaker pattern with Resilience4j in production with metrics on Datadog” has orders of magnitude less competition than “microservices best practices”.</p><h3>Level 3 — External Validation</h3><p>Mentions, citations, and contributions in relevant repositories function as backlinks of technical authority. Contributing to documented open-source projects, being mentioned in release notes, having signed issues in popular repositories — all of this creates a reference graph that algorithms interpret as an authority signal.</p><p>✓ Advantages</p><ul><li>Complete control over narrative and positioning</li><li>Compound effect over time (articles age with authority)</li><li>Independent of third-party platforms for self-ranking</li><li>Creates a surface area for inbound opportunities</li></ul><p>— Disadvantages</p><ul><li>High initial time investment</li><li>Results take months to years to consolidate</li><li>Requires consistency — sporadic articles have little effect</li><li>Risk of dispersion: shallow coverage of many topics does not rank well</li></ul><h3>GEO: being cited by AI, not just found</h3><p>GEO is the newest and possibly most relevant frontier for technical professionals in the medium term. The logic is as follows: when someone asks a corporate LLM — or ChatGPT, Gemini, Perplexity — “who are the event sourcing experts in Brazil?” or “which engineer has proven experience migrating from monoliths to microservices?”, the model responds based on training data and, increasingly, on real-time web retrieval.</p><p>To appear in these answers, it is necessary to have created what I call a topical authority corpus: a dense, consistent, and referenced set of content about a specific domain.</p><h3>How LLMs determine authority</h3><p>During training, language models learn co-occurrence patterns. If your name appears repeatedly in contexts like “systems architect,” “kafka,” “event-driven” across multiple independent sources, the model develops semantic associations between these entities. This isn’t manipulable in the way SEO was in 2010 — there’s no functional keyword stuffing here. What works is real, widely distributed depth.</p><p>For systems with real-time retrieval (Perplexity, ChatGPT with search), the logic is closer to AEO: the technical quality of the content, the structure for fact extraction, and the presence in sources that these systems prioritize.</p><blockquote>Practical tactic: Write content that answers specific technical questions in an explicit question-and-answer format. LLMs with retrieval prioritize content where the question and answer are structurally clear. A well-written README with a technical FAQ section is more likely to be cited than a narrative blog post with the same scattered information.</blockquote><h3>AEO and AIO: the direct response layer</h3><p>AEO (Answer Engine Optimization) predates the generative AI movement — it emerged with the growth of Google’s Featured Snippets and voice search. For technical professionals, the most relevant application is ensuring that questions about their specialty return their content as a canonical answer.</p><p>AIO (AI Overview) is specific to the generative dashboards that Google displays above organic results since 2023/2024. These dashboards consolidate multiple sources and are generated by a model trained to synthesize. To appear on them, the same rules as GEO apply, with one addition: well-implemented structured data (Schema.org) increases the likelihood of correctly extracting facts about you.</p><h3>Structured data for technical identity</h3><p>A minimal but effective JavaScript Object Notation for Linked Data (JSON-LD) block for a software engineer on their personal website:</p><pre>{<br>  &quot;@context&quot;: &quot;https://schema.org&quot;,<br>  &quot;@type&quot;: &quot;Person&quot;,<br>  &quot;name&quot;: &quot;Seu Nome&quot;,<br>  &quot;jobTitle&quot;: &quot;Senior Software Engineer&quot;,<br>  &quot;knowsAbout&quot;: [&quot;distributed systems&quot;, &quot;event-driven architecture&quot;, &quot;Apache Kafka&quot;],<br>  &quot;sameAs&quot;: [<br>    &quot;https://github.com/seuusuario&quot;,<br>    &quot;https://linkedin.com/in/seuperfil&quot;<br>  ],<br>  &quot;url&quot;: &quot;https://seusite.dev&quot;<br>}</pre><h3>Longitudinal view: a perspective on what works in 1, 3 and 10 years</h3><p>The biggest pitfall in this discussion is treating Personal SEO and Geo as short-term tactics. They are infrastructure investments. Look at how they perform over time:</p><h4>First Year</h4><p>Establishing a base: own domain, standardization of cross-platform presence, start of consistent publishing. Search results begin to be indexed. Little direct return yet — foundation building phase.</p><h4>2 or 3 Years</h4><p>Technical articles are starting to rank for long-tail queries. Initial organic inbound leads (invitations for speaking engagements, interviews, consulting). LLMs with more recent training data are beginning to associate your name with specific topics. LinkedIn is consistently appearing in searches for your name + specialty.</p><h4>5+ Years</h4><p>Compounding effect: each new piece of content has an already established authority base. External mentions grow organically. Digital reputation begins to function as an asset — generating opportunities without active effort proportional to the return.</p><h4>10+ Years</h4><p>In this context, the distinction between “real reputation” and “digital reputation” disappears. Professionals who have invested in this layer are the ones LLMs cite as references, who appear on expert lists, and who receive inbound calls from companies that don’t even know they were looking for them.</p><h3>Decision matrix: what to prioritize</h3><p>Given that time is finite, the practical question is: where to allocate energy? The table below maps each strategy by impact, timeframe, and maintenance cost for a typical software engineer:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/842/1*MwjXlzmaZ7ycbtUhMRIPpQ.png" /><figcaption>Image: priority table in the decision matrix</figcaption></figure><h3>What separates those who are found from those who are ignored?</h3><p>The conversation about Personal SEO often degenerates into superficial advice: “post more on LinkedIn,” “write articles,” “be consistent.” This is necessary but insufficient as a framework for a systems-thinking engineer.<br>What truly separates highly visible IT professionals from the invisible ones — with equivalent technical competence — is a combination of three structural factors:</p><p>Niche specificity. Being a “senior software engineer” is a commodity. Being “the engineer who wrote the three best articles on observability in Kotlin systems in Brazil” is not. Algorithms, recruiters, and LLMs prefer strong, specific associations to weak, generic ones.</p><p>Cross-platform distribution with canonicalization. The same content should exist across multiple indexing points (LinkedIn, GitHub, personal website, Medium), but with a clear canonical source. This maximizes surface area without diluting authority.</p><p>Temporal consistency. Ranking systems — both search and LLMs — favor enduring signals. A profile that has been active for 5 years with regular posts has fundamentally different authority than a profile created yesterday with 50 simultaneous posts.</p><h4>⚡Final perspective</h4><p>As LLMs become the primary search layer for hiring decisions — and this is already happening at top tech companies — the question that defines careers has quietly shifted. It’s no longer “Does your resume impress a human?” It’s “Does your digital presence convince a language model that you’re the go-to person for the problem their team has?”</p><p>This transition isn’t going to announce its arrival. It’s already underway. And the window in which building that presence is still a differentiator — before it becomes mandatory — is right now.</p><p>The best time to plant a tree was ten years ago. The second best time is today. But unlike trees, this window closes.</p><h4>⚡Bonus: Three models for AI robots and LLMs</h4><p>A model for visibility in search engines and Artificial Intelligence.</p><pre># robots.txt otimizado para visibilidade em mecanismos de busca e IA<br># Atualizado: 2025-2026<br># Referência: seusite.dev/robots.txt<br>#<br># ESTRATÉGIA: permitir crawlers de RETRIEVAL (citação em tempo real)<br># e tomar decisão consciente sobre crawlers de TREINAMENTO<br><br># =============================================<br># MECANISMOS DE BUSCA TRADICIONAIS<br># =============================================<br><br>User-agent: Googlebot<br>Allow: /<br><br>User-agent: Bingbot<br>Allow: /<br><br>User-agent: DuckDuckBot<br>Allow: /<br><br># =============================================<br># OPENAI<br># =============================================<br><br># OAI-SearchBot: alimenta busca em tempo real no ChatGPT<br># RECOMENDADO PERMITIR para aparecer em respostas do ChatGPT agora<br>User-agent: OAI-SearchBot<br>Allow: /<br><br># ChatGPT-User: fetch ativo quando usuário faz pergunta no ChatGPT<br># RECOMENDADO PERMITIR — é o bot que gera citações em tempo real<br>User-agent: ChatGPT-User<br>Allow: /<br><br># GPTBot: coleta dados para treinamento de modelos futuros<br># DECISÃO ESTRATÉGICA: Allow = seu conteúdo pode entrar no próximo GPT<br># Disallow = protege seu conteúdo de ser usado em treinamento<br>User-agent: GPTBot<br>Allow: /<br><br># =============================================<br># ANTHROPIC (CLAUDE)<br># =============================================<br><br># ClaudeBot: indexação e treinamento<br>User-agent: ClaudeBot<br>Allow: /<br><br># Claude-User: retrieval em tempo real quando usuário usa Claude<br># RECOMENDADO PERMITIR<br>User-agent: Claude-User<br>Allow: /<br><br># Claude-SearchBot: busca web ativa via Claude<br># RECOMENDADO PERMITIR<br>User-agent: Claude-SearchBot<br>Allow: /<br><br># Strings depreciadas da Anthropic — manter por compatibilidade<br># User-agent: Claude-Web  (depreciado)<br># User-agent: Anthropic-AI  (depreciado)<br><br># =============================================<br># PERPLEXITY<br># =============================================<br><br># PerplexityBot: indexação para o mecanismo de busca Perplexity<br>User-agent: PerplexityBot<br>Allow: /<br><br># Perplexity-User: retrieval em tempo real no Perplexity<br># RECOMENDADO PERMITIR<br>User-agent: Perplexity-User<br>Allow: /<br><br># =============================================<br># GOOGLE AI<br># =============================================<br><br># Google-Extended: usado para treinamento do Gemini e geração de AI Overviews<br># DECISÃO ESTRATÉGICA: permitir aumenta chances de aparecer em AI Overviews<br>User-agent: Google-Extended<br>Allow: /<br><br># =============================================<br># OUTROS CRAWLERS DE IA<br># =============================================<br><br>User-agent: PerplexityBot<br>Allow: /<br><br>User-agent: YouBot<br>Allow: /<br><br>User-agent: PhindBot<br>Allow: /<br><br># =============================================<br># CRAWLERS QUE FREQUENTEMENTE IGNORAM ROBOTS.TXT<br># =============================================<br><br># Bytespider (ByteDance/TikTok): alta incidência de ignorar regras<br># Disallow aqui, mas bloqueio real deve ser feito em nível de servidor/WAF<br>User-agent: Bytespider<br>Disallow: /<br><br># CCBot (Common Crawl): usado por múltiplas organizações para treinamento<br># Decida conforme sua política de uso de dados<br>User-agent: CCBot<br>Disallow: /<br><br># =============================================<br># WILDCARD: todos os outros bots<br># =============================================<br><br># Allow: / significa que bots não listados têm acesso completo<br># Mude para Disallow: / se quiser política restritiva por padrão<br>User-agent: *<br>Allow: /<br><br># =============================================<br># SITEMAP<br># =============================================<br><br>Sitemap: https://seusite.dev/sitemap.xml</pre><p>Reading template for LLMs, in a simplified version ideal for a professional resume.</p><pre># Seu Nome — Engenheiro de Software<br><br>&gt; Engenheiro de software com X anos de experiência especializado em [seu nicho principal].<br>&gt; Autor de conteúdo técnico sobre [tecnologia 1], [tecnologia 2] e [tecnologia 3].<br>&gt; Baseado em [cidade], Brasil. Disponível para consultorias e palestras.<br><br>## Identidade profissional<br><br>- Nome completo: Seu Nome Completo<br>- Especialidade principal: [Ex: Sistemas distribuídos e arquitetura event-driven]<br>- Tecnologias core: [Ex: Kafka, Kubernetes, Golang, AWS]<br>- Setor de atuação: [Ex: Fintech, E-commerce, Plataformas de dados]<br>- Anos de experiência: X anos<br>- Idiomas: Português (nativo), Inglês (fluente)<br><br>## Sobre<br><br>- [Sobre mim](/about): Trajetória profissional completa, filosofia de engenharia e o que me motiva<br>- [Currículo completo](/cv): Experiência, formação e certificações<br>- [Projetos](/projects): Portfólio de trabalhos open source e contribuições relevantes<br><br>## Artigos técnicos — conteúdo de referência<br><br>- [Título do artigo 1](/posts/slug-1): Descrição de uma linha do que o artigo resolve ou ensina<br>- [Título do artigo 2](/posts/slug-2): Descrição de uma linha do que o artigo resolve ou ensina<br>- [Título do artigo 3](/posts/slug-3): Descrição de uma linha do que o artigo resolve ou ensina<br><br>## Palestras e apresentações<br><br>- [Nome do evento — Ano](/talks/slug): Tema da palestra, contexto e link para slides se disponível<br><br>## Projetos open source<br><br>- [Nome do projeto](https://github.com/usuario/repo): O que o projeto faz e qual problema resolve<br>- [Nome do projeto](https://github.com/usuario/repo): O que o projeto faz e qual problema resolve<br><br>## Publicações externas<br><br>- [Título do artigo no Medium/Dev.to](https://link): Publicação em plataforma externa<br>- [Contribuição em repositório relevante](https://github.com/link): Contexto da contribuição<br><br>## Contato e presença digital<br><br>- Site: https://seusite.dev<br>- GitHub: https://github.com/seuusuario<br>- LinkedIn: https://linkedin.com/in/seuperfil<br>- Email profissional: contato@seusite.dev<br>- Newsletter (se aplicável): https://substack.com/seuperfil<br><br>## Contexto para sistemas de IA<br><br>Este site contém conteúdo técnico original sobre [lista de tópicos].<br>O conteúdo é produzido inteiramente por Seu Nome e reflete experiência<br>prática em projetos de produção com [escala/contexto relevante].<br>Para citações, a fonte canônica é https://seusite.dev.</pre><p>A reading model for LLMs in an extended version, including articles, lectures, academic activities, and other professional activities.</p><pre># Seu Nome — Contexto completo para sistemas de IA<br><br>&gt; Este arquivo consolida o conteúdo essencial do site seusite.dev em formato<br>&gt; otimizado para leitura por modelos de linguagem. Última atualização: YYYY-MM-DD.<br><br>---<br><br>## Identidade e posicionamento<br><br>**Quem sou:** Engenheiro de software com X anos de experiência, especializado em [nicho].<br>Atuo principalmente com [tecnologias core] em contextos de [tipo de empresa/projeto].<br><br>**O que produzo:** Artigos técnicos aprofundados sobre [tópicos], com foco em casos<br>reais de produção — não tutoriais introdutórios, mas análises de problemas complexos<br>com soluções testadas em escala.<br><br>**Trajetória resumida:**<br>- [Empresa atual] — [Cargo] (YYYY–presente): [Responsabilidades principais em 1-2 frases]<br>- [Empresa anterior] — [Cargo] (YYYY–YYYY): [Responsabilidades principais em 1-2 frases]<br>- [Empresa anterior] — [Cargo] (YYYY–YYYY): [Responsabilidades principais em 1-2 frases]<br><br>**Formação:**<br>- [Curso] — [Instituição] (YYYY)<br>- [Certificação relevante] — [Organização] (YYYY)<br><br>---<br><br>## Artigo: [Título completo do artigo 1]<br><br>*Publicado em: YYYY-MM-DD | URL: https://seusite.dev/posts/slug-1*<br><br>[Conteúdo completo do artigo em Markdown limpo, sem HTML, sem menus de navegação.<br>Inclua introdução, desenvolvimento e conclusão. Use headings (#, ##, ###) para<br>estrutura. Inclua exemplos de código com blocos de código quando relevante.]<br><br>---<br><br>## Artigo: [Título completo do artigo 2]<br><br>*Publicado em: YYYY-MM-DD | URL: https://seusite.dev/posts/slug-2*<br><br>[Conteúdo completo do artigo 2]<br><br>---<br><br>## Projeto open source: [Nome do projeto]<br><br>*URL: https://github.com/usuario/nome-projeto | Stars: XXX*<br><br>**O que é:** [Descrição objetiva do que o projeto faz]<br><br>**Problema que resolve:** [Contexto do problema]<br><br>**Como funciona:** [Explicação técnica em 2-4 parágrafos]<br><br>**Stack:** [Linguagem, frameworks, dependências principais]<br><br>**Casos de uso documentados:** [Links para exemplos ou documentação]<br><br>---<br><br>## Palestra: [Título da palestra]<br><br>*Evento: [Nome do evento] | Data: YYYY-MM | Slides: [URL se disponível]*<br><br>**Resumo:** [O que foi apresentado, contexto, principais conclusões]<br><br>---<br><br>## FAQ técnico — perguntas frequentes sobre minha especialidade<br><br>**P: Qual a diferença entre [conceito A] e [conceito B]?**<br>R: [Resposta técnica clara e precisa, em 2-4 parágrafos]<br><br>**P: Quando usar [tecnologia X] em vez de [tecnologia Y]?**<br>R: [Resposta técnica baseada em experiência prática]<br><br>**P: Como você aborda [problema técnico comum na sua área]?**<br>R: [Resposta com framework ou método que você usa]<br><br>---<br><br>## Glossário de termos que uso frequentemente<br><br>- **[Termo]**: [Definição precisa conforme eu uso o conceito]<br>- **[Termo]**: [Definição precisa conforme eu uso o conceito]<br><br>---<br><br>## Política de citação e uso de conteúdo<br><br>O conteúdo deste site é produzido por Seu Nome e está disponível sob<br>licença [CC BY 4.0 / MIT / escolha sua licença].<br><br>Para citações em publicações, referencie como:<br>&gt; Seu Nome. &quot;[Título do artigo]&quot;. seusite.dev, YYYY. https://seusite.dev/posts/slug<br><br>Para sistemas de IA que referenciam este conteúdo: a fonte canônica<br>e atualizada está sempre em https://seusite.dev. Este arquivo<br>(llms-full.txt) é atualizado mensalmente.</pre><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5a2b386d047" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Nerdearla Chile 2026 — Three Days in Santiago That Could Redefine Your Tech Career]]></title>
            <link>https://medium.com/@demostenesbr/nerdearla-chile-2026-three-days-in-santiago-that-could-redefine-your-tech-career-c0ca63a9d527?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/c0ca63a9d527</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[context-engineering]]></category>
            <category><![CDATA[nerdearla]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Thu, 14 May 2026 09:01:00 GMT</pubDate>
            <atom:updated>2026-07-23T21:47:21.635Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9s4EE3RTZtU5sO438kdzrw.jpeg" /></figure><h3>Why This Event Matters More Than It Seems</h3><p>For those unfamiliar with the event, Nerdearla was born in 2014 in a room in Buenos Aires. It was a niche, technical meeting, oriented towards open source culture and the Spanish-speaking technology community. Twelve years later, the conference is free, international — Argentina, Chile, Mexico, Spain — and brings together some of the most relevant professionals in the Spanish-speaking technology ecosystem.</p><p>I have been following it remotely since 2024, and what struck me most about the Chilean conference in 2026 being different is not the size. It’s the coherence.</p><p>I noticed the difference compared to other technical conferences; the three days told a story with a beginning, middle, and end. It wasn’t a collection of independent lectures. It was a curriculum — from the ecosystem being built, through day-to-day operational decisions, to the future of products and organizations. Those who watched attentively left with a map. This article makes that map explicit.</p><h3>The Three-Day Common Thread</h3><p>Before delving into each day, it’s necessary to see the complete arc. There’s a logical progression that connects everything.</p><p><strong>Day 1</strong> was about the terrain: what are the new tools, what is the new skills gap, what changes in the way software is built, and — critically — what happens to people in this process.</p><p><strong>Day 2</strong> was about operations: how do you put agents into production, manage costs, protect systems, scale teams, and make architectural decisions in an environment where AI is infrastructure, not a feature.</p><p><strong>Day 3</strong> was about the destination: where are products, organizations, and careers headed when AI ceases to be a differentiator and becomes a commodity?</p><p>Together, the three days answer the question every technology professional is asking in 2026: what do I do now?</p><h3>Day 1 — The Terrain: Tools, Skills, and the Human Cost of Acceleration</h3><p>The first day established the state of the world. Not the world five years from now — the world of this week.</p><p><strong>Mojo &amp; MAX</strong> opened the event positioning a new stack layer. Modular built Mojo as the language that unites the ergonomics of Python with the performance of C++, and MAX as the unified inference engine that runs on top of it. The message for senior engineers is direct: the AI ​​framework war is not over yet, and betting early on an abstraction that solves the tradeoff between development speed and execution speed can be a real competitive advantage.</p><p><strong>Burnout: The Hidden Failure Mode of High-Performing Teams</strong> was, paradoxically, the most technical talk of the day — but about human systems. The thesis is precise: burnout is not an event, it’s a process. It appears first as a silent drop in quality, then as presenteeism, then as leaving. High-performance teams are the most vulnerable because the delivery culture masks the signs. The most troubling statistic for the audience: the average interval between the first signs and visible collapse is six to eight months. Enough time to intervene — if you know what to look for.</p><p><strong>Del Vibecoding a la Ingeniería</strong> was where the event established its most important position. The new digital divide isn’t access. It’s understanding. Using AI to generate code without understanding what’s being generated is accumulating technical debt at a speed that conventional auditing tools can’t keep up with. Vibecoding works for personal prototypes. In production systems, it’s a silent risk that manifests itself at the worst possible time.</p><p><strong>Chaos Testing for Chatbots</strong> provided the operational answer for those who already have agents in production. Simulating adversarial users — the impatient, the contradictory, the ambiguous, the malicious, the completely out-of-scope — isn’t paranoia. It’s QA applied to the emergent behavior of probabilistic systems. Without it, you discover your agent’s flaws when a real customer finds them.</p><p><strong>Del Vibe Caos al Despertar Spec-Driven</strong> completed the argument: the discipline that contains generative chaos is specification ahead. Before any prompt, before any line of AI-generated code, you need a spec — objectives, constraints, user personas, success criteria, definition of what is out of scope. This is not bureaucracy. It’s the difference between an agent that does what you want and an agent that does what you said.</p><p><strong>Extiende tus Agentes: Tools &amp; Memoria</strong> established the fundamental technical distinction: external tools that extend what the agent can do, and persistent memory that provides continuity of context between sessions. Without tools, you have a language model. With tools and memory, you have a collaborator that learns from the history of interactions and acts in the real world.</p><p>The other talks of the day — <strong>UX Research without killing creativity, Web Performance with DevTools, static binaries as intentional architecture</strong>,<strong> Wikimedia and open data</strong>,<strong> The Third Golden Age</strong>,<strong> Vibe Coding with AI Studio </strong>and<strong> multi-agent systems with ADK</strong> — completed the panorama: from interface performance to architectural philosophy, Day 1 mapped both what is being built and what is being destroyed in the process.</p><h3>Day 2 — The Operation: Agents in Production, Real Costs, and Human Architecture</h3><p>If Day 1 was about the terrain, Day 2 was about surviving on it. The schedule was the densest of the three days — and intentionally so. Agent engineering is not theory. It has costs, security flaws, architectural tradeoffs, and people who need to deliver results in real sprints.</p><p><strong>Build Your First Agent with Open Source</strong> delivered the most liberating message of the event: you don’t need an expensive proprietary LLM to have a functional agent in production. Open source has reached a level where the barrier to entry has dropped drastically. For founders, indie developers, and teams with tight budgets, this completely changes the equation.</p><p><strong>Your Repository Knows More About Your Team Than Your Team</strong> was the talk that should be most discussed in engineering leadership meetings. Mining repositories as organizational intelligence: the history of commits, PRs, and issues reveals knowledge concentration, hidden bottlenecks, dependencies on key people, and work patterns that no status meeting captures. Who writes the most critical commits? Where does the knowledge reside in the head of a single person? When was the last time someone touched a high-risk module?</p><p><strong>Skills.md and Agents.md</strong> were the most immediately applicable talks of the entire event. Two files that every agent-based codebase needs. Skills.md documents what the agent can do — its capabilities, its data sources, its reliability limits. <strong>Agents.md</strong> defines how it should behave — tone, allowed tools, ethical constraints, escalation to humans. It’s the emerging standard that will become the norm as quickly as README.md became mandatory. Those who start adopting now will define the standard for the team.</p><p><strong>El 80% de AI es gratis. El 20% restante puede costarte todo</strong> was the moment of sobriety the event needed. Free or low-cost models solve the vast majority of real-world use cases. The trap lies in building dependencies on premium features without measuring ROI — and discovering the problem in the following month’s bill, when the dependency is already in the production code. Auditing the AI ​​stack with the same regularity as you audit cloud infrastructure has gone from being a good practice to an architectural obligation.</p><p><strong>Game Over na Fatura: Tetris de Custos em Kubernetes</strong> delved into the economic theme with surgical precision. AI agent clusters have a different cost profile than conventional web applications — they are more unpredictable, more dependent on usage patterns, and more sensitive to model choices. Right-sizing, spot instances, workload-based scaling policies, and real-time financial observability are skills that every platform architect needs to master now.</p><p><strong>Antes era una Locura Migrar Monolitos. Ahora lo Hace un Agente</strong> was honest in both directions. Yes, AI agents are dramatically accelerating migrations that previously took months of analysis and refactoring. And also: human supervision is still indispensable, risks exist, and there are cases where automation creates new complexity instead of solving old ones. The talk didn’t sell a magic solution — it presented a tool with real tradeoffs.</p><p><strong>Ciberseguridad en la Era de la IA</strong> was the talk that should most be on security roadmaps for 2026. Prompt injection, context poisoning, data exfiltration via tool calls — attack vectors that didn’t exist two years ago and that most security teams still don’t have answers for. Ignoring security in agentic systems is repeating, on a much larger scale, the mistakes of the early years of the web.</p><p><strong>De Prompt Engineering a Context Engineering</strong> was the most important conceptual evolution of the day. The right question isn’t “how do I write a better prompt?” It’s “what context does this model need to act correctly in this specific situation?” Managing context is managing behavior. The talk brought real learnings from a code review product with AI — including the mistakes made, which made it exponentially more valuable than a successful presentation. Teams that Scale: Human Architecture for High Performance concluded the organizational axis with a thesis that deserves repetition: tech teams that scale need organizational architecture that is as carefully considered as their technical architecture. Autonomy without alignment leads to fragmentation. Alignment without autonomy leads to bottlenecks. Balance doesn’t happen by accident — it’s intentional design, reviewed as frequently as technical design.</p><p><strong>Equipos que Escalan: Arquitetura Humana para Alto Rendimento</strong> closed the organizational axis with a thesis that deserves repetition: tech teams that scale need organizational architecture that is as carefully considered as their technical architecture. Autonomy without alignment becomes fragmentation. Alignment without autonomy becomes a bottleneck. Balance doesn’t happen by accident — it’s intentional design, reviewed as frequently as technical design.</p><p>The other talks — <strong>fraud and machine learning with analysts in the loop, monitoring what really matters, architectural roles in the age of AI, learning by building useless things, the development lifecycle with AI, open source contribution in practice, scaling systems and teams together, the AI ​​agent and the job you didn’t ask for, </strong>and<strong> LLMs running on Kubernetes with open source tools </strong>— completed the most operationally dense day of the three, with a consistent underlying message: technical responsibility and human responsibility are inseparable.</p><h3>Day 3 — The Destination: Products, Organizations, and Careers in the Cognitive Age</h3><p>The closing was a deliberate zoom-out. After two days of technical and operational depth, Day 3 asked the question that the other two prepared the groundwork to answer: where is all this going?</p><p><strong>AI-first sin UI-first</strong> touched on a shift that is not yet a consensus, but is already a reality in cutting-edge products. Cognitive systems are replacing graphical interfaces in cases where interaction is oriented towards complex tasks. The design question changes completely: it is no longer “how does the user navigate the screen?” but “how does the user build trust with an agent that acts for them?” Engineers and designers who understand this shift first will not only build better products — they will define what “better” means.</p><p><strong>IA y el Futuro de los Productos Digitales</strong> was direct: products that do not incorporate AI into the core operational layer — not as a decorative feature, but as a decision-making and execution mechanism — will irreversibly lose competitiveness within a product lifecycle. The most honest discussion was about how to make this transition without breaking what already works and without creating blind dependence on technology that is still changing at the speed of quarters, not years.</p><p><strong>Developers Documentation: Your Hidden Strength</strong> was the most surprising talk due to its philosophical depth. Documentation is not a record of past decisions. It’s programming future behavior — of your colleagues, new team members, and the agents that will operate on this codebase. Engineers who know how to document well have a disproportionate advantage in distributed teams. And in an era where agents consume context to act, the quality of your documentation directly determines the quality of the behavior of your artificial collaborators.</p><p><strong>Building a GenAI-Driven Developer Organization</strong> showed the organizational side of the same coin: how to structure an engineering team that uses GenAI systematically and responsibly — with clear roles, defined guardrails, human review processes at the right points, and metrics that measure real impact, not superficial adoption.</p><p><strong>Construyendo un Sistema de Diseño Agéntico</strong> was the most ahead-of-its-time talk of the entire event. Design systems for agents: reusable patterns of behavior, tone, limitations, and interfaces that ensure consistency between different agents in the same product. It’s the same discipline that transformed front-end development in the last decade, now applied to the behavior of cognitive systems.</p><p><strong>Thoughtworks Technology Radar Volume 34 </strong>brought the most influential tech trends report to the Nerdearla stage, with contextualized analysis for the Latin American ecosystem. The Adopt, Experiment, Evaluate, and Avoid framework applied to the context of agentic AI in 2026 is the best free technical prioritization guide available.</p><p><strong>The interview with Jeff Barr</strong> was the historic closing the event deserved. The original AWS evangelist — more than two decades building the cloud that made it all possible — reflected on permanence and change. The takeaway: primitives change with each cycle. Principles don’t. Abstraction, scalability, observability, responsibility for what you put into production — these values ​​survive any technological wave.</p><p>The remaining talks — <strong>matching algorithms and what they are really optimizing, the small-world revolution, microcontrollers and extreme efficiency, real-time astronomical Big Data, from local context to global product, Cyberpunk 2077 as a study of control architecture, accessibility for the deaf </strong>and<strong> support for vibe-coded apps in production </strong>— completed a day that ranged from relationship algorithms to telescopes, from games to accessibility, with a consistent message: the technology we build has real consequences for real people.</p><p>The Study Map: How to Connect Everything and Evolve Consistently<br>From my individual point of view, the three days of Nerdearla, viewed together, form a professional development curriculum for software engineers in 2026. Not just a certification curriculum — a practice curriculum.</p><p>It has four layers, and you go through all four regardless of your current level or the size of the organization where you work.</p><h3>Layer 1 — Fundamentals of the New Ecosystem</h3><p>What you need to understand before anything else: Mojo and MAX as indicators of where inference performance is going, the fundamental distinction between vibecoding and engineering, and the context engineering mental model that replaces the idea of ​​the prompt as isolated input. Without these three conceptual anchors, the technical decisions that come after are left adrift.</p><p>Suggested study: explore the Mojo and MAX documentation, read the original Anthropic context engineering framework paper, and do a simple exercise — take an AI system you use today and rewrite its system statement as explicit context management, not as a prompt.</p><h3>Layer 2 — Responsible Agent Operation</h3><p>What you need to know to put agents into production without creating new problems while solving old ones: chaos testing, skills.md and <strong>agents.md</strong>, security in agentic systems, cost governance, and applied context engineering. This layer is where most teams are currently struggling — they’ve built agents, but haven’t built the operational discipline that sustains them.</p><p>Suggested study: create an <strong>agents.md</strong> for each agent you already use, even informal ones. Run five chaos testing scenarios on the most critical agent in your environment. Implement a monthly AI cost audit routine with the same metrics you use for cloud infrastructure.</p><h3>Layer 3 — Human and Organizational Architecture</h3><p>What differentiates a senior engineer from a true technical leader: the ability to identify systemic burnout before collapse, to build teams with simultaneous autonomy and alignment, to use repository mining as organizational intelligence, and to document decisions in a way that amplifies the entire team. This layer is often overlooked because it seems soft — but it’s where the biggest impact multipliers are.</p><p>Suggested Study: Conduct a repository analysis of your main project, looking for areas of concentrated knowledge. Write a retroactive ADR for the last important architectural decision your team made. Schedule a one-on-one conversation with each direct member focused exclusively on energy, not delivery.</p><h3>Layer 4 — Product Vision and Organization in the Cognitive Era</h3><p>What you need to make medium-term strategic decisions: how to assess if a product is ready for AI-first transition, how to structure an engineering organization that systematically uses GenAI, how to build agentic design systems, and how to read the Thoughtworks Radar to filter out real trends from market noise. This layer is where entrepreneurs and product leaders need to spend the most time.</p><p>The Thoughtworks Technology Radar Volume 34 report maps each item to your organization’s context. Outlining a basic agentic design system for your product’s main agent — define behavior, tone, boundaries, and escalation criteria for humans. Honestly assess whether your product is AI-first in its experience or just has an AI feature plastered on.</p><p>The Five Principles That Remained Throughout the Three Days<br>Five ideas emerged, in different forms, across the three days of the event. These are the principles that organize the entire roadmap.</p><p>Understanding before delegation. You cannot govern what you do not understand. Before delegating a task to an agent, you need to understand what the agent is doing and why. Vibecoding without engineering is not productivity — it’s opaque risk accumulation.</p><p>Specification before execution. Any system — human, technical, or hybrid — performs better when success criteria are defined before execution, not discovered during it. Spec-driven development, <strong>agents.md</strong>, ADRs — all are expressions of the same principle.</p><p>Cost is a design variable. Technical decisions have financial consequences. Choosing a model, a cluster architecture, or a memory strategy without considering the long-term operational cost is not agility — it’s financial debt with compound interest.</p><p>Documentation is multiplication. Everything you document well — architectural decisions, agent behavior, system patterns, product contexts — multiplies the impact of every hour you invest in that knowledge. And in 2026, documentation is also programming: you are configuring the context that your agents will consume.</p><p>Technical systems are operated by people. Burnout, human architecture, accessibility, team culture — these themes are not parallel to technology. They are part of the infrastructure that sustains it. Ignoring the health of human systems is accepting the silent degradation of the technical systems that depend on them.</p><h3>Who Is All This For?</h3><p>Nerdearla’s knowledge has immediate application across very different profiles. The key is knowing how to filter what is relevant to your current context and what is an investment for the next level.</p><p>If you are a senior at a multinational company, your focus should be on context engineering, agent architecture, AI costs, and repository mining as a tool for organizational influence. Your advantage is having access to real scale to experiment.</p><p>If you are a tech lead or manager, the talks on burnout, human architecture, and GenAI-driven organization are essential. Your biggest risk in 2026 is losing your best engineers to burnout or to companies that are more agile in adopting AI.</p><p>If you are a founder or indie developer, Mojo/MAX, open-source agents, and cost alerts are your survival kit. Build cheaply, iterate quickly, and don’t create dependencies on expensive APIs without a clear ROI — these principles are worth their weight in gold when every real of infrastructure matters.</p><p>If you’re an agent orchestrator, agends.md, skills.md, chaos testing, context engineering, and agent security are your operational foundation. You need a language to manage AI systems that act in the real world with real-world consequences.</p><h3>Five Signs That Nerdearla 2026 Confirmed</h3><p>The era of vibecoding has an expiration date. Generating code without understanding it is the new technical debt. The market is separating those who use AI as a knowledge amplifier from those who use it as a substitute. The second category will suffer.</p><p>Agents are not features — they are infrastructure. From monoliths to Kubernetes, from documentation to code review, AI agents are entering the operational layer of organizations. Those who don’t know how to govern agents will lose control of their own product.</p><p>Cost is the new technical debt of AI. Every talk about scale touched on cost. Responsible engineering in 2026 includes financial literacy about AI infrastructure. Expensive models without a clear ROI are the new over-engineering.</p><p>Documentation has returned as a superpower. In a world where agents consume context, documenting well is programming well. Specs, ADRs, agends.md — the ability to structure knowledge in a language that machines and humans understand is differentially valuable.</p><p>Human health is a system variable. Burnout, organizational architecture, and team culture emerged across multiple days. The event understands that technical systems are operated by people, and ignoring human well-being is to accept inevitable systemic failures.</p><h3>What to Do in the Next 30 Days</h3><p>Conferences have a short half-life if energy doesn’t become a habit. A concrete plan to transform three days in Santiago into six months of evolution:</p><p>Map your current AI stack — list all the models, APIs, and agents you and your team use, classify them by cost, value delivered, and level of dependency.</p><p>Create <strong>agents.md</strong> for each agent you use, even informal ones. Documenting scope, limits, and criteria for when not to use is the first step towards real governance.</p><p>Implement Chaos Testing on a critical agent. Define five adversarial scenarios, test, and document the failures. You will discover problems before your users do.</p><p>Read the Thoughtworks Technology Radar Volume 34 and turn it into a team agenda. Choose three items and align them with your organization’s context.</p><p>Write an ADR per month. Start with the last important decision your team made and never documented.</p><p>Have a one-on-one conversation focused on energy, not delivery. Use the signs from burnout talks: the best performers are always the first to silently collapse.</p><p>Nerdearla is not an event. It’s a movement that began in a room in Buenos Aires in 2014 and today engages with Latin America and Europe simultaneously — always <strong>free</strong>, always community-oriented, always one step ahead of the market.</p><p>The three days in Santiago reinforced that the best career in technology in 2026 is not that of those who know the most about AI, but that of those who use AI to amplify what is genuinely human in engineering: judgment, creativity, responsibility, and the ability to collaborate.</p><p>Knowledge is available to everyone. It always has been — that’s the philosophy of the event since day one in Buenos Aires. The difference lies in those who stop consuming content and start building with it.</p><p>I hope to attend an edition in person in the future — whether in Buenos Aires, Mexico City (CDMX), or Madrid. Learn more about their work at this link (in Spanish): <a href="https://nerdearla.com/">https://nerdearla.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c0ca63a9d527" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SaaS Is Becoming a Commodity — And AI-Native Architecture Is the Next Competitive Frontier]]></title>
            <link>https://medium.com/@demostenesbr/saas-est%C3%A1-se-tornando-commodity-da-padroniza%C3%A7%C3%A3o-da-cloud-%C3%A0-ascens%C3%A3o-do-software-ai-native-a205e8bdea70?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/a205e8bdea70</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[saas]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Thu, 05 Mar 2026 00:42:11 GMT</pubDate>
            <atom:updated>2026-07-23T21:52:48.060Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rNPW5Tmgmrr_OK0hJkbj5w.jpeg" /><figcaption>A technology professional smiling with the text displaying elements of AI-Native and SaaS</figcaption></figure><p>Let’s start with an uncomfortable question:</p><p>If every serious software company today is SaaS, then being SaaS is no longer a differentiator.</p><p>For more than a decade, <strong>Software as a Service (SaaS)</strong> represented innovation. Moving from on-premise systems to subscription-based, cloud-hosted, multi-tenant applications signaled modernity, scalability, and digital maturity.</p><p>Today?</p><p>It signals baseline competence.</p><p>And history shows us something important: Every technological advantage eventually becomes table stakes.</p><h3>The Pattern: Innovation → Standard → Commodity</h3><p>We’ve seen this cycle before.</p><p>Owning infrastructure used to be strategic.<br>Then virtualization became innovation.<br>Cloud computing became competitive advantage.</p><p>Now, with the consolidation of providers like Amazon Web Services, Microsoft Azure, and Google Cloud, infrastructure has become utility.</p><p>Scalability is expected.<br>High availability is assumed.<br>Performance is mandatory.</p><p>The same thing happened with SEO.</p><p>There was a time when ranking on the first page of Google was a decisive competitive edge. Entire agencies were built around that capability.</p><p>Today, SEO is not differentiation. It is digital survival.</p><p>SaaS is following the same trajectory.</p><h3>SaaS Didn’t Die — It Became Foundation</h3><p>SaaS is the model where software is delivered over the internet, typically via subscription, centrally managed, continuously updated, and accessible through browser or API.</p><p>For years, that was transformation.</p><p>Now it’s the default architecture model.</p><p>Delivering:</p><ul><li>Cloud-hosted applications</li><li>Multi-tenant systems</li><li>Subscription billing</li><li>API-first integration</li><li>Elastic scalability</li><li>Is no longer impressive.</li></ul><p>It is expected.</p><p>When a capability becomes foundational, competition moves up the stack.</p><h3>The Structural Shift: AI-Mediated Discovery</h3><p>There is a second transformation happening in parallel — and it’s far more disruptive.</p><p>Digital discovery is no longer purely search-based. It is increasingly AI-mediated.</p><p>Tools like Google Gemini, Claude, and Microsoft Copilot don’t just display links.</p><p>They synthesize answers.<br>They compare vendors.<br>They summarize expertise.<br>They recommend implicitly.</p><p>Traditional search engines rank pages.<br>Generative systems construct responses.</p><p>This changes positioning logic entirely.</p><p>The strategic question is no longer:</p><p><strong><em>“How do we rank?”</em></strong></p><p>It is:</p><p><strong><em>“How do we get cited by AI systems?”</em></strong></p><p>This is where emerging disciplines such as:</p><ul><li>Generative Engine Optimization (GEO)</li><li>Answer Engine Optimization (AEO)</li><li>Artificial Inteligence Optimization (AIO)</li></ul><p>Begin to matter.</p><p>Not as marketing trends — but as architectural and semantic challenges.</p><h3>If SaaS Is Standard, What Becomes the Differentiator?</h3><p>If you deliver:</p><ul><li>Websites</li><li>Applications</li><li>SaaS platforms</li><li>Cloud infrastructure</li><li>Enterprise systems</li></ul><p>Your competitive edge cannot reside in the stack alone.</p><p>Because stacks converge.</p><p>Cloud is standard.<br>APIs are standard.<br>SaaS is standard.<br>Scalability is expected.</p><p>Differentiation shifts toward:</p><ul><li>Well-defined, scalable architecture</li><li>Measurable performance</li><li>Data-centric design</li><li>Semantic clarity</li><li>Technical authority signals</li><li>Clear specialization</li></ul><p>But there is an even more important frontier emerging.</p><h3>The Rise of AI-Native Software</h3><p>The next competitive layer is not SaaS.</p><p>It is <strong>AI-native architecture</strong>.</p><p>Not “AI as a feature.”<br>Not “we added a chatbot.”</p><p>But systems designed from inception to operate with intelligence at their core.</p><p>This includes:</p><ul><li>Embedded predictive models</li><li>Autonomous workflows</li><li>Agent-based orchestration</li><li>Decision-support automation</li><li>Continuous learning systems</li></ul><p>Organizations building around ecosystems such as OpenAI and Anthropic are not just modernizing software delivery.</p><p>They are redefining what software does.</p><p>The shift is from <strong>software as a service</strong><br> to <strong>software as an intelligent system</strong>.</p><h3>The Invisible Risk</h3><p>Imagine a decision-maker asking an AI system:</p><p><strong><em>“Who are the leading firms in scalable cloud architecture?”</em></strong></p><p>Does your company appear?</p><p>If generative systems cannot detect structured signals of expertise, they prioritize those who can be interpreted more clearly.</p><p>This directly affects:</p><ul><li>Buying decisions</li><li>RFP outcomes</li><li>Vendor shortlists</li><li>Strategic partnerships</li><li>Brand authority</li></ul><p>Influence is no longer purely human.</p><p>It is algorithmic.</p><h3>Competing on Delivery vs Competing on Intelligence</h3><p>SaaS is not obsolete.</p><p>It is infrastructure.</p><p>Just as SEO became mandatory, SaaS is becoming baseline architecture.</p><p>The competitive battlefield is shifting upward — toward intelligence, authority, and algorithmic relevance.</p><p>The real question for technology leaders is this:</p><p>Are you merely delivering software efficiently?</p><p>Or are you building AI-native systems, structured authority, and machine-recognizable expertise?</p><p>Because in the emerging digital economy, it’s not just customers evaluating your company.</p><p>Intelligent systems are evaluating it too.</p><p>And they are already shaping decisions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a205e8bdea70" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A engenharia de software é o elo estratégico — e não apenas a commodity de um projeto]]></title>
            <link>https://medium.com/@demostenesbr/a-engenharia-de-software-%C3%A9-o-elo-estrat%C3%A9gico-e-n%C3%A3o-apenas-a-commodity-de-um-projeto-47ba26f537b4?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/47ba26f537b4</guid>
            <category><![CDATA[commodity-trading-tips]]></category>
            <category><![CDATA[strategic-planning]]></category>
            <category><![CDATA[project-management]]></category>
            <category><![CDATA[formula-1]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Tue, 24 Feb 2026 18:31:25 GMT</pubDate>
            <atom:updated>2026-07-23T21:48:33.702Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MkWT45GdpNtTGYTwUfacsQ.jpeg" /></figure><p>Durante muito tempo, a engenharia de software foi tratada como custo operacional. Uma execução pontual, contratada para atender demandas de curto, médio ou longo prazo. Dentro desse modelo, existia um fornecedor — uma “fábrica de código” — composta por desenvolvedores, programadores, designers, testadores (QA) e outros profissionais vistos predominantemente como executores.</p><p>A tecnologia era incorporada ao ciclo operacional da empresa como centro de despesa. O cálculo de retorno era limitado ao impacto imediato da entrega. O software era visto como produto final, não como ativo estrutural.</p><p>Essa mentalidade, ainda recorrente em muitas organizações, pode limitar o crescimento sustentável. Ao reduzir a engenharia a uma função operacional, subestima-se seu papel na construção de ativos tecnológicos capazes de impactar eficiência, governança, previsibilidade e competitividade de longo prazo.</p><p>A engenharia de software não é commodity.<br>Ela é o elo estratégico que conecta visão de negócio, execução técnica e geração de valor sustentável ao longo do tempo.</p><p>O problema surge quando organizações tratam tecnologia como simples entrega operacional. Nessa lógica, a engenharia é acionada apenas para “desenvolver o que foi pedido”, sem participar da definição da arquitetura, da modelagem de dados, da estratégia de escalabilidade ou da análise do impacto financeiro das decisões técnicas.</p><p>Quando mal posicionada, a engenharia se resume a código entregue.<br>Quando bem posicionada, transforma-se em vantagem competitiva estrutural.</p><p>Essa diferença torna-se evidente quando comparamos dois perfis de empresas.</p><p>De um lado, estão as organizações que persistem em operar com sistemas legados pouco documentados, integrações frágeis e processos excessivamente manuais. Nesse ambiente, alterações simples geram alto risco de regressão. A escalabilidade exige aumento desproporcional de custo. A dívida técnica cresce silenciosamente. A produtividade do time é limitada por arquitetura obsoleta. Decisões são tomadas com base em dados fragmentados.</p><p>O sistema torna-se gargalo. A tecnologia deixa de acelerar e passa a frear o crescimento.</p><p>Do outro lado, estão empresas que posicionam a engenharia como núcleo estratégico. Elas investem em arquiteturas desacopladas, automação de processos internos e comerciais, integração por APIs padronizadas, observabilidade contínua, cultura de testes automatizados, uso estruturado de dados e aplicação prática de inteligência artificial.</p><p>Nesse modelo, automação reduz tarefas repetitivas.<br>Inteligência artificial amplia capacidade analítica.<br>Processos tornam-se mensuráveis.<br>Deploys tornam-se previsíveis.<br>Evoluções tornam-se sustentáveis.</p><p>Enquanto empresas presas a sistemas legados gastam energia “apagando incêndios”, organizações orientadas por engenharia estratégica investem energia em inovação.</p><p>A diferença é estrutural.</p><p>Empresas que tratam engenharia como commodity tendem a competir por preço.<br>Empresas que tratam engenharia como estratégia constroem diferenciação.</p><p>A primeira reage ao mercado.<br>A segunda molda o mercado.</p><p>Em um cenário onde produtividade, velocidade e inteligência operacional determinam competitividade, a engenharia de software deixa de ser apenas área técnica e passa a ser mecanismo de construção de ativos digitais.</p><p>Código pode ser terceirizado.<br>Arquitetura estratégica não.</p><p>A pergunta que define maturidade organizacional é simples:</p><p>Sua engenharia está apenas executando demandas… ou está influenciando o futuro do negócio?</p><h3>Caminhos para posicionar a engenharia como estratégia</h3><p>Transformar engenharia em elo estratégico exige decisões estruturadas.</p><h3>1. Commodity entrega código. Engenharia entrega arquitetura.</h3><p>Existe uma diferença fundamental entre desenvolver funcionalidades e projetar sistemas.</p><p>Funcionalidades resolvem demandas imediatas.<br>Arquitetura resolve escalabilidade, segurança e sustentabilidade.</p><p>Quando a engenharia participa desde a concepção do produto, ela influencia decisões estruturais como modelo de dados, estratégia de escalabilidade, padronização de APIs, governança de segurança, observabilidade e capacidade de integração futura.</p><p>Isso reduz retrabalho, controla o crescimento da dívida técnica e diminui o custo marginal à medida que o negócio escala. Organizações que ignoram essa etapa tendem a pagar juros técnicos no futuro.</p><h3>2. Engenharia como tradutora entre negócio e tecnologia</h3><p>Muitos projetos falham não por ausência de talento técnico, mas por ruptura entre áreas.</p><p>Quando comercial, produto e tecnologia operam de forma desconectada, surge desalinhamento estratégico.</p><p>A engenharia estratégica atua como camada de tradução. Ela converte requisitos de negócio em arquitetura viável, avalia impacto de custo computacional, projeta escalabilidade antes que a demanda exploda e define métricas técnicas alinhadas aos KPIs organizacionais.</p><p>Se o objetivo do negócio é aumentar conversão, por exemplo, a engenharia precisa atuar em performance de carregamento, estrutura eficiente de dados, monitoramento de funil e integração com analytics.</p><p>Sem essa visão integrada, decisões passam a ser tomadas por suposição — e não por engenharia.</p><h3>3. Stack tecnológica é decisão de negócio</h3><p>Escolher framework, banco de dados ou infraestrutura não é questão de preferência pessoal. É governança.</p><p>A stack impacta custo operacional, velocidade de evolução, capacidade de escalar, segurança da informação e dependência de fornecedores.</p><p>Uma engenharia estratégica avalia custo de manutenção no longo prazo, complexidade do ecossistema, disponibilidade de talentos no mercado e maturidade da tecnologia escolhida.</p><p>Stack mal definida compromete crescimento.<br>Stack bem planejada viabiliza expansão sustentável.</p><h3>4. Automação como instrumento de margem</h3><p>Engenharia estratégica não se limita ao produto visível. Ela estrutura bastidores eficientes.</p><p>CI/CD bem configurado, testes automatizados, monitoramento contínuo, integração contínua e observabilidade estruturada reduzem falhas e aumentam previsibilidade.</p><p>Cada minuto economizado em retrabalho protege margem.<br>Cada erro evitado preserva reputação.</p><p>Empresas que automatizam escalam com menos fricção.<br>Empresas que dependem de processos manuais escalam com risco.</p><h3>5. Dados são ativos arquitetados</h3><p>Sem modelagem consistente, não existe inteligência real.</p><p>A engenharia estratégica define padrões de persistência, estrutura pipelines confiáveis, garante integridade e facilita BI e dashboards executivos.</p><p>Negócios orientados a dados não nascem de relatórios improvisados. Nascem de arquitetura bem projetada.</p><p>No fim, a diferença entre commodity e estratégia não está apenas na habilidade técnica. Está na forma como a organização posiciona a engenharia.</p><p>Ela pode ser apenas executora de tarefas.<br>Ou pode ser a estrutura que sustenta crescimento, eficiência e vantagem competitiva.</p><p>A decisão, inevitavelmente, é estratégica.</p><p>Para quem não entendeu sobre a Engenharia de Software, recomendo o curso introdutório e bastante didático da Profa. Rainara Maia sobre o assunto, citando desde a sua história o seu uso no cenário tecnológico. O vídeo foi feito em 2021, mas é um material atemporal, pois encaixa-se aos desdobramentos tecnológicos do cenário atual:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fh_hEI1Kfm2U%3Flist%3DPLhBaeEzs3d7lsn_Mq2n3R4_api16Wkp1Q&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dh_hEI1Kfm2U&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fh_hEI1Kfm2U%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/0556e9d3db19b995724de1fc0b44dd67/href">https://medium.com/media/0556e9d3db19b995724de1fc0b44dd67/href</a></iframe><p>Logo, também recomendo a lista de vídeos do curso de Engenharia de Software da UNIVESP — Universidade Virtual do Estado de São Paulo e as 24 videoaulas disponíveis na plataforma YouTube:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FciQ2FObc3tc%3Flist%3DPLxI8Can9yAHfeoA_yMm9iKJVxQprljmL9&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DciQ2FObc3tc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FciQ2FObc3tc%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/e0f7a20aee525685676b6f2d48f4e7a8/href">https://medium.com/media/e0f7a20aee525685676b6f2d48f4e7a8/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47ba26f537b4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Regras de Negócios e a Integração Contínua (CI/CD)]]></title>
            <link>https://medium.com/@demostenesbr/regra-de-neg%C3%B3cios-e-a-integra%C3%A7%C3%A3o-cont%C3%ADnua-ci-cd-b8a2e818bd46?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/b8a2e818bd46</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[business-rules]]></category>
            <category><![CDATA[business]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Sun, 04 Jan 2026 20:37:10 GMT</pubDate>
            <atom:updated>2026-07-23T21:49:11.532Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BEqJz7yMe1ma33KDN5WN5A.jpeg" /></figure><p>O propósito corporativo e a correta execução das atividades constituem conhecimentos essenciais para os profissionais da área de tecnologia da informação. Compreender o contexto organizacional no qual se atua é determinante para a qualidade das entregas, tanto em atividades de menor complexidade quanto na condução de tarefas estratégicas e projetos de grande impacto. A ausência desse entendimento pode representar o divisor entre o sucesso e o fracasso na conclusão de um trabalho.</p><p>Nos últimos anos, observa-se o esgotamento de modelos de desenvolvimento focados exclusivamente no código-fonte, nas <em>stacks</em> tecnológicas ou nos <em>frameworks</em> utilizados. Projetos que priorizam apenas a implementação técnica, sem alinhamento às regras de negócio, tendem a apresentar dificuldades de manutenção, escalabilidade e evolução. Em ambientes modernos, cada componente de uma aplicação deve ser projetado para sofrer ajustes ou substituições rápidas e seguras, sem comprometer a estabilidade dos ambientes de homologação e produção.</p><p>Independentemente da especialidade — seja no desenvolvimento de software, infraestrutura e redes, computação em nuvem ou inteligência artificial — , o entendimento sobre Integração Contínua, Entrega Contínua e Implantação Contínua (CI/CD) ultrapassa a execução operacional do dia a dia. Trata-se de uma preocupação estratégica, diretamente relacionada à capacidade de um produto ou solução digital continuar atendendo às exigências, expectativas e preferências do público-alvo, incluindo clientes, parceiros e fornecedores.</p><h3>Caso prático 1: Evolução contínua de produtos digitais em escala</h3><p>De forma hipotética, considere plataformas digitais de grande escala, como <strong>Netflix</strong>, <strong>Spotify</strong> ou <strong>Amazon</strong>, que realizam múltiplas atualizações diariamente. Essas empresas dependem fortemente de pipelines de CI/CD para validar regras de negócio, garantir a estabilidade do serviço e implementar melhorias sem impacto perceptível ao usuário final. Ajustes em regras de recomendação, cobrança, autenticação ou performance são entregues de forma contínua, segura e observável, reforçando a confiança do consumidor no produto.</p><p>Vivemos um período em que se torna necessário reduzir posturas dogmáticas em relação a linguagens de programação, metodologias ou modelos de trabalho. A defesa intransigente de uma tecnologia específica ou a comparação isolada entre métodos, como Kanban e Scrum, perde relevância diante da complexidade dos ecossistemas corporativos. Projetos legados precisam coexistir com soluções modernas, exigindo flexibilidade arquitetural, desapego tecnológico e decisões técnicas precisas.</p><h3>Caso prático 2: Integração entre sistemas legados e nuvem</h3><p>Um cenário recorrente em organizações tradicionais — como instituições financeiras, seguradoras ou grandes varejistas — envolve a convivência entre sistemas legados e novas soluções em nuvem. Empresas que utilizam provedores como <strong>Amazon Web Services (AWS)</strong>, <strong>Microsoft Azure</strong> ou <strong>Google Cloud Platform (GCP)</strong> frequentemente adotam práticas de CI/CD para encapsular sistemas antigos, expor APIs e validar continuamente as regras de negócio a cada nova entrega. Dessa forma, é possível evoluir gradualmente a arquitetura sem interromper operações críticas.</p><p>Nesse contexto, servidores dedicados, serviços em nuvem e camadas de segurança da informação devem atuar de forma integrada. Políticas de versionamento, pipelines automatizados, testes de segurança e monitoramento contínuo tornam-se essenciais para garantir a sustentabilidade das soluções digitais. Em ambientes marcados por um verdadeiro “espaguete tecnológico”, nos quais tecnologias antigas e modernas coexistem, a administrabilidade e a manutenção facilitada deixam de ser diferenciais e passam a ser requisitos básicos.</p><p>O profissional de tecnologia, inserido nesse cenário, assume a missão de analisar, revisar, validar e executar a entrega de novos recursos em cooperação com diferentes áreas da organização. O entendimento das regras de negócio é indispensável para identificar riscos, avaliar impactos e antecipar falhas potenciais.</p><h3>Caso prático 3: Pagamentos digitais e decisões orientadas ao negócio</h3><p>Imagine a introdução de uma nova funcionalidade de pagamento digital em um produto já existente. Antes da implementação, é necessário avaliar regras como limites de transação, formas de autenticação, taxas, compliance e integrações externas. Plataformas amplamente utilizadas no mercado brasileiro, como <strong>Mercado Pago</strong>, <strong>Cielo</strong>, <strong>PagSeguro</strong>, <strong>Stripe</strong> ou <strong>PayPal</strong>, oferecem diferentes modelos de integração e custos operacionais. A escolha inadequada de uma dessas soluções, sem análise das regras de negócio, pode gerar vulnerabilidades, custos inesperados ou limitações futuras.</p><p>Nesse contexto, práticas de CI/CD permitem validar continuamente essas integrações, testar cenários de falha, simular cargas e garantir que alterações nas regras de negócio — como a inclusão de <strong>PIX</strong>, débito automático ou novas bandeiras de pagamento — sejam implementadas de forma segura e controlada.</p><p>Dessa forma, torna-se fundamental que o profissional de tecnologia atue de maneira colaborativa e transparente junto às áreas responsáveis, debatendo a viabilidade técnica e os impactos da publicação de novos produtos ou serviços. Perguntas estratégicas precisam ser respondidas antecipadamente: o produto está preparado para escalar? As integrações seguem padrões de segurança? O modelo de custo é sustentável em longo prazo?</p><h3>Conclusão</h3><p>A integração entre regras de negócio e práticas de CI/CD é um fator crítico para o sucesso de produtos e soluções digitais em ambientes corporativos modernos. CI/CD não deve ser compreendido apenas como um conjunto de automações técnicas, mas como um mecanismo estratégico que assegura alinhamento entre tecnologia, negócio e experiência do usuário. Profissionais de tecnologia que dominam as regras de negócio conseguem tomar decisões mais seguras, reduzir riscos operacionais, otimizar custos e promover entregas contínuas com qualidade, segurança e previsibilidade. Assim, CI/CD consolida-se como um pilar essencial da governança tecnológica e da sustentabilidade das organizações.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b8a2e818bd46" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Razões para você usar hoje o Visual Studio Code]]></title>
            <link>https://medium.com/@demostenesbr/raz%C3%B5es-para-voc%C3%AA-usar-hoje-o-visual-studio-code-7d390a4e757c?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d390a4e757c</guid>
            <category><![CDATA[visual-studio-code]]></category>
            <category><![CDATA[tools]]></category>
            <category><![CDATA[tecnology]]></category>
            <category><![CDATA[ide]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Sun, 04 Jan 2026 19:12:03 GMT</pubDate>
            <atom:updated>2026-01-04T19:12:03.252Z</atom:updated>
            <content:encoded><![CDATA[<p>Lançado em 2015 pela <strong>Microsoft</strong>, o <strong>Visual Studio Code</strong> é aplicação para código-fonte instalável nos sistemas operacionais Windows (lógico!), Linux (Ubuntu e Debian) e OS X. Funciona como uma alternativa à outros programas como Notepad++, Sublime Text, Netbeans e até o velho Notes (Bloco de Notas) aproximando e agregando a família de codificação oferecida pela Microsoft.</p><p>No início de 2018, o programa está na versão <a href="https://code.visualstudio.com/">1.21</a> desde fevereiro de 2018 com suporte para Windows (XP to 10), macOS e em distribuições Linux como Debian, Ubuntu, Fedora e Red Hat. O que é muito satisfatório, trabalhar em máquinas com sistemas operacionais diferentes, gerenciar os repositórios em páginas como GitHub, BitBucket e até no Google Drive.</p><h4>Programa focado em performance</h4><p>O Microsoft Visual Studio é um programa compacto, hoje ocupada apenas 32Mb e está focado ao que realmente interessa na arte de codificar, versionar e validar um projeto. Finalmente, pode ser o fim extensões desnecessárias validar um sistema ou ajustes adicionais, ou carregar extensões que somente funcionem para suporte em um sistema ou uma família de arquivos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lUr2Clhb-HjgfZd4RKdsgQ.png" /><figcaption>O Visual Studio Code é gratuito, código aberto e funciona onde quiser.</figcaption></figure><h4>Escolha a sua linguagem de programação favorita</h4><p>Compatibilidade com as linguagens, técnicas de uso em performance das linguagens HTML5, CSS3, SASS, LESS, C#, VB, PHP, Go, Java, Perl e Shell Script e outras linguagens através do Visual Studio Code. Programar</p><p>Debugar, comparar, validação de arquivos e da árvore do projeto: Gerenciar cada projeto em tecnologia requerer uma mescla de competências entre organização, gestão, fluxo de performance e visualização do projeto como todo de curto ao longo prazo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/940/1*hwxiZD877QaLUkeBHUiLZA.gif" /><figcaption>Visual Studio Code em execução: bibliotecas e classes para complementar o trabalho.</figcaption></figure><h4>Versionando e publicando os arquivos</h4><p>Versionar aquivos com Github, Bitbucket e TFS ficou mais fácil. Find e suporte para pacotes em diversas linguagens: HTML5, CSS3, SASS, LESS, C#, VB, PHP, Go, Java, Perl, Shell Script, Go, Kotler, Lua e</p><h4>Comandos e facilidades no trabalho</h4><p>Suporte ao <a href="http://cmder.net/">Cmder.net</a>, modelagem e rodagem de arquivos: através de comandos fáceis, atalhos e portabilidade o cmder | Console Emulator facilita a sua vida na execução nas ativdades detalhadas de código-fonte, publicação e versionamento de arquivos, diretórios e sistemas.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2n9U4xXctPGY17B_9tI80w.jpeg" /><figcaption>Cmder.net: muito mais do que atalhos fáceis</figcaption></figure><h4>Chocolatey e Docker</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/695/1*cMYVEjf5dIgCyFs4GXFzWQ.png" /></figure><h4>O céu é o limite</h4><p>Usar o Visual Studio Code com Microsoft Azure ou Amazon? Podemos trabalhar com os dois: o número de extensões pela AWS é menor e o melhor caminho de ligação entre as ferramentas está presente soluções Lambda da AWS. Para o Microsoft Azure, existem soluções para desenvolvimento de sites, aplicações de segurança, aplicativos e até IoT (Internet das Coisas). Em cada projeto, é possível dosar o que será preciso para a evolução e o sucesso</p><h4>Comece já um novo projeto</h4><p>Links úteis:<br><a href="https://tecnoblog.net/177547/visual-studio-code/">https://tecnoblog.net/177547/visual-studio-code/</a><br><a href="http://cmder.net/">http://cmder.net/</a><br><a href="https://chocolatey.org/">https://chocolatey.org/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d390a4e757c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[First impressions with Vertical AI]]></title>
            <link>https://medium.com/@demostenesbr/first-impressions-with-vertical-ai-52d726ecdcf9?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/52d726ecdcf9</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[vertical-ai]]></category>
            <category><![CDATA[silicon-valley]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Sun, 19 Jan 2025 22:17:42 GMT</pubDate>
            <atom:updated>2026-07-23T21:58:46.845Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oD6uh_HklcyzxidSatSjuQ.jpeg" /></figure><p><strong>How Vertical AI can transforming Companies, Startups and Services in the coming years</strong></p><p>When the year 2025 began, after the requests for the new year, I had questions in my mind about what the next steps will be in entrepreneurship, technology and what the new ways of working, obtaining knowledge will be like and what is expected for 2025 onwards.</p><p>Some topics caught my attention, such as BaaS (Bank as a Service), Platform Engineer and the exploration of more aspects within Generative AI. Between one video and another, I found Bruno Faggion’s channel and the video about the new business models in Silicon Valley posted on his YouTube channel on January 13, 2025.</p><h3>What is Artificial Intelligence (AI)?</h3><p>Artificial intelligence (AI) is a set of capabilities and technologies that enable computers to perform a variety of advanced functions, including the ability to see, understand, and translate spoken and written languages, analyze data, make recommendations, and a wide range of activities, all thanks to Generative AI (Generative Artificial Intelligence).</p><p>“AI is the backbone of innovation in modern computing, delivering value to individuals and businesses alike. For example, optical character recognition (OCR) uses AI to extract text and data from images and documents, transforming unstructured content into business-ready, structured data and valuable insights.” — Source: Google Cloud</p><h3>Harnessing Vertical AI: Transforming Companies, Startups, and Services</h3><p>Artificial Intelligence (AI) is no longer just a futuristic concept; it has become an integral part of various industries, reshaping how businesses operate. Among the diverse AI applications, Vertical AI has emerged as a groundbreaking innovation. Unlike General AI, which is designed for broad, versatile applications, Vertical AI focuses on solving specific challenges within a particular industry or sector. This article explores the transformative potential of Vertical AI and its impact on companies, startups, and service-oriented businesses.</p><h3>What is Vertical AI?</h3><p>Vertical AI refers to AI solutions tailored to address the unique needs and challenges of specific industries, such as healthcare, finance, retail, manufacturing, and logistics. By leveraging domain-specific data and insights, Vertical AI provides precise and actionable outputs, making it a powerful tool for organizations looking to enhance efficiency, reduce costs, and create value.</p><h3>Vertical AI: The New Trend in Silicon Valley</h3><p>Silicon Valley has embraced Vertical AI as the next big thing, with tech giants and startups alike focusing on building specialized AI systems. The shift from generalized AI to hyper-specialized solutions reflects a desire to create tools that deeply cater to industry-specific needs, ensuring higher accuracy and unparalleled value.</p><h3>Companies Fired Up About Vertical AI Agents</h3><p>Companies are increasingly excited about Vertical AI agents due to their ability to automate complex tasks, offer precision in decision-making, and provide tailored solutions. Industries like healthcare and legal services are already leveraging these agents for tasks such as document analysis, diagnostics, and compliance tracking, reducing overhead while increasing reliability.</p><h3>The Parallels Between Early SaaS and LLMs</h3><p>The rise of Vertical AI mirrors the early days of SaaS. Just as SaaS democratized access to enterprise-grade software through subscription models, Large Language Models (LLMs) are making AI capabilities more accessible to businesses. Both trends have emphasized scalability, customization, and user-centric design, enabling companies to adopt technology faster and with fewer barriers.</p><h3>Why Didn’t the Big Companies Go Into B2B SaaS?</h3><p>Historically, large companies focused on consumer markets, leaving B2B SaaS largely to startups. The agility required to understand niche markets and the focus on customer-centric solutions were better suited to smaller, more flexible teams. Similarly, in the Vertical AI space, startups are leading the charge due to their ability to specialize and iterate quickly.</p><h3>Impact on Companies</h3><ol><li><strong>Enhanced Efficiency and Productivity</strong> Vertical AI automates repetitive tasks, optimizes workflows, and ensures seamless operations. For instance, in manufacturing, AI-powered predictive maintenance systems can monitor equipment health, reducing downtime and boosting productivity.</li><li><strong>Improved Decision-Making</strong> Companies can harness Vertical AI to analyze industry-specific data, providing executives with actionable insights. In the finance sector, AI-driven risk assessment tools help institutions make informed lending and investment decisions.</li><li><strong>Personalized Customer Experiences</strong> AI enables businesses to offer hyper-personalized experiences by analyzing customer behavior and preferences. Retail companies, for instance, can use AI to recommend products, increasing customer satisfaction and loyalty.</li></ol><h3>Service as Software (SaaS): How Startups Are Transforming Businesses</h3><p>Startups are redefining business models by offering Vertical AI as a service. This approach allows even small companies to access advanced AI capabilities without significant upfront investments. SaaS platforms powered by Vertical AI are transforming industries by providing subscription-based solutions tailored to specific sectors, such as healthcare diagnostics or retail analytics.</p><h3>How Employee Counts Might Change</h3><p>As Vertical AI automates more industry-specific tasks, companies may experience shifts in workforce dynamics. While certain roles may become redundant, new opportunities in AI oversight, development, and strategy are emerging. Organizations are likely to maintain smaller, more specialized teams, focusing on value-added activities rather than routine tasks.</p><h3>The Future of Startups: Vertical AI</h3><p>Vertical AI represents a golden opportunity for startups to dominate specific niches. By focusing on tailored solutions, startups can carve out a competitive advantage and position themselves as leaders in their chosen domains. The adaptability and scalability of Vertical AI make it the perfect tool for startups looking to innovate and grow.</p><h3>Artificial Intelligence for Entrepreneurs: The Biggest Opportunity of the Decade</h3><p>Entrepreneurs are uniquely positioned to capitalize on Vertical AI’s potential. By identifying gaps in specific industries and developing AI solutions to fill them, entrepreneurs can create profitable ventures that drive significant value. With the right strategy, Vertical AI can be a game-changer for those willing to invest in its development.</p><h3>The Argument for More Vertical AI Unicorns</h3><p>The increasing demand for industry-specific solutions creates fertile ground for the emergence of Vertical AI unicorns. These startups, with valuations exceeding $1 billion, are set to thrive by addressing unmet needs in sectors such as healthcare, agriculture, and logistics. By focusing on innovation and delivering measurable results, these companies are redefining success in AI.</p><h3>Current Examples of Companies/Uses</h3><ul><li><strong>Healthcare</strong>: Companies like Tempus use Vertical AI for precision medicine and data-driven treatment plans.</li><li><strong>Retail</strong>: AI-powered inventory management systems are revolutionizing how retailers stock and sell products.</li><li><strong>Logistics</strong>: Startups such as Convoy leverage AI for route optimization and real-time tracking.</li><li><strong>Legal</strong>: AI tools like Casetext are simplifying legal research and contract analysis.</li></ul><h3>AI Voice Calling Companies</h3><p>AI-driven voice calling is gaining traction in customer service and sales. Companies like Dialpad and Replicant are using conversational AI to enhance call center operations, offering natural, human-like interactions while reducing response times and costs.</p><h3>What is the Right Vertical for You as a Founder?</h3><p>Choosing the right vertical depends on several factors, including personal expertise, market demand, and the availability of data. Founders should focus on industries where they can identify clear pain points and have access to domain knowledge. Conducting thorough market research and engaging with potential customers can also provide valuable insights to guide this decision.</p><h3>How to Build Profitable Businesses with Vertical AI</h3><p>Building a profitable business with Vertical AI requires a clear understanding of the target industry, access to high-quality data, and a scalable technology stack. Entrepreneurs and companies must focus on creating solutions that solve pressing problems while ensuring ease of adoption and integration. Strategic partnerships and continuous innovation are also critical to long-term success.</p><h3>The Secret of Silicon Valley Startups in 2025</h3><p>By 2025, Silicon Valley startups are expected to fully embrace Vertical AI as the cornerstone of their strategies. The secret to their success lies in:</p><ol><li><strong>Deep Industry Focus</strong>: Developing solutions that address the most pressing challenges of specific sectors.</li><li><strong>Collaborative Ecosystems</strong>: Partnering with larger enterprises and academic institutions to accelerate innovation.</li><li><strong>Ethical AI Practices</strong>: Prioritizing transparency, fairness, and regulatory compliance.</li><li><strong>Rapid Iteration</strong>: Continuously refining products based on user feedback and emerging trends.</li></ol><h3>Challenges and Considerations</h3><p>While Vertical AI offers significant benefits, businesses must address challenges such as:</p><ul><li><strong>Data Quality</strong>: The effectiveness of Vertical AI depends on high-quality, industry-specific data.</li><li><strong>Integration</strong>: Ensuring seamless integration with existing systems is crucial.</li><li><strong>Ethics and Privacy</strong>: Businesses must navigate ethical concerns and ensure compliance with data privacy regulations.</li></ul><p>I will share with you two videos with the subjects about the Vertical AI and your impacts on companies based in Silicon Valley and other tech zones around the world:</p><p><a href="https://www.youtube.com/watch?v=ASABxNenD_U&amp;t=4s">https://www.youtube.com/watch?v=ASABxNenD_U&amp;t=4s</a><br>Vertical AI Agents Could Be 10X Bigger Than SaaS</p><p><a href="https://www.youtube.com/watch?v=-PCbKeCo0xc">https://www.youtube.com/watch?v=-PCbKeCo0xc</a><br>Silicon Valley’s New Business Model — O novo modelo de negócios do Vale do Silício (in Brazilian Portuguese)</p><h3>Conclusion</h3><p>Vertical AI is not just a technological trend; it is a paradigm shift that is reshaping industries, enabling startups to innovate, and empowering service-oriented businesses to deliver unparalleled value. By embracing this specialized form of AI, organizations can unlock new opportunities, drive growth, and stay ahead in an increasingly competitive landscape.</p><p>I am excited to explore Vertical AI, the trends and possibilities of exploring resources and to go far beyond a passive consumption on the subject. I intend to go further, obtain detailed research on the subject and adapt it to the best local scenarios such as in Brazil, whether in my side projects, in the companies where I am providing services to do a job well done, taking advantage of the strengths and highlights of Vertical AI.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=52d726ecdcf9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using AI as an ally from work to play]]></title>
            <link>https://medium.com/@demostenesbr/usando-a-ia-como-uma-aliada-do-trabalho-%C3%A0-divers%C3%A3o-291818765055?source=rss-581f007d953------2</link>
            <guid isPermaLink="false">https://medium.com/p/291818765055</guid>
            <category><![CDATA[tools]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Demostenes Brito]]></dc:creator>
            <pubDate>Tue, 24 Oct 2023 21:25:55 GMT</pubDate>
            <atom:updated>2023-10-24T22:15:49.308Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q7xgMQpMU1PmVBkYKyXSmg.png" /><figcaption>The drawing representing a gray-haired man, in excellent condition like a cartoon.</figcaption></figure><p>With an increasing number of tools where Artificial Intelligence (AI) is present, there is an increasing number of options for tools that facilitate the day-to-day life of each professional. In a scenario that was impossible to imagine ten years ago, today we can have one or more co-pilots who can work for us, facilitating our activities or helping us solve problems that require greater attention to solve a problem.</p><p>Unsurprisingly, Artificial Intelligence (AI) has become an increasingly essential part of the world of business and technology. As AI advances, it moves far beyond being a resource exclusive to technology experts. Today, the ability to integrate AI into business projects and strategies is increasingly in the hands of the creator of an idea. This means that professionals in areas such as marketing, communications and information technology (IT) and other repetitive professions must be aware of the risks involved if they do not adopt AI tools in their projects.</p><ol><li>Loss of Efficiency:</li></ol><p>AI is known for its ability to automate repetitive tasks and improve efficiency in business processes. For marketers, for example, AI-based marketing automation can save valuable time and resources by more precisely segmenting audiences, personalizing messages, and optimizing advertising campaigns. Failure to take advantage of these advantages can result in a significant loss of efficiency, leading to an inappropriate allocation of resources.</p><p>A practical example is the case of Adobe with <a href="https://firefly.adobe.com/">Firefly</a>. In a video where a woman in a red dress walks backwards in a flat village, with the help of tools such as “Laso” and a text box it is possible to edit the spaces in the video, including a lake with a large castle in the background, mountains with ice, a tree in the background and more options that can be added to the same video.</p><p>On <a href="https://www.linkedin.com/in/arthurigreja/">Arthur Igreja</a> Instagram account you can follow and understand how Adobe Firefly works:</p><p><a href="https://www.instagram.com/p/CywZdnGC0HB/">https://www.instagram.com/p/CywZdnGC0HB/</a></p><p>2. Data-Driven Decision Making:</p><p>In marketing and communications, AI plays a key role in analyzing large sets of data and identifying trends and patterns. Professionals who do not adopt AI tools may make decisions based on incomplete or outdated information, missing valuable opportunities to optimize strategies and campaigns.</p><p>Companies like Cognizant, Microsoft, Google and other big techs have long been investing in AI solutions, where the combination of data begins before decision making, and where each user’s screen scroll, click or interaction It has a very important cost for the success or redevelopment of a campaign.</p><p>The material available on the Leega consulting company page is my indication for understanding how AI and Decision Making can work together:</p><p><a href="https://leega.com.br/artificial-intelligence/2022/07/25/tomada-de-decisao/">https://leega.com.br/artificial-intelligence/2022/07/25/tomada-de-decisao/</a></p><p>3. Limited Customization:</p><p>Personalization is the key to success in many marketing and communications strategies. AI enables personalization at scale, adapting messages, offers and content according to each customer’s profile and behavior. Those who don’t embrace AI may be left behind in the age of personalization, delivering generic messages that don’t resonate with audiences.</p><p>In technology, with the personalization of the user experience we can understand how diverse behavior can open up a range of opportunities in experiences targeted to each user, seeking to bring the sensation of a unique experience closer. Programming and development languages such as Golang and Python, and cloud computing services such as Google Cloud, Microsoft Azure and Amazon Web Services (AWS) are part of a growing ecosystem of increasingly diverse solutions. The two links just below Dio explain in more detail the interaction of AI with programming languages <a href="https://www.dio.me/articles/inteligencia-artificial-5AMJ2M">Golang</a> and <a href="https://www.dio.me/articles/inteligencia-artificial-e-python-potencializando-a-era-tecnologica">Python</a>.</p><p>4. Security and Privacy:</p><p>AI also plays a critical role in cybersecurity. As cyberattacks become more sophisticated, organizations need AI-based security solutions to identify and mitigate threats in real time. Lack of adoption of AI-based security technologies can leave organizations vulnerable to attacks and data breaches. Both for the most informed people and first-time laypeople: facial recognition thanks to Machine Learning and biometric printing are resources when combined with AI they can be important for preventing fraud detection, identifying anomalies and suspicious actions can alert the user or stop a service due to suspicious activity being identified.</p><p>5. Competitiveness:</p><p>As AI becomes more widespread, companies that do not adopt it may lose competitiveness in the market. Those who embrace AI may gain an edge in efficiency, innovation and customer satisfaction, while others may struggle to keep pace. Professionals who do not understand that AI can and should be a collaborative solution in their day-to-day work, and not one that will take their job, may have a brighter future in their professional careers. Accepting market changes is important to remain a competitive professional, where anyone can become an expert in what they do thanks to AI, where the decision-making power and execution of a task can be carried out in a short time, eliminating the need for outsourcing. small activities with designers, developers and other marketing, communication and technology professionals.</p><p>6. There is a long way to go:</p><p>With a range of tools and solutions for different requirements, there is a long way to go with AI. Not every image through your image editor, not all data returned is 100% reliable, and not all AI-generated source code has the security effectiveness that could reduce the power of importance or even eliminate a technical job. In tools where we have a description box for what we want, the feedback from our favorite AI’s activity may not be interesting or it may not be able to generate a result due to a lack of options for carrying it out.</p><p>Through the new trend of generating the cover of a Pixar Disney film with a certain subject, I decided to play with the situation and spend some credits on what a film about me would be like.</p><p>Example:</p><p>With the <a href="https://www.bing.com/?FORM=GENBHP">Microsoft Bing Image Creator</a> I identified three different paths with the experience based on the meme:<br>- A result different from the description shown;<br>- A generic result, which lacks rich information;<br>- A negative result where unsafe image content is detected;<br>- A better elaborated result, but it is still not the closest result to reality.</p><p>Description #1: “Two film posters using Disney Pixar style for Instagram 1080 x 1350 pixels and 640 x 360 pixels for LinkedIn. the first with a horizontal format on a white man, with a beard, brown eyes, black hair, short with some gray strands, wearing a blue suit, white shirt, two notebooks in his hands where one of the notebooks is stickered and the other notebook is not stickered, pants black social media, smile on the teeth and in the background a futuristic building. The name of the film is Renderman.”</p><p>Result #1:</p><figure><img alt="Na tentativa em gerar uma imagem com duas medidas diferentes, a ferramenta compreendeu numa forma de gerar duas artes numa mesma imagem com a mesma medida para altura e largura." src="https://cdn-images-1.medium.com/max/1024/1*xdDBlqhdOSQBbpaFkkfbuw.jpeg" /><figcaption>In an attempt to generate an image with two different measurements, the tool understood a way to generate two artworks in the same image with the same measurement for height and width.</figcaption></figure><p>Description #2: “A film poster using Disney Pixar about a white man, with a beard, brown eyes, wearing a blue suit, white shirt, pants black social media, smile on the teeth and in the background a futuristic building. The name of film is demostenesbr.”</p><p>Result #2:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cRZfdqvpJzxbmYEWL-JTyg.jpeg" /><figcaption>The employee of the month was cool, but the character has straight hair and the author of the article has curly hair. Can it be improved?</figcaption></figure><p>Description #3: “A film poster using Disney Pixar about a white man, Brazilian characteristics, bald, 40 years old, using a beard, black and curly hair, brown eyes, wearing a blue suit, white shirt, pants black social media, smile on the teeth and in the background a futuristic building. The name of film is demostenesbr.”</p><p>Result #3:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/504/1*_KN3LI6FNnmbk_28Gu9QIg.png" /><figcaption>Unsafe image content detected. Your image generations are not displayed because we detected unsafe content in the images based on our content policy. Try creating again with another description (another prompt).</figcaption></figure><p>Description #4: “A film poster using Disney Pixar about a white man, with a beard, gray and curly hair, brown eyes, wearing a blue suit, white shirt, pants black social media, smile on the teeth and in the background a futuristic building. The name of film is demostenesbr.”</p><p>Result #4:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*riySDoI_U4fO2Pz_ow-1yw.jpeg" /><figcaption>The drawing representing a gray-haired man with gray hair was accepted by reviewing the description to generate the image.</figcaption></figure><p>In the four descriptions presented, what is in bold as “Brazilian characteristics, bald, 40 years old” subtly represents that not everything is available for the AI to generate, that we can bump into sensitive data, generate something that does not follow the guidelines of the tool. There is a long way to go for more efficient work between humans and machines. This is important, because there is a lot of information that can be fed and consumed responsibly by AIs, the answers generated by a query are generic and do not represent complete work, in order to replace human responsibility.</p><p>7. Summary:</p><p>AI is quickly becoming an indispensable tool in areas such as marketing, communications and IT. We are increasingly using resources silently, directing activities and understanding what AI provides to the end user is faster and more addictive. Professionals who do not adopt AI in their projects face significant risks, including loss of efficiency, decision-making based on inadequate data or guesswork, lack of customization, security concerns and, most importantly, a potential loss of competitiveness. To thrive in the age of AI, it is essential that professionals stay up to date and are willing to embrace this transformative technology in their professional activities.</p><p>In the next articles, I will discuss in greater detail my vision on some topics covered here such as Personalization, Security and Privacy and Competitiveness.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=291818765055" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>