Biotech startup NewLimit secures a major funding round after claiming a ‘breakthrough discovery’ of a prototype medicine that rewinds cellular aging in the liver.
The CEO of a data platform startup knows that his business will be a public company eventually, but he insists that it won’t happen in the near future.
As artificial intelligence (AI) takes over many entry-level tasks, early career roles are becoming harder to find and land, with postings declining by 35% since 2023. This decrease has created an experience gap. Entry-level candidates lack the skillset employers are looking for and at the same time, traditional pathways to gain those skills are disappearing. For years, entry-level roles were the natural starting point for a career. But in reality, they served a deeper purpose. These roles were how new graduates learned to operate in the workforce, providing an opportunity to gain and practice skills, contribute to business outcomes and build confidence. The challenge is that this model assumes employers will continue to invest in early talent development, but that is no longer a given. If AI can successfully offload entry-level tasks, the business case for training early-stage workers becomes more difficult to justify. With Entry-Level Roles Disappearing, Education Must Bridge the Gap Closing the experience gap has traditionally been framed as a shared responsibility between employers and educational institutions. That model is breaking down. As entry-level roles shrink, it is increasingly unrealistic to expect them to continue to carry the responsibility for developing that talent. This doesn’t eliminate the need for partnership between employers and institutions, but it does require that institutions take the lead in designing learning environments that mirror the first one to two years of professional work, ensuring students graduate with the foundational skills and experience that reflect the realities of the modern workplace. Designing Education Around Real-World Application Success starts in the classroom. Rather than separating learning from application, institutions must begin embedding real-world experience directly into coursework. Advances in technology are making this more accessible across industries. Simulation tools, virtual and augmented reality allow students to engage in hands-on learning that reflects actual job settings, from technical trades to professional services. This approach ensures that students aren’t just learning concepts but applying them in context. The result is a more continuous, integrated opportunity to gain experience. Creating a Continuous Pipeline of Experience Gaining relevant, career-aligned skills can often be more valuable than a purely academic education. Structured co-op and work-integrated learning models led by institutions offer a way to build a steady pipeline of experience, providing students the ability to alternate between classroom learning and real-world work throughout their education, at a time when internships are difficult to come by. Today, internship applications are nearly twice as competitive as they were just a year ago and as a result, more than half of students (56%) seeking an internship are unable to secure one. Externships can complement this by offering flexible, project-based experiences that expand access to real experience. Externships allow students to gain experience without needing to secure a full-time role, like an internship. They often are short-term, unpaid opportunities for learners to shadow professionals and explore a career path quickly. Together, these approaches create repeated exposure to workplace expectations, allowing students to build valuable skills and confidence over time, while giving employers earlier access to emerging talent. Institutions that remain responsive to the needs of today’s learners and employers by intentionally integrating hands-on experience throughout the student journey are best positioned to deliver meaningful, career-oriented outcomes. Northeastern is a strong example of an institution that has long distinguished itself by its connection to evolving, real-world job requirements. Extending Learning Beyond Graduation Closing the experience gap doesn’t end at graduation. As traditional entry-level roles continue to evolve, the need to build and grow skills extends well into the early years of a career. In today’s job market, early career readiness is no longer a nice-to-have, it is a necessity for accessing career opportunity. For those employers that do choose to partner with institutions and invest in entry-level talent, there are several clear benefits. By participating in experiential learning programs, employers can help shape a strong talent pipeline that is better prepared from day one. The result is faster employee ramp up, stronger performance and a more diverse pool of candidates. As AI reshapes entry-level work, fewer companies will see it as their responsibility to cultivate early career skills. The system that was in place for developing early experience has fractured. If we don’t redesign how and where experience is built, more young workers will find themselves locked out of opportunity. The future of work doesn’t just demand new skills; it demands a new way of gaining those skills and now is the moment to design it.
L’Oréal passed. The internet didn’t.
In the 1980s, Big Tobacco started playing a major role in America’s food industry, buying up companies like General Foods, Kraft, and Nabisco. And that role, according to a new study, included using R&D from the cigarette business to make ultra-processed foods, such as the enduringly popular Lunchables brand of kid-friendly prepackaged meals and snacks. Such foods, the study reveals, were engineered “for consumer pleasure and appeal” with help from cigarette-related research on flavor engineering, packaging developments, and more. The study was recently published in the American Journal of Public Health. It’s one article in an entire issue focused on ultra-processed foods, their connection to chronic diseases and addiction, and how tobacco companies essentially created “the modern ultra-processed foods industry.” Why Big Tobacco got into the food industry Laura Schmidt, a professor of medicine and health policy researcher at the University of California, San Francisco, who authored the study, explored the topic by asking why Big Tobacco companies got into the food business in the first place. Tobacco giant Philip Morris bought General Foods in 1985 and acquired Kraft in 1988. (It was spun out in 2007.) This coincided with the rise of public health concerns, criticisms, and lawsuits against the tobacco industry. That expansion was more than just a way to diversify at a time when Big Tobacco’s business was facing scrutiny. “The reason they got into the food business was because they wanted to use tobacco R&D assets to make food,” Schmidt says. She outlines how that happened through a case study of Lunchables, using internal documents made public from litigation of the tobacco industry. Lunchables were in premarket development at General Foods when Philip Morris bought that company; the products were later released by Kraft. (The two food divisions merged in 1989.) Today the brand is manufactured by Kraft Heinz. In a statement to Fast Company, Nicolas Amaya, president of Kraft Heinz’s North American market, emphasized that the company hasn’t had any affiliation with Philip Morris since 2007. “What we can speak to is the food we make today,” Amaya said. “Our portfolio today includes affordable options with more protein, more whole grains, less sugar and sodium, and no artificial dyes.” Lunchables are still considered an ultra-processed food, part of a category the study connects to the tobacco industry’s R&D process. However, Amaya said, Kraft Heinz continues to “evolve our portfolio based on consumer preferences and feel proud of the role we play in helping people live delicious, healthy and balanced lives.” “Technical synergies” between food and cigarettes While Philip Morris owned Kraft, executives spoke about the “technical synergies” between its food and tobacco product lines. It even created a Technical Synergies Committee to streamline R&D for chemical additives and processing and packaging technologies across both cigarettes and food. One way Philip Morris used its experience creating cigarettes to make products like Lunchables was through its R&D of flavorings, colors, and chemicals. “When Philip Morris bought up these food companies, they already had these massive libraries of flavor and colors,” Schmidt says. Sometimes the company would develop chemical additives that didn’t make it into cigarettes (it had patents on artificial sweeteners, for example), but that R&D was “on the shelf” when it got into the food industry. The two product lines even shared supply chains for “refined agricultural ingredients and chemical additives,” Schmidt’s paper says, noting that they also used the same processing and packaging technologies, including “chemically encapsulated flavors.” Fast Company reached out to Altria Group, parent company of Philip Morris USA, for comment. What distinguishes ultra-processed foods Philip Morris’s approach, Schmidt says, led to the rise in ultra-processed foods, which now account for 70% of packaged foods in the U.S. and 62% of the calories in children’s diets. Studies have linked ultra-processed foods to a rise in obesity and health concerns like diabetes and cardiovascular disease. Lots of foods can be considered “processed” (like bread, for instance). And many of the approaches to industrial manufacturing, processing, and marketing could be considered common strategies for all consumer packaged goods companies that reach global audiences. There isn’t one definition for ultra-processed foods. But food scientists, relying on a validated food classification system, say they are distinguished by their cosmetic additives (which change flavor, texture, and color) and their extreme processing, which includes breaking ingredients down to the molecular level and recombining them with additives in industrial facilities to make what Schmidt calls “something that looks like food.” “Those two things—that pretty much is the R&D that Philip Morris took from its cigarette division and applied to Lunchables,” Schmidt says. “They’re the same business model. They’re the same formulation strategy. They use the same ingredients. They use the same industrial processing technologies. They use the same product design technologies.” Tobacco-owned ultra-processed foods, her paper notes, were “disproportionally hyper-palatable,” meaning they contained high levels of fat, sugar, sodium, or carbs, all factors that have been linked to overeating. “Why should we see them so separate?” As food companies hopped on the low-fat craze of the 1990s, Philip Morris brought its nicotine flavor researchers over to Kraft to consult on how to make such foods taste better. Kraft researchers credited those nicotine researchers, documents show, as crucial to the development of products like low-fat Lunchables. Another commonality comes from the strategy of “consumer-driven product development,” which Big Tobacco pioneered. This used psychological research to discover and target customers’ subconscious desires. For cigarettes, that resulted in marketing like the Marlboro Man and products like Virginia Slims, targeted at women. For the food business, that shaped the design of Lunchables “around what children really want and need, which is autonomy and freedom to play,” Schmidt says. “That’s why Lunchables were designed to really function like a toy.” All are examples of how cigarettes and food were approached with similar corporate strategies. The study includes a quote from then-Philip Morris CEO Hamish Maxwell noting “all of our major businesses share common characteristics.” To Schmidt, this means the two industries should also face similar approaches when it comes to public health policies and regulations. “If the people who designed and made these foods think that they’re very similar businesses, why should we see them so separate?” she says. “It does start to make you wonder, really, how different are ultra-processed foods from cigarettes?”
Finally, some good news for weary travelers. After battling long TSA lines and wait times earlier this year, airline passengers departing from Boston Logan International Airport now have the option to skip the lines and head straight to the gate, thanks to a new “first-in-the-nation” remote check-in pilot program, according to the Transportation Security Administration. “This is going to be a game-changer for so many people,” Massachusetts Governor Maura Healey told CBS News Boston. The pilot program, launched June 1, is being run in partnership with the Massachusetts Port Authority (Massport)—an independent public authority that runs and operates the Commonwealth’s three major airports. It is the first off-airport security checkpoint in North America. How does remote check-in work? Instead of going through security at the airport, eligible travelers can print boarding passes, check bags, and clear security at a new, remote terminal in Framingham, Massachusetts, about 23 miles southwest of Logan, before boarding a secure bus. “[Passengers will] go through a full TSA screening checkpoint lane, very similar to like you would at the airport,” said Peter Howe, Massport’s deputy director of roadway management. Framingham’s Logan Express airport bus has long been a popular park-and-ride spot for locals looking to avoid heavy airport traffic and hefty airport parking fees. A bus ticket from the remote terminal costs $9 each way (kids younger than 18 ride free with a ticketed family member) and parking costs $7 per day, compared to $37 to $46 per day at Logan. Currently, only passengers flying on Delta Air Lines and JetBlue Airways between the hours of 5:30 a.m. and 4 p.m. will be able to use this service, though there are plans to expand to additional airlines in the future, according to Massport. Buses from Framingham run about every half hour. Passengers can book tickets from 90 days to 90 minutes before departure, though they’re encouraged to book early to allow for flexibility as the pilot program gets underway. TSA said it is working on “rolling out more potential remote passenger screening options nationwide.”
Enterprise AI today feels strangely familiar: The infrastructure is powerful. The capabilities are real. The demonstrations are impressive. Models can write, summarize, reason, code, search, retrieve, translate, classify, plan, and increasingly act. The raw machinery is there. And yet, inside companies, the same pattern keeps repeating: pilots everywhere, transformation nowhere near the promise. The first article in this series argued that large language models were never built to run a company because companies operate through memory, context, feedback, constraints, state, incentives, and dependencies — not through isolated sequences of text. The second argued that enterprise AI must move from answers to outcomes, from prompts to constraints, and from copilots to systems of action. The third argued that when enterprise AI finally works, it will not look like a better chatbot. It will look like intelligence embedded into the organization itself. The next question is obvious: If all of that is true, where are we in the historical cycle? My answer is simple: Enterprise AI is in 1991. It has TCP/IP. But it does not yet have the web. The internet worked before the web The analogy matters because it prevents us from confusing infrastructure with industrialization. In 1991, the internet already worked. TCP/IP moved packets. Email connected people across institutions. FTP moved files. Telnet enabled remote access. Universities, research labs, and technically sophisticated organizations could use the network. But for a normal company, the internet was still not a business environment in the modern sense. It was powerful, but not yet consumable. Then the World Wide Web added a thin but decisive layer: URLs, HTTP, HTML, servers, and browsers. CERN’s history of the web explains that by Christmas 1990, Tim Berners-Lee had already defined the basic concepts of HTML, HTTP, and URLs, and written the first browser/editor and server software. In 1991, CERN released the WWW software more broadly and announced it on internet newsgroups, allowing the idea to spread beyond its original context. That layer did not invent networking. It made networking legible, usable, and buildable for the rest of the world. That is exactly the distinction that enterprise AI is missing today. Models are not the web Large language models are extraordinary infrastructure. They are probably one of the most important technological substrates of our time. But infrastructure is not the same as an application layer. A company using LLMs today often resembles a bookstore trying to sell online before the web existed. The network is there. Packets move. Servers exist. But every transaction would require custom machinery: custom protocols, custom interfaces, custom logic, custom deployment, custom integration . . . custom everything. That is not commerce. That is engineering. This is why the current enterprise AI market still depends so heavily on pilots, bespoke deployments, forward-deployed engineers, and consulting-heavy implementations. The problem is not that the underlying intelligence is fake. It is that the layer that makes it consumable by ordinary organizations is still immature. A model can generate an answer. But a company needs a system that knows where that answer fits, what data it can use, what constraints apply, who has permission to act, what process is being affected, what outcome matters, and how the system learns from what happens next. That is not a prompt. That is a missing layer. The missing layer has specific properties This is the important part. The gap is not vague. It is identifiable. Enterprise AI does not simply need “more AI.” It needs the equivalent of the web layer: a structured application layer that turns raw intelligence into something organizations can use repeatedly, safely, and at scale. That layer has to provide at least seven things: Persistent context: The system cannot behave as if every interaction begins from zero. Business semantics: It must understand customers, products, policies, workflows, roles, and constraints in company-specific terms. Process state: It must know where work is, what has happened, what is pending, and what depends on what. Permission and governance models: It must operate inside organizational boundaries, not around them. Feedback loops: It must learn from outcomes, not merely generate outputs. Interoperability: It must connect to systems of record, tools, data, and workflows without bespoke reconstruction every time. Repeatability: It must be deployable as architecture, not as artisanal consulting. This is why Anthropic’s recent emphasis on context engineering is so revealing. Its engineering team explicitly describes context as a critical but finite resource for agents, and argues that the challenge is now to curate and manage the information that surrounds the model — not merely write better prompts. That is the direction of travel: The model is no longer the whole product. The environment around the model becomes the product. The second analogy: Pre-ERP enterprise software The web analogy explains the missing application layer. But there is a second analogy that is just as useful: Enterprise AI is also in the pre-industrial phase of enterprise software. Before ERP systems became standardized platforms, corporate software was often a patchwork of custom implementations, integrations, internal systems, and consulting projects. SAP’s history shows the long arc from specialized business software toward enterprise application platforms, with SAP eventually becoming the market leader in enterprise application software. That evolution mattered because it did not merely digitize individual functions. It industrialized a way of representing the company: finance, inventory, procurement, manufacturing, HR, logistics, and reporting became standardized enough to create repeatable implementations and a partner ecosystem. The same happened later in CRM and SaaS. Salesforce’s own history shows how AppExchange became a marketplace for independent software vendors and applications, turning Salesforce from a product into a platform ecosystem. That is the difference between a category that depends on custom projects and a category that scales. Today, enterprise AI is still too often stuck in the custom-project phase. Each company needs its processes mapped, its data cleaned, its permissions understood, its workflows reconstructed, its constraints encoded, and its outcomes defined. That work is necessary. But when it has to be done manually in every deployment, it proves the platform layer has not yet arrived. Why the next winners may not be the model providers This is where the analogy becomes strategically uncomfortable: In the web transition, the critical question was not who owned the cables. It was who defined the layer that made the network usable. In enterprise software, the critical question was not who owned the database or the server hardware. It was who defined the system of business representation and built the ecosystem around it. The same may be true in AI: The winners of the next phase may not be the companies with the largest models or the biggest clusters. Those companies will matter enormously, just as telecom providers, server vendors, and infrastructure companies mattered enormously. But the category-defining power may belong to whoever builds the missing application layer: the layer that allows enterprise intelligence to become persistent, governed, contextual, process-aware, and repeatable. That is why the current obsession with model performance, context windows, and benchmark scores is both understandable and incomplete. Better models are necessary, but they are not sufficient. As McKinsey’s 2025 research on AI adoption notes: Companies seeing the most value are not just deploying tools; they are redesigning workflows and embedding AI into processes. Deloitte reaches a similar conclusion in its work on agentic AI: Many organizations are hitting a wall because they are trying to automate processes designed for humans instead of reimagining how the work should actually be done. In other words, the bottleneck is moving up the stack. Industrialization always looks obvious in retrospect The strange thing about these transitions is that they are difficult to see while they are happening and are obvious afterward. Before the web, the internet looked like a domain for specialists. After the web, it became a business environment. Before ERP and SaaS platforms matured, enterprise software looked like custom automation. Afterward, it became repeatable architecture. Before cloud platforms matured, infrastructure looked like procurement and systems administration. Afterward, it became programmable capacity. Enterprise AI is now approaching the same kind of threshold. The current phase still looks artisanal: pilots, prototypes, integrations, forward-deployed engineers, consulting-heavy engagements, custom workflow mapping. That is normal. Every powerful technology goes through a phase in which experts have to carry it across the gap manually. But that phase is not the destination. The destination is the layer that makes the expert intervention less central. This is why the next 5 years matter The web did not turn the internet into a commercial civilization overnight. ERP did not standardize the enterprise overnight. Salesforce did not create a platform ecosystem in a single release. These transitions take years. But the decisive moment is usually the same: Someone defines the missing layer well enough that everyone else can build on it. That is where enterprise AI is now. We have the models. We have the infrastructure. We have the early agents. We have the consulting wave. We have the pilots. We have the frustration. We have the proof that isolated tools are not enough. We have the emerging recognition that context, workflows, constraints, memory, and outcomes matter more than prompts. What we do not yet have is the equivalent of the browser, the URL, the ERP layer, the AppExchange — the standard application layer that makes enterprise AI consumable by ordinary companies. And until that appears, the industry will remain trapped in a paradox: extraordinary intelligence delivered through extraordinary effort. Where’s the web for enterprise AI? That is the question. Not “which model is best?”Not “which chatbot is most impressive?”Not “which copilot has the slickest interface?” The real question is who will define the layer that turns intelligence into enterprise infrastructure? Because once that layer appears, the current debate will look very different. Forward-deployed engineers will not disappear, but they will become less central. Custom deployments will not vanish, but they will stop being the dominant pattern. Pilots will not go away, but the path from pilot to production will become far shorter. Artificial intelligence will stop being something companies experiment with and become something companies are built on. That is the industrial era of enterprise AI. And it has not arrived yet. But if history is any guide, once the missing layer appears, it will feel as if it was obvious all along.