 |
|
|
|||||
Nostalgia has been one of the dominant themes of 2025, from AI-generated scenes of the good ol days to the resurgence of analog hobbies. Retro, a friends-only photo journal, recently launched a new feature which taps into this mindset, turning your camera roll into a personal time machine. The Rewind feature, launched this week, resurfaces camera roll memories from this time last year. These are private to you unless you choose to share with others. People are taking more photos than ever but theyre actually doing less with them. Its almost as if those photos go into the ether, Nathan Sharp, cofounder and CEO of Retro, tells Fast Company. We built Retro to change that. Our mission is to bring friends closer and help you appreciate the important moments in life. The Rewind feature does that by surfacing forgotten photos and making it easy to share memories with the people who matter most, Sharp adds. On the app, Rewind can be launched from the end of the row of shared photos, or from the middle tab in the bottom navigation bar on the app. Users have the option to share or send the photos to a friend, or hide those theyd rather not see. Theres also a dice icon, which takes users to a random memory when tapped. The idea of dusting off old photo albums is nothing new. Facebooks On This Day feature performs a similar function, while Apple Photos has been known to make emotional slideshows of ephemera in its camera roll or surprise you with long-forgotten photos of an ex. Its not the solo nostalgia you get from apps built to store or manage photos. Its also not the same as social platforms that prioritize links and news over friends content, says Sharp. That’s the difference: we’re building for genuine connection with real friends, not algorithms, likes, or audience growth. Sharp, who previously spent over six years at Meta, founded the photo-sharing startup with Ryan Olson, Retros CTO, in 2022. Now with roughly a million users, Retro just hit #1 in photo apps in 12 countries, is the #1 overall app in six countries (including Germany, Austria, Finland, the Netherlands, Sweden, and Switzerland) above Instagram and ChatGPT, and is climbing fast in the U.S. It was also selected as a finalist for Apples 2025 Cultural Impact Award. The apps main function is sharing unfiltered photos of whats happening during your week with a private group of friends, or in shared albums. No public likes, algorithm-induced doomscrolling, or pressure to curate an aesthetic photo dump. A wider pushback against performativity and, in turn, surveillance culture, has internet users turning to online spaces and apps that exist beyond influencer culture, social clout and e-commerce. Here, the internet is restored to its original purpose: facilitating moments of authentic connection both ad-free and slop-free. Gen Z is actively looking for an alternative to algorithmic feeds dominated by influencers and brands, says Sharp. We see social moving toward digital sanctuaries where connection is easy and authentic, not performative. That’s what we’re building.
Category:
E-Commerce
Thomas Kuhn was a philosopher whose groundbreaking 1962 book, The Structure of Scientific Revolutions, is credited with bringing the term paradigm shift to pop culture. Kuhn described how scientific communities stick to established paradigms, even as evidence of their limitations mounted. Widely accepted paradigms for understanding and interpreting knowledge dont crumble under the weight of mere data. Instead, they tend to persist until a crisis emergeswhen anomalies become so disruptive that a shift to a new paradigm is unavoidable. Zoning was established in the early 20th century as a way to protect homeowners from unwanted industrial developments nearby. It was pitched as a way to separate heavy industry from residential areas, which made practical sense at a time when factories polluted neighborhoods. Early industrial cities were notorious for their noise, filth, sickness, and all-around misery. The wealthy had options, so theyd put some distance between themselves and factory life. You can imagine that the elite would want to guarantee never having to deal with the industrial riffraff. Zoning would give such guarantees. You can also imagine that social workers and other empaths would want to guarantee the poor and middle class had the same separation from the dirty parts of a city as the elites had. Zoning would give such guarantees. {"blockType":"creator-network-promo","data":{"mediaUrl":"","headline":"Urbanism Speakeasy","description":"Join Andy Boenau as he explores ideas that the infrastructure status quo would rather keep quiet. To learn more, visit urbanismspeakeasy.com.","substackDomain":"https:\/\/www.urbanismspeakeasy.com\/","colorTheme":"blue","redirectUrl":""}} But zoning wasnt used merely as a tool to separate heavy industry from residential zones. Local power brokers segregated all the land usesseparating single-family homes from apartments, office buildings from retail, residential from retail, and so on. The regulatory framework became so normalized in America that its hard for people to imagine life without it: Without zoning, my neighbor might build a strip club and a paper mill. Unintended consequences Normal science, the activity in which most scientists inevitably spend almost all of their time, is predicated on the assumption that the scientific community knows what the world is like. Much of the success of the enterprise derives from the communitys willingness to defend that assumption, if necessary, at considerable cost. As Kuhn wouldve predicted, the normal science of zoning has produced a number of anomalies that increasingly contradict zonings purported benefits. Housing Expense and Shortage: By restricting a variety of housing sizes and types, zoning codes limit the supply of housing, driving up prices and making places unaffordable for many residents. Environmental Degradation: Zoning encourages urban sprawl by pushing residential development outward into zones that are only practically reachable by car. Zoning codes create low-density, car-centric development, at great expense to our natural environment. Social Segregation: Zoning is a devilish segregation tool. Throughout pre-zoning history, cities had opportunities for people from all walks of life, social standing, and economic standing. Economic Stagnation and Opportunity Costs: By prohibiting a mixture of land uses in a neighborhood, zoning limits economic activity, making it difficult for small businesses to thrive in residential neighborhoods or for residents to access amenities without a car. Car Dependency: Neighborhood pharmacies are outlawed, so you drive to CVS just to get a birthday card. Neighborhood restaurants are outlawed, so you drive your kids to Chick-fil-A. Neighborhood salons are outlawed, so you drive to get your nails done. A resilient paradigm Changing a paradigm isnt just about accepting new facts, its about challenging an entire worldview, and thats something humans are generally reluctant to do. And in spite of all its harms, the zoning paradigm remains resilient among the experts because: Planning departments are organized around zoning administration. Professional credentialing still lionizes zoning codes. University programs train students to use zoning for the greater good. Thousands of attorneys specialize in zoning law. Lobbying pressure remains intense from industries that benefit from strict land-use policies. There are powerful incentives to preserve the system, even among professionals who privately acknowledge its failures. Kuhn observed that paradigms persist not because they work well, but because entire careers, departments, and professional identities are built upon them. Challenging zoning means threatening not just an idea, but the livelihoods and expertise of countless people. Much like a fundamentalist belief system, zoning has developed a language of justification that makes it difficult to challenge. Clever defenses like preserving neighborhood character or protecting property values are invoked to defend restrictive zoning policies, even when these policies have been proven to harm the vast majority of people. Zoning defenders use language not to inform, but to deflect and manipulate. A tipping point Kuhn would say a paradigm shift requires a moment of crisis, a point at which the old framework can no longer explain or accommodate the reality of a situation. I think were getting there with zoning, because the accumulating anomalies are becoming too severe to ignore. Scientific revolutions reshaped how we understand the world. A zoning revolution has the potential to transform our small towns, big cities, and sprawling suburbs in positive ways we have yet to fully imagine. We have 100 years of evidence that zoning has brought more harm than good. {"blockType":"creator-network-promo","data":{"mediaUrl":"","headline":"Urbanism Speakeasy","description":"Join Andy Boenau as he explores ideas that the infrastructure status quo would rather keep quiet. To learn more, visit urbanismspeakeasy.com.","substackDomain":"https:\/\/www.urbanismspeakeasy.com\/","colorTheme":"blue","redirectUrl":""}}
Category:
E-Commerce
The flap of a butterflys wings in South America can famously lead to a tornado in the Caribbean. The so-called butterfly effector sensitive dependence on initial conditions, as it is more technically knownis of profound relevance for organizations seeking to deploy AI solutions. As systems become more and more interconnected by AI capabilities that sit across and reach into an increasing number of critical functions, the risk of cascade failureslocalized glitches that ripple outward into organization-wide disruptionsgrows substantially. It is natural to focus AI risk management efforts on individual systems where distinct risks are easy to identify. A senior executive might ask how much the company stands to lose if the predictive model makes inaccurate predictions. How exposed could we be if the chatbot gives out information it shouldnt? What will happen if the new automated system runs into an edge case it cant handle? These are all important questions. But focusing on these kinds of issues exclusively can provide a false sense of safety. The most dangerous AI failures are not the ones that remain confined to one particular area. They are the ones that spread. How Cascade Failures Work While many AI systems currently operate as isolated nodes, it is only when these become joined up across organizations that artificial intelligence will fully deliver on its promise. Networks of AI agents that communicate across departments; automated ordering systems that link customer service chatbots to logistics hubs, or even to the factory floor; executive decision-support models that draw information from every corner of the organizationthese are the kinds of AI implementations that will deliver transformative value. But they are also the kinds of systems that create the biggest risks. {"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/creator-faisalhoque.png","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/faisal-hoque.png","eyebrow":"","headline":"Ready to thrive at the intersection of business, technology, and humanity? ","dek":"Faisal Hoques books, podcast, and his companies give leaders the frameworks and platforms to align purpose, people, process, and techturning disruption into meaningful, lasting progress.","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"https:\/\/faisalhoque.com","theme":{"bg":"#02263c","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#ffffff","buttonHoverBg":"#3b3f46","buttonText":"#000000"},"imageDesktopId":91420512,"imageMobileId":91420514,"shareable":false,"slug":""}} Consider how quickly problems can multiply: Corrupted data at a single collection point can poison the outputs of every analytical tool downstream. A security flaw in one model becomes a doorway into every system it touches. And when several AI applications compete for the same computing resources, a spike in demand can choke performance across the boardoften at the worst possible moment. When AI is siloed, failures are contained. When AI is interconnected, failures can propagate in ways that are difficult to predict and even harder to stop. The 2010 Flash Crash in the U.S. stock markets showed how algorithms can interact in unexpected ways, causing problems on a scale that can be hard to imagine. On the morning of May 6th, more than a trillion dollars was wiped off the value of the Dow Jones Industrial Average in a matter of minutes as automated systems triggered a spiral of sell-offs. Despite several years of investigation, the exact cause of the crash is still unknown. What the Flash Crash revealed is that when autonomous systems interact, their combined behavior can diverge dramatically from what any single system was programmed to do. None of the algorithms were designed to crash the market and none of them would have done so if they were operating independently. But the interactions between themeach responding to signals created by othersproduced an unexpected result at the systemic level that was divorced from the goals of any one part of that system. This is the nature of cascade risk. The danger lies not in any individual AI system failing, but in the unpredictable ways that interconnected systems can amplify and spread failures across organizational boundaries. The Hidden Connections Several characteristics make AI systems particularly susceptible to cascading failures. Shared data dependencies create hidden connections between seemingly independent systems. Two AI applications might appear to be completely separate, but if they rely on the same underlying data sources, a corruption or error in that data may affect both simultaneously. And a simultaneous failure may have consequences that are more severe than the sum of the individual failures. These kinds of dependencies and their possible outcomes often go unmapped until a failure forces the organization to take notice. Shared infrastructure creates similar vulnerabilities. Multiple AI systems running on common cloud resources or the same on-site computational infrastructure can all be affected by a single point of failure. During high-demand periods, competition between systems for resources can degrade performance across the board in ways that are difficult to predict or diagnose. Feedback loops between AI systems can amplify small errors into large disruptions. When one systems output feeds into another systems input, and the second systems output then influences the first system, the potential for runaway effects increases. What begins as a minor anomaly can be magnified through successive iterations until it produces significant failures. Integration with critical operations also raises the stakes dramatically. When AI becomes embedded in systems that organizations depend onsupply chains, financial operations, customer service, manufacturingcascade failures dont just create technical problems. They disrupt the core functions that keep the business running. The Organizational Blind Spot Perhaps the greatest challenge in managing cascade risk is organizational rather than technical. The systems that interact to create cascade failures often span different departments, different teams, and different areas of expertise. No single person or group has visibility into all the connections and dependencies. This means that cascade risk management requires cross-functional coordination that cuts against traditional organizational structures. It requires mapping dependencies that cross departmental boundaries. It requires testing failure scenarios that involve multiple systems simultaneously. And it requires governance structures that can make decisions about acceptable risk levels across the organization as a whole, not just within individual units. Organizations that treat AI implementation as a series of independent projectseach managed by its own team, each evaluated on its own meritswill inevitably create the conditions for cascading failures. The connections between systems will emerge organically, without deliberate design or oversight. And when failures occur, they will propagate through pathways that no one fully understood. The alternative is to treat the entire AI ecosystem as an interconnected whole from the beginning. This means thinking about how systems will interact before they are built. It means maintining visibility into dependencies as systems evolve. And it means accepting that the reliability of any individual system is less important than the resilience of the system of systems. Four Ways to Protect Your Organization from AI Cascade Failures 1. Map your AI dependencies before they map themselves. Most organizations discover their system interdependencies only after a failure reveals them. Dont wait. Conduct a systematic audit of how your AI systems connectwhat data they share, what infrastructure they rely on, what outputs feed into other systems inputs. Create a visual map of these dependencies and update it as your AI ecosystem evolves. The goal isnt to eliminate connections (interconnection is often where value comes from) but to understand them well enough to anticipate how failures might propagate. 2. Design circuit breakers into your architecture. Financial markets use automatic trading halts to prevent cascading crashes. Your AI systems need equivalent mechanisms. Build monitoring systems that can detect unusual patternssudden spikes in error rates, unexpected resource consumption, anomalous outputsand automatically pause operations before small problems become large ones. These circuit breakers buy time for human operators to assess situations and intervene. The cost of brief pauses is far less than the cost of cascading failures. 3. Test failure scenarios across system boundaries. Traditional testing evaluates whether individual systems work correctly. Cascade risk requires testing how systems fail together. Run exercises that simulate failures in one system and trace the effects through connected systems. What happens to your customer service AI when your data pipeline delivers corrupted information? How does your inventory system respond when your demand forecasting model produces anomalous predictions? These cross-boundary tests reveal vulnerabilities that single-system testing will never find. 4. Establish cross-functional AI governance. Cascade risks emerge from the gaps between organizational silos. Managing them requires governance structures that span those silosa cross-functional team with visibility into AI implementations across departments and the authority to make decisions about system interactions, acceptable risk levels, and required safeguards. This team should own the dependency map, oversee cross-boundary testing, and ensure that new AI implementations are evaluated not just for their individual merits but for how they affect the broader ecosystem. The butterflys wings are already flapping. The organizations that thrive will be those that see the tornado comingnot by monitoring any single system, but by understanding how all their systems connect. {"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/creator-faisalhoque.png","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/faisal-hoque.png","eyebrow":"","headline":"Ready to thrive at the intersection of business, technology, and humanity? ","dek":"Faisal Hoques books, podcast, and his companies give leaders the frameworks and platforms to align purpose, people, process, and techturning disruption into meaningful, lasting progress.","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"https:\/\/faisalhoque.com","theme":{"bg":"#02263c","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#ffffff","buttonHoverBg":"#3b3f46","buttonText":"#000000"},"imageDesktopId":91420512,"imageMobileId":91420514,"shareable":false,"slug":""}}
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||