What changes when AI enters a business
We follow AI once it stops being novelty and starts changing cost structures, workflows, control, technological dependence, and competitive advantage.
What we are watching
Compute infrastructure, agents, enterprise software, restricted model distribution, and decisions that turn AI into a layer of power, not just productivity.
Where it is being decided
In the cloud, inside workflows, in the relationship between provider and client, in model governance, and at the point where automation starts changing who gets to decide.
Why it matters
Because adopting AI is not just adding a tool. It means accepting new dependencies, new costs, and a new way of organising judgment, speed, and control.
Featured
Artificial Intelligence

Physical AI in the field has no technology problem — it has a design problem
There is a conversation that takes place at almost every agribusiness conference these days: the promise of autonomous robots that harvest, spray, monitor and transport without a single human worker having to bend their back under the sun. The presentations are flawless. The renders, impressive. And in the room, seated near the exit, are the people whose jobs are being automated — people nobody invited to co-design anything. That is not a minor detail. It is the structural fault line running through the entire discussion about Physical Artificial Intelligence applied to agriculture.
Isabel Ríos9 minLatest articles
LangChain Raised $300 Million. What It Bought Is Not Technology
There is a difference between building tools for developers to experiment with artificial intelligence agents and building the control layer from which companies operate them responsibly. LangChain spent most of its history on the former side. The week of November 20, 2026, with $300 million freshly raised in a Series C led by CapitalG and a valuation of $3 billion, it made clear that its bet is on the latter.
IFF's AI Knows How to Generate Concepts, but Someone Has to Know How to Use It
There is a moment in product development that nobody wants to say out loud: the moment when the innovation team stares at a blank page and does not know where to begin. It is not a lack of talent. It is that the early ideation process — that period when everything is possible and nothing is validated — remains, in most organizations, a mixture of intuition, experience bias, and the speed of the budgetary clock.
Why Companies Spend Millions on AI and Keep Producing the Same Results as Last Year
There is an experiment that repeats itself in thousands of companies and always ends the same way. The team identifies a repetitive task, connects a language model, automates the visible steps, and announces that they now have AI up and running. Six months later, the financial indicators have not moved.
When programming robots stops being a data problem
There is an implicit bet in most of the capital that has entered robotics over the past three years: that with enough human demonstrations, robots will learn to do what humans do. The logic is seductive because it already worked once, with language models. But robotics is not text, and that difference is beginning to cost those who assumed the same recipe would produce the same result.
Most Popular
Most voted on this page
The Recurring Revenue of AI Startups No Longer Guarantees What It Once Promised
There is a number that circulates through Silicon Valley pitch decks with the force of a closed argument: ARR, or Annual Recurring Revenue. For years it was the metric that separated serious startups from those simply burning cash on hope. According to data published in 2026 by venture capital firm Madrona, 77% of companies reassess their artificial intelligence vendors every six months or even on a continuous basis.
90
The most powerful model is not the one that wins in business
There is a number that the president of Alibaba.com chose to open his argument, and it is an uncomfortable number: 61.7%. That is the percentage of real-world commercial tasks successfully completed by the best artificial intelligence agent they tested. Across 107 real e-commerce operations, the most capable system available failed in four out of ten.
89
IFF's AI Knows How to Generate Concepts, but Someone Has to Know How to Use It
There is a moment in product development that nobody wants to say out loud: the moment when the innovation team stares at a blank page and does not know where to begin. It is not a lack of talent. It is that the early ideation process — that period when everything is possible and nothing is validated — remains, in most organizations, a mixture of intuition, experience bias, and the speed of the budgetary clock.
88
Why Companies Spend Millions on AI and Keep Producing the Same Results as Last Year
There is an experiment that repeats itself in thousands of companies and always ends the same way. The team identifies a repetitive task, connects a language model, automates the visible steps, and announces that they now have AI up and running. Six months later, the financial indicators have not moved.
88

Enterprise AI Already Masters the Task, but Doesn't Know What the Organization Does
In its 2026 global survey, 80% of professionals reported improvements in their personal productivity thanks to artificial intelligence. But when the question shifts from 'did it help you?' to 'did it generate financial value for the company?', the number drops to 37%. What that figure describes is not a technology problem, but a translation problem.

Why Respect Is Replacing Authenticity as the Currency of Creator Marketing
Ten years ago, brands were debating whether they should even be on Instagram. Today, Neil Waller, co-founder and co-CEO of Whalar Group, argues that the debate is no longer about whether creators matter, but about how to structure a commercial relationship that works when both parties have something real to lose and something real to gain. His central thesis is that 'authenticity' is too vague an aspiration to serve as an operational principle.

Salesforce Paid $2.7 Billion for the Human Voice in Customer Service
There is a decision Marc Benioff has made twelve times in 2026: buying what Salesforce could not build as fast as the market demanded. The latest, and probably the most revealing of all, is the acquisition of Cresta, a company that automates customer service calls in real time, for a figure that The Information places at $2.7 billion. Salesforce confirmed the deal on November 2, 2026, though it did not disclose the price.

52% of enterprise AI projects never reach production — and the problem is not technical
According to 2026 research by CIO&Leader, 52% of companies surveyed reported that fewer than one in four AI pilots ever becomes an operational system. Only 8% managed to have more than three-quarters of their experiments cross that line. And just 12% were able to demonstrate significant, measurable return on investment.

95% of the AI Models Companies Build Never Reach Production
The chief technology officer of a large global corporation recently told Abhijit Kakhandiki, senior vice president at BMC Software, something that captures the state of the market with uncomfortable precision: his data science teams had built more than 200 artificial intelligence models. Fewer than ten were in production. A conversion rate of less than five percent.
FAQ
Artificial Intelligence
What changes when AI stops being a pilot and enters operations?
It changes how costs are allocated, how work is coordinated, and where control lives. AI stops being an isolated tool and starts touching the operating architecture of the company.
When does an AI agent create advantage and when does it only add complexity?
It creates advantage when it removes friction, expands capacity, or improves decisions in an important process. It adds complexity when it is inserted without clear governance, useful metrics, or a specific bottleneck to solve.
What risks appear when a company depends on a model or compute provider?
Cost risk, availability risk, slower iteration, and loss of strategic control. When the provider concentrates too much power, adoption can harden into structural dependence.

Borrowed Autonomy: Why Enterprise AI Agents Cannot Simply Be Switched On
There is a difference between a system that works in a demonstration and one that does not destroy value when no one is watching. With AI agents, that difference is measured in modified records, processed payments, and decisions made without human intervention. The problem is not technological. It is one of governance.
Clara Montes8 min
DeepSeek banned from federal agencies: how a Chinese startup reconfigured the geopolitics of software
The U.S. Department of Homeland Security made public in October 2026 an order dating from July of the same year: federal civilian agencies must remove DeepSeek — and any third-party products built on its implementations — from their information technology systems. This is not a cautionary notice. It is a procurement ban and an active removal mandate.
Martín Soler9 min
AI Agents in Your Text Messages and Why That Reshuffles Power Over Your Attention
There is a precise moment when a technology stops being a new tool and becomes a distinct market condition. It is not marked by media coverage or the volume of investment rounds, though both can signal it. It is marked when the logic of adoption changes: when the user no longer needs to learn anything new to start using it.
Gabriel Paz9 min
AMD Acquires Fei-Fei Li's Startup for $8.2 Billion and Reshapes Its Bet on Physics and Artificial Intelligence
On September 28, 2026, AMD announced the acquisition of World Labs in an all-stock transaction valued at approximately $8.2 billion. The stated goal: to move decisively into the branch of artificial intelligence that does not process text but instead models the physical world. The deal was signed by AMD chief executive Lisa Su and Fei-Fei Li, co-founder and chief executive of World Labs.
Sofía Valenzuela9 min
The Return on Enterprise AI Is an Architecture Problem, Not an Intelligence Problem
There is a figure that CIOs are memorizing with discomfort: 72% of organizations admit that their AI investments are, at best, breaking even. At worst, losing money. What is failing is not the machine's intelligence. What is failing is the business architecture surrounding it.
Lucía Navarro8 min
The most powerful model is not the one that wins in business
There is a number that the president of Alibaba.com chose to open his argument, and it is an uncomfortable number: 61.7%. That is the percentage of real-world commercial tasks successfully completed by the best artificial intelligence agent they tested. Across 107 real e-commerce operations, the most capable system available failed in four out of ten.
Andrés Molina9 min
ScrollEd bets the problem was never the scroll, but its content
There is a paradox that the education industry has spent years ignoring: while young people spend hours navigating vertical feeds with a level of concentration any classroom would envy, learning systems keep sending them PDFs and textbooks that no one opens. ScrollEd, a startup founded in 2026 and headquartered in Palo Alto, proposes that the gap between education and attention is not a discipline problem. It is an interface problem.
Diego Salazar8 min
CADDi Reaches $1.2 Billion Valuation by Solving the Problem Nobody Had Properly Digitized
There is a problem that any manufacturing operations director recognizes instantly: their company buys the same part, from different suppliers, at different prices, without knowing it had already purchased it before. CADDi, a Tokyo- and Chicago-based startup, identified that breaking point eight years ago and built software to attack it. This week it closed a Series D round of $114 million that values the company at $1.2 billion.
Tomás Rivera9 min
Oracle Spends $2.8 Billion to Reinvent Itself: This Is What the Real Cost of the AI Transition Looks Like
Larry Ellison cancelled in September 2026 a plan to sell up to 50 million Oracle shares, equivalent to roughly $7.5 billion at that Friday's closing price. No official explanation was given. What did appear in a regulatory filing submitted that same week was another figure: Oracle expanded its fiscal 2026 restructuring plan by an additional $700 million, bringing the total expected cost of the programme to approximately $2.8 billion.
Gabriel Paz8 min
Why Ellison Withdrew Oracle's $7.5 Billion Share-Sale Plan
Larry Ellison, co-founder and executive chairman of Oracle Corporation, adopted on June 22, 2026, a trading plan to sell up to 50 million ordinary shares of the company. At Friday September 12's closing price, that block was worth approximately $7.5 billion. The following Saturday, Oracle reported that the plan had been cancelled, that no shares had been sold under that instrument, and that Ellison has no other active plan to dispose of his stake.
Mateo Vargas8 min
When AI Acts Without Permission, the Problem Is Not the Model
For years, the conversation about artificial intelligence risks revolved around the same axis: the model hallucinates, invents figures, cites sources that do not exist, confuses facts. It was a real problem, costly in some cases, embarrassing in others. But it was, at its core, a problem of output quality. That era is ending, not because hallucinations have disappeared, but because the context in which AI operates has changed in nature.
Ignacio Silva9 min
Maven Robotics raised $100 million without having a single physical robot
There is a scene that captures what Maven Robotics is building better than any investor presentation. It was 2024, the company had been in existence for a matter of weeks, and, in the words of its own CEO, Hamza Derbas, the only tangible thing they could show was "a cartoon of a robot and a team of people." A large-scale consumer goods company was in Silicon Valley meeting with four robotics firms to evaluate automation projects.
Elena Costa9 min
Enterprise AI Is Still Waiting for Its Platform Moment
There are technologies that exist long before anyone makes them usable. Email had been around for decades before Hotmail showed your mother how to use it. GPS navigation existed in military devices years before a hundred-dollar gadget told a delivery driver which way to turn.
Simón Arce8 min
The Recurring Revenue of AI Startups No Longer Guarantees What It Once Promised
There is a number that circulates through Silicon Valley pitch decks with the force of a closed argument: ARR, or Annual Recurring Revenue. For years it was the metric that separated serious startups from those simply burning cash on hope. According to data published in 2026 by venture capital firm Madrona, 77% of companies reassess their artificial intelligence vendors every six months or even on a continuous basis.
Martín Soler7 min