From Vibe Coding to Martech’s Reinvention (Part 7)

Martech’s Transformation

For B2C companies drowning in revenue taxes and AdWaste—those hidden costs that silently erode profitability—vibe coding promises not just new tools but a fundamental reimagining of the marketing technology stack. The future isn’t about choosing between human creativity and AI efficiency; it’s about orchestrating their convergence to finally solve the problems that have plagued digital marketing for decades.

To make this happen, here is what my colleague Chirag Patnaik recommends: “Traditional enterprise platforms are built as monolithic experiences optimized for human navigation. The AI-native approach demands service decomposition—breaking complex platforms into discrete, autonomous capabilities. Instead of offering “marketing automation software,” forward-thinking companies should provide: customer data infrastructure accessible via clean APIs, message delivery services for email, SMS, and messaging platforms, segmentation and personalization engines that can be called programmatically, analytics and reporting systems that return structured data, not visual charts, and campaign orchestration tools that can be automated without human intervention. This isn’t about building better APIs—it’s about fundamentally restructuring how services are architected and exposed… The companies that thrive in the next decade won’t be those with the most polished interfaces, but those whose services are most seamlessly accessible to AI agents through intelligent orchestration layers. The transition from human-navigated software to AI-orchestrated infrastructure has begun.”

In the AI-native world, martech must unbundle into composable services exposed via APIs—so agents orchestrate, not humans navigate. This architectural shift unlocks three fundamental transformations in how marketing technology operates.

  1. From Dashboards to Dialogues

The marketing platforms of yesterday forced users into predetermined workflows, clicking through endless menus and dashboards to find the insights they needed. Tomorrow’s platforms will be conversational, intuitive, and infinitely flexible.

Imagine a marketing manager starting their day not by logging into multiple dashboards but by having a conversation: “Show me yesterday’s campaign performance, but focus on the segments that underperformed expectations. For each underperforming segment, suggest three alternative approaches based on what’s worked for similar cohorts. And prepare A/B tests for the top recommendation in each case.”

Within seconds, the platform responds with visualisations, insights, and ready-to-launch experiments. No clicking through reports. No manual segment creation. No complex test setup. Just outcomes delivered through natural dialogue.

This interface revolution goes deeper than convenience. It democratises advanced marketing capabilities. A junior marketer can now access the same sophisticated analyses that previously required years of platform expertise. The conversation becomes the interface, and the interface adapts to each user’s needs and expertise level.

The MCP (Model Context Protocol) Advantage

Behind this conversational interface, MCP servers maintain context across sessions, learning from every interaction. They remember that this particular brand manager always wants to see mobile-first metrics, that the CMO prefers visual summaries over tables, that the retention team needs cohort analyses defaulted to 30-day windows.

This isn’t just personalisation—it’s the platform becoming an extension of each user’s thought process. The system doesn’t just respond to requests; it anticipates needs, suggests opportunities, and alerts users to anomalies they haven’t even thought to look for.

Vibe-Coding Custom Workflows

But here’s where it gets revolutionary: marketers can now vibe-code their own workflows directly within the platform: “Create a custom workflow that monitors our top 100 customers’ engagement scores daily. If anyone’s score drops by more than 20%, automatically trigger a personalised win-back campaign using their purchase history, but exclude anyone who’s received a promotional email in the last 48 hours. Also, alert their account manager on Slack.”

The platform doesn’t just execute this—it builds it as a reusable component that can be shared, modified, and improved by the entire team. Marketing operations transforms from configuration to conversation.

  1. The Age of Marketing Agents

The next evolution goes beyond interface changes to fundamental capability expansion through dual-agent architectures that operate at superhuman scale and precision.

Marketing Operations Agents

These agents don’t just automate tasks—they multiply human capability by orders of magnitude. Where a human marketer might manage 10 customer segments, agents manage 10,000 micro-segments simultaneously, each with its own strategy, budget, and performance targets.

Consider campaign optimisation. Today’s marketer might A/B test subject lines. Tomorrow’s agent simultaneously tests hundreds of variations across thousands of micro-segments, learning and adapting in real-time. It’s not replacing human creativity—it’s amplifying it to previously impossible scales.

These agents work 24/7, constantly monitoring, adjusting, and optimising. They catch anomalies before they become problems, identify opportunities before competitors, and execute complex strategies that would require armies of human marketers.

BrandTwins: Customer Representative Agents

The truly revolutionary concept is BrandTwins—AI agents that represent individual customers, learning their preferences, predicting their needs, and engaging with brand communications on their behalf.

Imagine every customer having a digital twin that:

  • Filters marketing messages to surface only truly relevant offers
  • Negotiates personalised deals based on lifetime value
  • Provides feedback to brands about product preferences
  • Manages subscription preferences intelligently

This isn’t science fiction—it’s the logical evolution of customer empowerment. And forward-thinking platforms like Netcore are preparing for this reality by building systems that can interact with millions of customer agents simultaneously.

The TwinLedger Revolution

Perhaps most transformative is the TwinLedger concept—maintaining a P&L for every single customer. Not segments. Not cohorts. Individual customers.

Every customer interaction, every marketing dollar spent, every purchase made gets recorded in their personal ledger. The system knows precisely:

  • Customer Acquisition Cost (and reacquisition costs)
  • Lifetime Value (updated in real-time)
  • Profitability trajectory
  • Optimal investment levels for retention

This granular financial visibility enables true 1:1 marketing economics. The platform can automatically adjust spend based on individual customer profitability, ensuring marketing efficiency at a level never before possible.

  1. Partners, Not Vendors

The fundamental relationship between Martech providers and their customers must evolve. The old model—charging for seats and features regardless of outcomes—is dying. The new model aligns vendor success with customer success.

Outcome-Based Pricing

Instead of charging $50,000 per year for platform access, imagine pricing based on:

  • Revenue generated through the platform
  • Cost savings achieved through automation
  • Improvement in customer lifetime value
  • Reduction in customer acquisition costs

This isn’t just fairer pricing—it’s aligned incentives. When martech companies only succeed if its customers succeed, the entire relationship dynamic changes. The platform becomes a true partner in growth, not just a cost centre.

This shift toward outcome-based pricing aligns closely with ideas we’ve been pioneering at Netcore—ZeroBase, where brands pay only for incremental outcomes, and Progency, which fuses product and AI-powered services into shared-success models. Both reframe martech not as software cost, but as a profit engine.

Solving the Real Problems: Revenue Taxes and AdWaste

B2C companies face two critical challenges that traditional Martech hasn’t solved:

Revenue Taxes: The 15-20% of revenue paid to Google and Meta for customer acquisition, with diminishing returns as costs rise and effectiveness falls. Vibe-coded solutions can help by:

  • Creating alternative acquisition channels
  • Optimising spend allocation with AI precision
  • Building owned media assets that reduce dependence on paid channels

AdWaste: The billions spent on advertising that doesn’t convert, doesn’t engage, or worse, annoys potential customers. The solution isn’t better targeting—it’s fundamental restructuring of how brands communicate with customers.

The vision for martech companies like Netcore should address both by:

  • Shifting focus from acquisition to retention (where economics are better)
  • Enabling true 1:1 communication that eliminates waste
  • Building direct customer relationships that bypass expensive intermediaries

Thinks 1734

FT: “How many grammes of protein do we require anyway? Does everybody need to eat like Superman? In the US, the Recommended Dietary Allowance for protein is only 0.8 grammes per kilogramme of body weight. The UK’s Reference Nutrient Intake is 0.75g per kg per day. “For most people, a good target is around 20-30g of protein per meal,” says Kat Chan, a nutritionist who suggests that some people could benefit from more. “Protein requirements increase during life phases that involve growth, repair or hormonal transition, so that includes adolescence, pregnancy, post-partum, perimenopause, menopause and later life. Protein also helps regulate blood sugar, which is important for mood, focus and steady energy. This can be particularly helpful for people navigating irregular meals, digestive issues or chronic stress.””

SaaStr: “The QBR isn’t about showing off your product or patting yourself on the back. It’s about aligning with your customer’s goals and showing how you’re helping them achieve those goals. Start by asking: What are their business objectives? What did they want to achieve when they signed up with you? Are you delivering on those objectives? Be honest about where you’re succeeding and where you’re falling short. What’s next for them? Use the QBR to understand their roadmap for the next 6-12 months and how you can support it. And importantly — ask them ahead of time what they want to discuss.  Too few do this for real.  And start there.”

FT: “The world’s largest economies will lack the workers they need to power growth and keep prices stable in the coming decades unless they attract more foreigners, top central bankers have warned…Across rich economies birth rates are at historically low levels, while people are living much longer. That has raised so-called dependency ratios, meaning that a far higher share of the population is no longer of working age.”

Cass Sunstein: “Hayek and the Mont Pelerins (and Posner and Epstein) seemed to be fighting old battles, and in important ways to be wrong. With respect to authoritarianism and tyranny, and the power of the state, of course they were right; but still, those battles seemed old. But those battles never were old. In important ways, Hayek and the Mont Pelerins (and Posner and Epstein, and Becker and Stigler) were right. Liberalism is a big tent. It’s much more than good to see them under it. It’s an honor to be there with them.”

From Vibe Coding to Martech’s Reinvention (Part 6)

SaaS and Martech

I asked Claude about how vibe coding will impact SaaS and Martech.

The Martech landscape has long been dominated by comprehensive platforms attempting to be everything to everyone. Salesforce, HubSpot, and Adobe Experience Cloud built their empires on the promise of unified, all-in-one solutions. But as one industry veteran noted, these platforms offer “a zillion features, but most companies really only need two or three features.”

Vibe coding is catalysing a shift toward hyper-customised solutions. Instead of adapting workflows to fit software limitations, marketing teams can build tools that perfectly match their processes. A B2B SaaS company can create a lead scoring system incorporating their unique market insights. An e-commerce brand can build attribution models reflecting their specific customer journey.

This trend is already visible in the numbers. The 2025 Martech landscape grew to 15,384 solutions, but perhaps more significantly, “custom-built or other” platforms jumped from 2% to 10% of marketing stacks, signaling a surge in bespoke solutions.

AI Agents as Marketing Stack Orchestrators

The average enterprise uses 130+ SaaS applications, with marketing departments often managing 20-30 specialized tools. Integration and orchestration have been persistent challenges, spawning entire categories like iPaaS (Integration Platform as a Service) and reverse ETL.

AI agents are emerging as a new orchestration layer. Instead of complex integration configurations, marketers can describe desired outcomes in natural language: “When someone downloads our whitepaper, wait 2 days, then send a personalised follow-up email unless they’ve visited our pricing page, in which case alert sales immediately.”

These agents don’t just connect systems—they make intelligent decisions. They can recognize patterns, suggest optimizations, and even autonomously adjust campaigns based on performance. We’re moving from rule-based automation to intelligent orchestration.

Impact on Marketing Automation and Campaign Management

Traditional marketing automation required technical expertise to implement. Building a multi-touch nurture campaign meant understanding trigger logic, conditional branches, and merge fields. Vibe coding eliminates these barriers.

Modern platforms are introducing conversational interfaces where marketers describe campaigns in plain English. “Create a campaign targeting enterprise prospects who’ve engaged with our content but haven’t requested a demo. Send them case studies relevant to their industry, and if they engage, have an SDR reach out within 24 hours.”

The AI doesn’t just create the campaign—it suggests improvements based on historical performance, automatically A/B tests variations, and adjusts timing based on engagement patterns. This isn’t just automation; it’s augmentation.

Rise of the Marketing Citizen Developer

Marketing operations professionals are becoming the vanguard of citizen development. They understand business needs, have basic technical literacy, but previously lacked the coding skills to build custom solutions. Vibe coding bridges this gap perfectly.

Real examples are compelling:

  • A marketing ops manager built a custom attribution model incorporating offline touchpoints their commercial platform couldn’t track
  • A demand generation team created a lead enrichment tool combining multiple data sources in ways vendors hadn’t anticipated
  • A content marketer developed an AI-powered tool analysing competitor content and suggesting strategic gaps to exploit

These aren’t replacing enterprise Martech—they’re filling gaps, adding customization layers, and solving the “last mile” problems that generic solutions can’t address.

Business Model Disruption

The traditional SaaS model relies on recurring revenue from feature-rich platforms. Vibe coding threatens this by enabling alternatives:

  • Build vs. Buy Dynamics: When custom development takes days instead of months, the economics shift dramatically. Why pay $50,000 annually for a platform when you can build exactly what you need for the cost of AI tokens and minimal hosting?
  • Feature Unbundling: SaaS platforms bundle features to justify pricing. Vibe coding enables cherry-picking—building just the features you need. This forces vendors to reconsider packaging and pricing strategies.
  • Democratised Competition: A small team can now compete with established vendors by rapidly building and iterating on specialized solutions. The barriers to entry in SaaS have never been lower.

The Pricing Crisis

The emergence of reasoning models has created what insiders call a “margin crisis.” Traditional SaaS pricing assumes predictable costs—servers, bandwidth, support. AI-powered vibe coding introduces variable compute costs that can spiral unexpectedly. The industry is scrambling to shift from seat-based to usage-based pricing, fundamentally altering the SaaS business model.

This creates opportunities for innovative pricing strategies:

  • Outcome-based pricing: Charging for results rather than access
  • Token pooling: Enterprise agreements with shared AI compute budgets
  • Hybrid models: Base platform fees plus usage-based AI features

Thinks 1733

Ezra Klein: “I don’t know whether A.I. will look, in the economic statistics of the next 10 years, more like the invention of the internet, the invention of electricity or something else entirely. I hope to see A.I. systems driving forward drug discovery and scientific research, but I am not yet certain they will. But I’m taken aback at how quickly we have begun to treat its presence in our lives as normal. I would not have believed in 2020 what GPT-5 would be able to do in 2025. I would not have believed how many people would be using it, nor how attached millions of them would be to it. But we’re already treating it as borderline banal — and so GPT-5 is just another update to a chatbot that has gone, in a few years, from barely speaking English to being able to intelligibly converse in virtually any imaginable voice about virtually anything a human being might want to talk about at a level that already exceeds that of most human beings. In the past few years, A.I. systems have developed the capacity to control computers on their own — using digital tools autonomously and effectively — and the length and complexity of the tasks they can carry out is rising exponentially.”

FT: “Within the next few decades, scientists will probably be able to build mirror life — organisms built from molecular components that are mirror images of the versions used in nature. I am a synthetic biologist. Engineering cells and bacteria is my trade. I was part of the team that, in 2010, created the world’s first living bacterial cell with a chemically synthesised genome. For decades, my colleagues and I have sought to scale the beneficial applications of this technology, helping to create vaccine strains, biofuels, pharmaceuticals and other molecules that could create new sources of clean energy, cure diseases and clean up the planet. Mirror life represents a profound break from this work. It would be created using entirely different building blocks, not the same molecules that are found in all known life. The organism’s DNA would twist to the left where ours twists to the right; their ribonucleic acids, essential to biological functions, would loop and bulge in the opposite direction… The solution is clear: we should choose not to build mirror life and pass laws to ensure nobody can.”

Andy Kessler: “Will robots replace house cleaners? Housekeepers? Nannies? Mr. Wineland thinks they can “help caretakers do more important things with children, do more things around the home, rather than picking up Legos yet again. It eliminates mundanity. They’re so additive, freeing humans to do more and more executive things.” This is exactly what technology should be doing. It’s early. Progress and affordability will come task by task. But get ready for the robot home invasion—and nicely folded T-shirts.”

SaaStr: “The venture capital market has officially split into two radically different universes. And this is super important for B2B founders to understand. In Universe A, solid B2B and SaaS companies raise at reasonable multiples following traditional venture math. In Universe B, AI-powered startups command valuations that defy every historical precedent. The data from Carta’s latest analysis reveals something unprecedented: we’re not just seeing a “hot market” — we’re witnessing the complete takeover of winner-take-all economics in venture capital.”

From Vibe Coding to Martech’s Reinvention (Part 5)

Writings – 2

Jason Lemkin: “The future of B2B software isn’t just AI-powered—it’s conversation-powered. It’s time for Vibeable SaaS. You tell the app, in plain English, the outcomes and workflows you want.  And it just creates them for you.  Bespoke.  In real time. It works in Replit, Lovable, etc.  At least mostly.  Now it should work the same way inside the B2B apps we all know, pay for, and rely on…Imagine this [on] your marketing automation platform: “Our Q4 campaign isn’t converting well. I need you to A/B test different email subject lines, but also try sending at different times for different segments. West Coast prospects should get emails at 9am their time, East Coast at 8am theirs. Anyone who hasn’t opened after 48 hours gets a follow-up with a different angle. And create a dashboard that shows me which combinations are working best.”… This is what I call a Vibe-able app. You communicate your intent in plain English, and the application morphs itself to match your vision… MCP gave us the infrastructure to connect AI to applications, and to get data from them. Now we need to go further. We need applications that don’t just work with AI—applications that can remake themselves to meet the customers’ needs.  Safely and securely, but for real. Your users shouldn’t need to learn your software. Your software should learn them.  ChatGPT taught us to start thinking this way.  Your customers are going to demand it soon from your B2B app, too. It’s time to make your app Vibe-able.”

Karthik S: “Vibe coding is a one way street. Once a program has been vibe coded, it is so verbose and unreadable that it is impossible for a human to edit it or change it. The only way to deal with this further is to further vibe code it. You need to keep this in mind while you make the decision on whether to vibe code or not. Basically, treat vibe coding as a black box and you will be good. The code will not look like what you might write. It will not look like what you think is good code. It will basically do the job, though. If you want control over your code, write the first version by yourself and then get Cursor to make small changes for you (making sure you edit them into your style while accepting them).”

Fast Company: “Over the last couple of years, AI coding tools like Claude Code (Anthropic), Codex (OpenAI), Cursor, Lovable, and Replit have reached far beyond auto-completing lines of code; they can generate entire apps and features from a plain-language prompt, even for users with little or no coding experience. But even as enterprise execs hope the tools will speed up their software production, many in the development community are finding that while vibe coding may be great for slapping together demos, it’s not so great for building secure, reliable, and explainable software. And the problems created by AI-generated code may only surface long after the software has shipped.”

FT: “Perhaps vibe coding will go the same way as DIY: plenty of people will experiment with projects at home and enjoy the process. Some people will become really good at it. But for the complicated jobs, many of us will discover a newfound respect for the professionals. We might just have to live through a few DIY disasters first.”

The New Stack: “Vibe coding is not about to eliminate programming jobs — at least not for work that matters. The future probably looks more like sophisticated autocomplete than wholesale code replacement, [Simon Ritter] said. This isn’t necessarily bad news. Programming languages have always evolved toward higher-level abstractions. We moved from assembly to C to Java to modern frameworks, each step making developers more productive while maintaining the precision that serious software requires, Ritter said. AI will undoubtedly play a bigger role in that evolution. But the fundamental tension between natural language ambiguity and software precision suggests human programmers are not going anywhere soon.”

VentureBeat: “The era of “vibe coding” – the conversational, often exploratory practice of prompting an AI for code, … – has given way to a more disciplined concept: agentic swarm coding… “Even Karpathy’s vibe coding term is legacy now. It’s outdated,” Val Bercovici, chief AI officer of WEKA, told me in a recent conversation. “It’s been superseded by this concept of agentic swarm coding, where multiple agents in coordination are delivering… very functional MVPs and version one apps.””

Claude summarised the issues with vibe coding:

  • Technical Debt Acceleration: AI-generated code often works but lacks the architectural elegance and maintainability of human-crafted solutions. When modifications are needed, developers find themselves untangling complex, inefficient logic that no human fully understands.
  • Security Vulnerabilities: Without human review, AI-generated code can contain subtle security flaws. One documented case involved a Replit AI agent deleting an entire database despite explicit instructions not to make changes.
  • The Debugging Nightmare: When vibe-coded applications break, debugging becomes exponentially harder. Traditional debugging requires understanding the code’s logic and structure—impossible when no human wrote or reviewed it.
  • Cost Explosions: The emergence of reasoning models has created what industry insiders call “token shock.” Companies like Cursor and Anthropic have had to implement emergency rate limits after individual users generated over $10,000 in monthly compute costs on $200 plans.

Thinks 1732

WSJ: “The internet transformed how people live and do business, but it took much longer than its enthusiastic early boosters would have predicted in the 1990s. For instance, it took a decade for U.S. home broadband to go from near zero penetration in 2000 to more than 60% of adults’ subscribing to it, according to figures from the Pew Research Center. The AI boom is different in many ways, but it could follow a similar trajectory: a burst of enthusiasm followed by a leveling-off as it bleeds into society and business, with the true scope of the benefits only clear years later.”

Business Standard: “Experts say companies will hire based on three key parameters going forward: tech expertise (knowledge of AI skills), domain expertise (banking, retail, life sciences), and client-specific expertise, which involves understanding the customer better through agentic AI.  This approach requires deeper contextual knowledge, or “contextual engineering”, where engineers must understand not just what they are building with, but also what they are building for. The more contextual awareness an engineer or manager has, the better they can monitor code and agents, preventing errors before they occur.“

FT: “How exactly will AI change the nature of browsing? More automation implies less human intervention. Three different models for how this might evolve are starting to emerge. In one, AI embedded into a browser supercharges the user experience by taking over some of the browsing functions. You could, for instance, ask the AI to open multiple web pages to compare airline fares or look at several reviews of a new movie. The browser performs routine work and provides some hand-holding, but a person stays in charge…A second approach also involves AI taking on browsing duties, though it takes place from within an AI app…In the third, AI agents use purpose-built tools to work online, rather than operating through browsers that were designed for humans. They tap into online services or databases through APIs (or programming “hooks”), and they use new web protocols, with names like MCP and A2A, that are designed to facilitate “agentic” action online.”

SaaStr: “The core of Palantir’s valuation premium lies in an unprecedented growth story. As one analyst noted: “The growth is frigging breathtaking, right? It goes from 12% growth at about $2 billion revenue in 2023 to almost 45% growth at $4 billion in ARR.” This level of re-acceleration is rare. According to Rory’s analysis: “I look across 20 years about only one in three companies re-accelerate for one year and about one in nine, one in ten, re-accelerates for two years.” Palantir has now sustained significant re-acceleration for two-plus years at scale. The trajectory: 12% → 23% → 45% growth rates.”

From Vibe Coding to Martech’s Reinvention (Part 4)

Writings – 1

Christopher S. Penn: “If you had to summarize this in one sentence, vibe coding is basically letting AI do the coding. You provide the ideas and the feedback, but you never write a line of code. Instead, the machine does all the coding, testing, correction, and ultimately deployment. What makes this different than other forms of AI content creation is the hands off nature. Prior to the current generation of AI language models, there was a lot of copy pasting involved as you went back and forth with AI and your code, error messages, and testing. AI was the workhorse, but you were the micromanager, heavily involved in assembling the pieces it produced. Prior to vibe coding, AI was like an IKEA manufacturing plant. It spit out pieces, and then it was up to you – usually with confusing directions – to figure out how to put it all together. Vibe coding arose with coding tools like Cursor, Cline, Aider, Claude Coder, and many more. These are billed as agentic coding, but they’re really much closer to automation.”

Techcrunch: “Lovable specializes in helping people build apps and websites, especially people with no coding experience. It’s one of the standouts in the popular AI category known as vibe coding, which lets users guide AI models as they produce code, websites, or whole applications. It’s been an attractive proposition for users: In just eight months, the Swedish company said it surpassed $100 million in ARR and raised a $200 million Series A at a $1.8 billion valuation — making it Europe’s fastest-growing unicorn. The Financial Times reported that investors are already hoping to launch a Series B, offering deals that would value the company at $4 billion….Speaking to TechCrunch, Osika laid out a vision for Lovable as the best place to build software products: a platform that can take users — founders specifically — through all the stages of product development so they can build AI-native companies more easily. “If you’re running a business, there are a lot of things you want to set up, like payments, understanding your users, and in the future, maybe even like ‘I need to incorporate my company,’” he said. “I want Lovable to help with all these things.” In late June, Lovable released an agent to help users read files, debug errors, search the web, generate images, and locate files — a first step toward making good on that vision. Lovable says it now has more than 2.3 million active users, 180,000 of whom are paying subscribers. Osika said the company chose its pricing simply by deciding what would help the company cover its own costs. His favorite Lovable use cases include a marketer building a sales training platform and an engineer running multiple small businesses on the platform.”

Lauren Goode writes after doing vibe coding for two days at Notion: “These changes—this invasion of AI code—has all happened within the past four to six months. Notion even has an AI engineer assigned to its enterprise sales team now, teaching software salespeople how to use AI in their own work. And it’s not just at Notion. It’s everywhere. My vibe-coding experiment, while solipsistically insightful, was already behind the curve. “The world is heating up in many ways, and the sense I have is not ‘I freed up more time’ but that there’s more urgency than ever to use these tools,” Simon [Last, Notion cofounder] said. The shift both exhilarates him and makes him anxious. He told me he looks back fondly on the not-too-distant past when he was simply coding and building stuff, “when there wasn’t, like, a crazy societal tidal wave happening. I think it would be crazy not to be a bit scared.””

WSJ: “In the next three years, market research and IT consulting firm Gartner predicts that 40% of new software for businesses will be created with techniques involving AI bots translating plain English prompts into usable code. Combined with the growing popularity of AI-powered assistants and editors, vibe coding’s rise in the enterprise reflects a significant shift in how quickly apps are being conceived and delivered, with many implications for professional developers… Jude Schramm, CIO of Fifth Third Bank, said the regional bank’s 700 full-time engineers may be entirely vibe coding in a few years’ time. Schramm said he’s already thinking more about the value of his developers as business problem-solvers rather than as code authors. “Engineering of the future is going to be much more a world of test, quality assurance and validation,” Schramm said. “That’s what we’re trying to train our folks for.””

Andrew Chen: “We are in the command line interface days of vibe coding. For the majority of creators, vibe coding will eventually fade, and vibe designing (with a visual paradigm) will come to dominate. People ultimately think better in a GUI-like format than a CLI-like format. Thus, in vibe designing you will show the AI the design outcomes you want, and then everything else is done for you. Yes, you may end up with tools to tweak the design details for extra controllability, and provide additional mockups that then get filled in underneath with code. But maybe folks will build software without seeing or learning a programming language… If vibe coding makes software trivial to build, then the bottlenecks shift to other places: 1) consistent creativity that stays ahead of everyone else. Anyone can write a tweet, but the best creators are the ones who consistently come up with new ideas. 2) distribution and network effects, where the first vibe coded product doesn’t win, but rather the first vibe coded product that hits scale that wins.”

Thinks 1731

FT: “Once an assembler that built its success on suppliers’ components, Xiaomi is aiming to reinvent itself as a manufacturing powerhouse. Xiaomi had humbler ambitions when it was formed just 15 years ago. Its name is Chinese for millet, with founder Lei Jun saying the company was created in 2010 in the spirit of “millet plus rifles”, a reference to Mao Zedong’s description of the Communist party’s modest military resources during the civil war. But in the space of its first three years, it leapfrogged incumbents to become the world’s third largest handset vendor and expanded its product line-up to include everything from rice cookers to robot vacuums.”

Scott Sumner: “Almost everything is downstream of integrity.”

Alex Tabarrok: “Across India, Greece, and Brazil the story converges: overpaying government workers distorts education, job search, and firm dynamics. The waste shows up as socially unproductive effort devoted to entering the echelons of government employment and a private sector which is drained of top talent causing it to be less productive and to grow more slowly. In short, rent seeking and misallocation from overly generous government compensation generate large macroeconomic losses. As relative compensation tends to be higher the poorer the economy, high government pay can be a development trap.”

Morgan Housel: “You should obsess over risks that do permanent damage and care little about risks that do temporary harm, but the opposite is more common…The only way to build wealth is to have a gap between your ego and your income…Having no FOMO might be the most important investing skill…Money’s greatest intrinsic value is its ability to give you control over your time.”

Ishan Bakshi on India and China: “While addressing these issues is harder than many appreciate, in recent years, it does seem that the policy focus in both countries has shifted away from prioritising growth. And so, in the absence of deep reforms, China persists with its debt-fuelled investment-export led model of growth, the limits of which are being tested in a world that is either unable or unwilling to absorb its excess capacity, while India, in the face of sluggish manufacturing and investment growth, continues to rely on domestic consumption, fuelled by debt and tax giveaways. The question is whether the required policy changes can be engineered or will the laws of path dependency make progress difficult.”

From Vibe Coding to Martech’s Reinvention (Part 3)

Landscape

The vibe coding ecosystem has rapidly stratified into distinct categories, each serving different user needs and technical sophistication levels:

Beginner-Friendly Platforms

Replit Agent leads the no-code revolution with its browser-native environment. Users describe their application in plain English, and Replit generates, deploys, and hosts the entire solution. Its strength lies in eliminating all technical barriers—no installation, no configuration, just pure creation through conversation.

Lovable (formerly GPT Engineer) focuses on building production-ready web applications with beautiful, modern designs. It excels at creating SaaS MVPs and marketing sites, automatically handling responsive design, animations, and user interactions that would typically require frontend expertise.

Bolt and Google AI Studio round out the beginner tier, offering streamlined experiences for specific use cases. Bolt specialises in full-stack web applications, while Google AI Studio provides powerful integration with Google’s ecosystem and Gemini models.

Professional-Grade Tools

Cursor has become the darling of professional developers embracing vibe coding. Built as a fork of VS Code, it maintains familiar development workflows while adding powerful AI capabilities. Its “Composer” feature allows developers to describe complex changes affecting multiple files, and the AI implements them simultaneously.

Windsurf, recently involved in a $3 billion acquisition drama between OpenAI and Google, represents the cutting edge of professional vibe coding. It combines the conversational interface of consumer tools with the power and flexibility demanded by enterprise development teams.

Claude Code and GitHub Copilot Workspace leverage their parent companies’ powerful AI models (Anthropic and OpenAI/Microsoft, respectively) to offer command-line and IDE-integrated vibe coding experiences that appeal to developers who prefer staying in their terminal or existing development environment.

The Hidden Giants: Consumer AI Platforms

Perhaps the most disruptive players aren’t positioning themselves as development tools at all. ChatGPT, Claude, and Gemini—with their millions of non-technical subscribers—are creating billions of micro-applications behind the scenes. When a user asks ChatGPT to analyse their sales data or create a custom calculator, the AI often writes and executes Python code invisibly, delivering only the results.

This “invisible software” phenomenon represents the ultimate expression of vibe coding—users getting custom software solutions without even knowing software was created.

The “Software for One” Revolution

Kevin Roose of the New York Times coined the term “software for one” to describe applications built for individual users’ specific needs. His experiments included an app that analysed his fridge contents to suggest lunch options—too specific for commercial development but perfect for personal use.

This concept extends far beyond personal productivity. In enterprises, vibe coding enables every employee to build tools addressing their unique workflow challenges. A marketing manager can create a custom dashboard combining data from multiple sources. A sales rep can build a prospect research tool tailored to their industry niche. The long tail of software needs—too specific for IT to address—suddenly becomes solvable.

Democratisation at Scale

The implications of democratised development extend beyond individual productivity. We’re witnessing the emergence of a new class of creator: the domain expert developer. These aren’t trained programmers but professionals who deeply understand their field’s problems and can now build solutions directly.

Consider a veteran marketer who understands customer journey mapping intricately. Previously, they’d submit requirements to IT, wait months for development, and often receive something that missed crucial nuances. With vibe coding, they can build, test, and refine solutions in real-time, ensuring perfect alignment with actual needs.

The Syntax to Problem-Solving Shift

Traditional programming education focuses heavily on syntax, data structures, and algorithms. Vibe coding shifts the essential skills toward:

  • Problem Decomposition: Breaking complex challenges into manageable components
  • Clear Communication: Articulating requirements precisely in natural language
  • Systems Thinking: Understanding how components interact without knowing implementation details
  • Iterative Refinement: Recognising what works and describing improvements effectively

This shift doesn’t eliminate the need for traditional developers—it changes their role from code producers to architects, reviewers, and quality guardians. The most valuable developers become those who can bridge the gap between vibe-coded prototypes and production-ready systems.

Thinks 1730

Gary Leff: “Flight delays get a lot of attention, and certainly mechanical and staffing issues are the fault of the airline. There’s also air traffic control which creates congestion – it isn’t just responsible for delays but also for longer flight times that get built into schedules. We don’t talk enough about that. Maybe the biggest failure in air travel is something we don’t talk about at all. How is it possible that people are being told to show up at the airport 2.5 to 3 hours before their flight, and that isn’t considered a failure of massive proportions?”

Andy Kessler: “With today’s flurry to invest in any company that can spell AI, Netscape’s biggest lesson is that nothing lasts forever. AOL bought Netscape in 1998 for $4.2 billion in stock. AOL’s peak value reached $222 billion in December 1999, and in January 2000 it bought Time Warner for $182 billion. The dot-com boom peaked a month later, and the merger was a massive failure. AOL stopped supporting Netscape browsers in 2008. In 2015 AOL was sold to Verizon for peanuts, and in 2021, along with the remains of Yahoo, it was sold to private-equity firm Apollo. In the current bull run, the “Magnificent Seven” stocks, growing on the backs of Netscape’s innovation, are worth $19 trillion. Netscape is a footnote in history but proof that the spark of a new idea and the freedom to pursue it, coupled with robust capital markets and even a dose of runaway speculation, can change everything. Even if the first movers, the original innovators, don’t make it.”

SaaStr: “Most B2B companies assume if customers aren’t complaining, they’re happy. If they’re not demanding attention, they don’t need it. If renewal conversations aren’t contentious, everything’s fine. This is one of the most expensive myths in B2B software. The customers who complain aren’t your biggest churn risk — they’re engaged enough to fight for a better experience. They care enough to push back. They’re giving you a chance to fix things. It’s the quiet ones you need to worry about. The ones who just… stop using your product as much. The ones who find workarounds. The ones who start looking elsewhere … but never tell you about it.”

Steve Stewart-Williams: “Psychologists have spent many decades sifting through the noise of human personality, and what they’ve found is surprisingly simple: five big traits that capture much of what makes you you. They’re called the Big Five, and they’re the closest thing psychology has to a periodic table of personality. Extraversion: Cheerfulness; sociability; assertiveness; Neuroticism: Proneness to anxiety, sadness, and other negative emotions; emotional volatility; Conscientiousness: Orderliness; reliability; industriousness; Agreeableness: Friendliness; politeness; compassion; Openness: Intellectual curiosity; creativity; openness to experience.” [via Arnold Kling]