When someone opens the door and enters a hospital room, wearing a stethoscope is a telltale sign that theyre a clinician. This medical device has been around for over 200 years and remains a staple in the clinic despite significant advances in medical diagnostics and technologies.
The stethoscope is a medical instrument used to listen to and amplify the internal sounds produced by the body. Physicians still use the sounds they hear through stethoscopes as initial indicators of heart or lung diseases. For example, a heart murmur or crackling lungs often signify an issue is present. Although there have been significant advances in imaging and monitoring technologies, the stethoscope remains a quick, accessible, and cost-effective tool for assessing a patients health.
Though stethoscopes remain useful today, audible symptoms of disease often appear only at later stages of illness. At that point, treatments are less likely to work and outcomes are often poor. This is especially the case for heart disease, where changes in heart sounds are not always clearly defined and may be difficult to hear.
We are scientists and engineers who are exploring ways to use heart sounds to detect disease earlier and more accurately. Our research suggests that combining stethoscopes with artificial intelligence could help doctors be less reliant on the human ear to diagnose heart disease, leading to more timely and effective treatment.
History of the stethoscope
The invention of the stethoscope is widely credited to the 19th-century French physician René Theophile Hyacinthe Laënnec. Before the stethoscope, physicians often placed their ear directly on a patients chest to listen for abnormalities in breathing and heart sounds.
In 1816, a young girl showing symptoms of heart disease sought consultation with Laënnec. Placing his ear on her chest, however, was considered socially inappropriate. Inspired by children transmitting sounds through a long wooden stick, he instead rolled a sheet of paper to listen to her heart. He was surprised by the sudden clarity of the heart sounds, and the first stethoscope was born.
Over the next couple of decades, researchers modified the shape of this early stethoscope to improve its comfort, portability, and sound transmission. This includes the addition of a thin, flat membrane called a diaphragm that vibrates and amplifies sound.
The next major breakthrough occurred in the mid-1850s, when Irish physician Arthur Leared and American physician George Philip Cammann developed stethoscopes that could transmit sounds to both ears. These binaural stethoscopes use two flexible tubes connected to separate earpieces, allowing clearer and more balanced sound by reducing outside noise.
These early models are remarkably similar to the stethoscopes medical doctors use today, with only slight modifications mainly designed for user comfort.
Listening to the heart
Medical schools continue to teach the art of auscultationthe use of sound to assess the function of the heart, lungs, and other organs. Digital models of stethoscopes, which have been commercially available since the early 2000s, offer new tools like sound amplification and recordingyet the basic principle that Laënnec introduced endures.
When listening to the heart, doctors pay close attention to the familiar lub-dub rhythm of each heartbeat. The first soundthe lubhappens when the valves between the upper and lower chambers of the heart close as it contracts and pushes blood out to the body. The second soundthe duboccurs when the valves leading out of the heart close as the heart relaxes and refills with blood.
Along with these two normal sounds, doctors also listen for unusual noisessuch as murmurs, extra beats, or clicksthat can point to problems with how blood is flowing or whether the heart valves are working properly.
Heart sounds can vary greatly depending on the type of heart disease present. Sometimes, different diseases produce the same abnormal sound. For example, a systolic murmuran extra sound between first and second heart soundsmay be heard with narrowing of either the aortic or pulmonary valve. Yet the very same murmur can also appear when the heart is structurally normal and healthy. This overlap makes it challenging to diagnose disease based solely on the presence of murmurs.
Teaching AI to hear what people cant
AI technology can identify the hidden differences in the sounds of healthy and damaged hearts and use them to diagnose disease before traditional acoustic changes like murmurs even appear. Instead of relying on the presence of extra or abnormal sounds to diagnose disease, AI can detect differences in sound that are too faint or subtle for the human ear to detect.
To build these algorithms, researchers record heart sounds using digital stethoscopes. These stethoscopes convert sound into electronic signals that can be amplified, stored, and analyzed using computers. Researchers can then label which sounds are normal or abnormal to train an algorithm to recognize patterns in the sounds it can then use to predict whether new sounds are normal or abnormal.
Researchers are developing algorithms that can analyze digitally recorded heart sounds in combination with digital stethoscopes as a low-cost, noninvasive, and accessible tool to screen for heart disease. However, a lot of these algorithms are built on datasets of moderate-to-severe heart disease. Because it is difficult to find patients at early stages of disease, prior to when symptoms begin to show, the algorithms dont have much information on what hearts in the earliest stages of disease sound like.
To bridge this gap, our team is using animal models to teach the algorithms to analyze heart sounds to find early signs of disease. After training the algorithms on these sounds, we assess their accuracy by comparing them with image scans of calcium buildup in the heart. Our research suggests that an AI-based algorithm can classify healthy heart sounds correctly over 95% of the time and can even differentiate between types of heart disease with nearly 85% accuracy. Most importantly, our algorithm is able to detect early stages of disease, before cardiac murmurs or structural changes appear.
We believe teaching AI to hear what humans cant could transform how doctors diagnose and respon to heart disease.
Valentina Dargam is a research assistant professor of biomedical engineering at Florida International University.
Joshua Hutcheson is an associate professor of biomedical engineering at Florida International University.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
When Accenture announced plans to lay off 11,000 workers who it deemed could not be reskilled for AI, the tech consulting giant framed the decision as a training issue: some people simply cannot learn what they need to learn to thrive in the world of AI. But this narrative fundamentally misunderstandsand significantly underplaysthe deeper challenge.
Doug McMillon, the CEO of Walmart, pointed to this bigger challenge recently when he said, AI is going to change literally every job. Now, if this turns out to be true, every role will have to be reimagined. And when every role changes, this is more than a change in each job or even a specific field. It implies a profound and systemic change in the nature and meaning of the work itself.
For instance, when a customer service reps job changes from answering questions to managing AI escalations, they are no longer doing old-fashioned customer servicethey are doing AI supervision in a customer service context. Their supervisor isnt managing people anymore; they are orchestrating a hybrid intelligence system composed of humans and AI. And HR isnt evaluating communication skills; they are assessing humanAI collaboration capacity. The job titles remain the same, but the actual work has become something entirely different.
{"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.","ctaText":"Learn More","ctaUrl":"https:\/\/faisalhoque.com","theme":{"bg":"#02263c","text":"#ffffff","eyebrow":"#9aa2aa","buttonBg":"#ffffff","buttonText":"#000000"},"imageDesktopId":91420512,"imageMobileId":91420514}}
You cannot prepare people for this disruption by sending them to a three-day workshop on how to prompt more effectively. When the change is as systemic as this, the real question is not whether individuals can be separately reskilled. It is whether organizations can transform themselves at the scale and speed AI demands.
Two types of transformation
To understand the reskilling demands created by AI transformation, it helps to distinguish between bounded and unbounded transformations.
Bounded transformations are organizational changes that follow a predictable path, starting from specific areas of operation with well-defined capabilities to develop. They unfold in distinct stages, allowing companies to master one phase before moving to the next.
Unbounded transformations, on the other hand, are sweeping changes that affect all parts of an organization at the same time, with no single point of origin. Because they simultaneously alter job functions, competencies, processes, and performance measures in interconnected ways, they can’t be tackled piecemeal or rolled out sequentiallythey demand a holistic, coordinated strategy.
The AI revolution is a paradigmatic example of an unbounded transformation, as it fundamentally reshapes how we think, work, and create value across every industry, function, and level of the organizationredefining not just individual tasks but the very nature of human contribution to work itself.
And that means that it is not enough to simply reskill employees for AI. Instead, business leaders will need to transform the entire ecosystem of workthe infrastructure, the interconnected roles, and the culture that enables change. And they will often need to do all of this across the entire organization at oncenot sequentially, not department by department, but everywhere simultaneously.
There are three key dimensions that organizations need to address if they are to successfully transform themselves and reskill their workers for the AI revolution.
1. Rebuilding the infrastructure of work
Most reskilling budgets cover workshops and certifications. Almost none cover what actually determines success: rebuilding the systems people work within.
For example, AI often now handles routine inquiries in contact centers while humans tackle complex cases. As McKinsey argues, successfully implementing this shift demands far more than teaching agents to use AI tools. Businesses must rethink operating models, workflows, and talent systemscreating escalation protocols that integrate with AI triage, metrics that measure human-AI collaboration rather than individual ticket counts, and training that builds the judgment needed to handle the ambiguous cases that AI cant decide. Career paths and team structures must evolve to support hybrid human-AI capacity.
Very little of this work is training in any classical senserather, it is organizational architecture and system-building. And the organizations that do not undertake this work will find that their AI reskilling programs will inevitably fail.
2. The network effect: why roles must transform together
Organizational roles do not exist in isolation. They are interconnected nodes in an organizational network. When AI transforms one role, it also transforms every other role it touches.
For example, when AI chatbots handle routine customer inquiries, frontline agents typically shift to managing only complex situations, which may be more emotionally charged for the client. This immediately transforms the role of their trainers and coaches, who must now redesign their curriculum away from teaching efficient delivery of scripted informational responses toward teaching de-escalation techniques, empathy skills, and complex judgment calls. Further, team supervisors will now no longer be able to evaluate performance based on call handle times and throughputthey must instead develop new frameworks for assessing emotional intelligence and problem-solving under pressure.
The result is that holistic and comprehensive role redesign is essential if employees are to be successfully reskilled for AI. AI transformation requires synchronized change across interconnected roleswhen one piece of the network shifts, every connected piece must shift with it.
3. Cultural transformation
As Peter Drucker almost said, culture eats reskilling for breakfast. It is crucial for organizations to understand that cultural transformation is not a nice-to-have follow-on that comes after technical change. Rather, it is the prerequisite that determines whether technical change takes root at all. Without the right culture, training budgets become write-offs and transformtion initiatives become expensive failures.
Consider a financial services firm training analysts on AI tools. If the culture punishes AI-assisted mistakes more harshly than human mistakes, adoption dies. If success metrics still reward heroic individual effort, collaboration with AI will be undermined. If executives do not visibly use AI and acknowledge their own learning struggles, teams will treat it as optional theater rather than strategic imperative.
The culture that enables AI reskilling is one built on curiosity, not certainty. This culture prizes experimentation over perfection and treats failure as data, not disgrace. Indeed, because AI tools evolve so quickly, the defining capability of an AI-ready culture is not mastery but continuous learning. Relatedly, psychological safety becomes essential: people must feel free to test, question, and sometimes get it wrong in public.
And the signal for all of this comes from the top. When leaders openly use AI, admit what they dont know, and share their own learning process, they make exploration permissible. When they do not, fear takes its place.
In short, successful AI cultures dont celebrate competencethey celebrate learning.
Conclusion
AI reskilling is not a training challengeit is an organizational transformation imperative. Companies that recognize this will rebuild their infrastructure, redesign interconnected roles, and cultivate learning cultures. Those that dont will keep announcing layoffs and blaming workers for failures that were always about systems, not people.
{"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.","ctaText":"Learn More","ctaUrl":"https:\/\/faisalhoque.com","theme":{"bg":"#02263c","text":"#ffffff","eyebrow":"#9aa2aa","buttonBg":"#ffffff","buttonText":"#000000"},"imageDesktopId":91420512,"imageMobileId":91420514}}
AI is often sold as the ultimate productivity hack. Just imagine: the report you dreaded writing, drafted in seconds. The spreadsheet you didnt want to touch, analyzed instantly. The code that once took you days, generated before lunch. For professionals who already struggle with overwhelm and the daily battle to manage their time, AI feels like salvation.
At Lifehack Method, where we help clients master time management and build systems for living fulfilling, balanced lives, we see this every day. People are desperate for tools that promise to take the weight off their shoulders. AI seems like the next logical step in that search. Theres no denying the dopamine hit of a blank page suddenly filling with words or lines of code. AI gives the illusion of acceleration, and in the moment, that feels like productivity. Youre doing something, and the grind of starting from scratch is gone.
But theres a problem: faster doesnt always mean more productive, and saved time doesnt always translate into better outcomes. The real test of productivity isnt how quickly you start, but whether you finish with work thats accurate, useful, and aligned with your goals. Thats where cracks begin to show.
AI can make you feel productive without actually being productive
A recent MIT study found that 95% of generative AI pilots in companies produced little to no measurable impact on profit and loss, despite $3040 billion in enterprise investment, because most GenAI systems do not retain feedback, adapt to context, or improve over time. In other words, the time people think theyre saving isnt translating into organizational productivity.
A similar story shows up among software developers in a recent controlled study. After trying AI coding assistants, developers estimated they experienced 1030% productivity gains. But in actuality, experienced coders took 19% longer when using AI tools on codebases they knew well. They not only lost time in practicethey walked away convinced theyd saved it. Thats a dangerous mismatch.
McKinseys research adds nuance: AI can indeed help with repetitive or shallow work tasks like painstakingly referencing large documents or analyzing invoices. But the productivity boost shrinks when tasks are complex or require deep, sustained attention. In other words, AI may help you clear the easy stuff off your plate, but its harder to get it to do the work that really moves the needle.
Why is that?
The 90% mirage
Heres the paradox of AI: it often gets you 90% of the way there, which feels like a huge time savings. But that last 10%checking for errors, refining details, making sure it actually workscan eat up as much time as you saved. The most common mistake is assuming 90% is good enough and shipping it.
Jeff Escalante is an engineering director at Clerk, puts it bluntly: Anything that you ask it to do, it will more than likely end up making one or more mistakes in what it puts out. Whether thats fabricating statistics, or making up things that are not real . . . or writing code that just doesnt work, he says. Its something that is really cool and really interesting to use, but also is something that you have to know you cant trust and cant rely on. It needs to be reviewed by an expert before you take what it puts out and deliver it, [especially if] its sensitive or important.
His advice? Treat AI like an intern: great for low-level work, occasionally useful when given training, but absolutely not someone youd send into a client meeting unsupervised. And if youre hoping eventually itll be foolproof, think again.
Jeff Smith, PhD is the founder of QuantumIOT and a serial technology entrepreneur. He says its important to think of the AI as an assistant because it still makes mistakes and it will make mistakes for a long time. Its probabilistic, not deterministic.
If youre a domain expert, you can spot and fix that last 10%. If youre not, you risk handing off work that looks polished but is quietly broken. That means wasted time correcting mistakesor worse, reputational damage. Many ambitious employees eager to level up with AI end up doing the opposite: walking into client pitches with beautiful decks full of hallucinated insights and an action plan that doesnt match the Statement of Work.
So should we throw AI out the window? Not exactly. But definitely stop treating it like a self-driving car and more like a stick shift: powerful, but only if you actually know how to drive.
How to use AI without losing control of your time
The most productive people dont hand over the keys to AI. They stay in the drivers seat. Here are a few rules emerging from early research and expert guidance:
Be the subject matter expert. If you dont know what excellent looks like, AI can lead you astray. The time you save drafting could vanish in endless rounds of corrections.
Use AI as a draft partner, not a finisher. The sweet spot is breaking inertiahelping you brainstorm, sketch a structure, or generate a starting point. Iterative prompting is the key to better AI outputs, but the final say will always belong to you.
Automate the shallow, protect the deep. Let AI knock out routine, low-value worksummaries, boilerplate, admin, certain emails. Guard your deep-work hours for the kind of thinking that actually moves the needle. Real productivity isnt about speed; its about aligning time with your top priorities.
Track actual outcomes. Dont confuse the feeling of speed with actual results. Measure it. Did the AI really shave an hour off your workflowor just generate more drafts to wade through?
And keep some perspective: were still in the early-adopter stage. As Smith puts it, Itll be a bit of a rocky road [but] theres tons of great tools that are going to come your way.
Productivity is still human business
At its best, AI helps remove the drudge work that crowds our days, giving us more room to think, plan, and focus on what matters. At its worst, it tricks us into mistaking busywork for progress.
AI wont manage your time for you. It wont choose your priorities or tell you which meetings to skip. That disciplineof mastering your schedule, focusing on high-leverage work, and knowing where your energy should gostill rests on human shoulders. Once that foundation is in place, AI can be a powerful ally. Without it, AI risks amplifying the chaos.
AI is a fast, powerful, occasionally unreliable tool. But like any tool, it only works if you weld it with intention. Youre still the driver. AI can help you go faster, but only if you know where you want to go.
So long, nine-to-five. There’s a new work schedule that’s taking over. The grueling “996” schedulewhich stands for 9 a.m. to 9 p.m., six days a weekis gaining momentum across the U.S., especially in certain industries.
If a 72-hour work week sounds all-consuming, that’s precisely the point. The 996 schedulewhich became popularized in China, eventually leading to protests and even claims that it led to a handful of worker deathsis meant to foster a eat-sleep-work lifestyle. Keith Spencer, a career expert at FlexJobs, told Fast Company that the trend is most commonly being seen across AI startups that “are embracing this approach to accelerate growth and remain competitive on a global scale.”
While the intense work ethic sounds overwhelming, Spencer says that some young and hungry workers may actually be drawn to it. “Certain employees, especially younger workers, may even welcome this level of intense dedication, particularly when additional pay or incentives are offered,” he explains.
That may be especially true as the rise in 996 culture has been touted by major tech leaders like Elon Musk, who have long promoted a work ethic that asks employees to make some major sacrifices. Musk opened up about the need for increased time commitments on X back in 2018 in a tweet promoting working for his companies as being revolutionary, but requiring immense dedication. “There are way easier places to work, but nobody ever changed the world on 40 hours a week,” Musk wrote.
When a commenter asked the Tesla CEO what the right number of hours a week was, he replied that it “varies per person, but about 80 sustained, peaking above 100 at times. Pain level increases exponentially above 80.
With that same hardcore work ethic in mind, companies embracing the lifestyle seem only to be interested in hiring employees who are “obsessive,” a word that appears on New York City-based AI startup Rilla’s career page to describe those who work there. Rilla explains on its applications that candidates who aren’t “excited” about working “70 hrs/week in person with some of the most ambitious people in NYC” should not apply.
Will Gao, the companys head of growth, previously told Wired about the benefits of the schedule. There’s a really strong and growing subculture of people, especially in my generationGen Zwho grew up listening to stories of Steve Jobs and Bill Gates, entrepreneurs who dedicated their lives to building life-changing companies, Gao explained. Kobe Bryant dedicated all his waking hours to basketball, and I dont think there are a lot of people saying that Kobe Bryant shouldnt have worked as hard as he did.
At Cognition, a San Francisco startup that is building an AI software engineer, the mansion workspace has living quarters for employees who don’t have time to go home. The company’s CEO Scott Wu explained what’s expected on X. “Cognition has an extreme performance culture, and were up-front about this in hiring so there are no surprises later,” Wu wrote. “We routinely are at the office through the weekend and do some of our best work late into the night. Many of us literally live where we work.”
The 996 trend seems to be taking off in the U.S. at a time when burnout is already at an all-time high. A 2025 report from online marketplace Care.com found that burnout was more impactful than employers thought. Companies believed 45% of their workers were at risk of burnout. But a staggering 69% of employees said they were actually at moderate to high risk.
For that reason, Spencer warns that companies should “exercise caution” when leaning into the 996 schedule. In addition to burnout and overwhelm, Spencer says that overworking can even trigger “a quarter-life career crisis” when employees feel disconnected with their career as a result of overworkingwhich isn’t great for the employee and doesn’t serve the company either.
Winter Peng, founder and CEO of Silveroak Capital Academy, an elite career coaching and mentorship firm, agrees that the hustle culture can backfire. She tells Fast Company that it “destroys the creativity that drives real innovation.” Peng continues: “U.S. startups adopting 996 are trading innovation for compliance” and says that ultimately, “their best talent will simply leave” in favor of companies who believe in work-life balance.
E-commerce continues to eat up ever-increasing share of the U.S. retail market: Americans bought more than $3.3 billion of items online every day in the second quarter of last year, according to U.S. Census Bureau data. Online retails share of spending is increasing with every year that passes.
Traditionally, thats meant typing a term or phrase into a search bar and clicking through to a shopping basket. But the AI revolution is poised to swamp online retail, too, with agentic AI set to shop on behalf of customers. The e-commerce sector is rapidly preparing for whats about to comean influx of non-human customers acting on behalf of humans.
We avoid hype around technology, but AI agentic shopping could bring huge changes to retail if it is widely adopted, says Clare Walsh, director of education at the Institute of Analytics, a professional body for data analytics experts.
The usually staid professional organization is full-throated in its belief that agentic AI shopping could change society. AI-empowered agentic shoppersrobots that learn your shopping needs and preferences and then shop for youhave the potential to be as disruptive for e-commerce as moving bricks and mortar retail online, Walsh says.
Those within the retail sector are equally enamored with the concept of AIs arrival. For many years now, eCommerce shopping experiences have consisted of a search bar and a long list of item responses,” Doug McMillon, CEO of Walmart, said in a statement announcing his retailers partnership with OpenAI to enable shoppers to buy things directly through ChatGPT with the aid of AI agents. That is about to change.
The early data suggests that the reality is matching the hype. AI-driven traffic to retail sites was up 4,700% in the U.S. in the last 12 months, according to Visa. The future isnt coming, its already checking out, says Rubail Birwadker, global head of growth at the credit-card company.
Shoppers want AI to help them, according to Birwadker, who points to research that 85% of shoppers say AI agents improved their experience. Separate research, provided to Fast Company by consumer insights company GWI, suggests one in five people are comfortable receiving product recommendations from AI agents. Data from consultancy Kearney indicates 60% of consumers plan to use AI agents to shop in the next year.
But ensuring those shopping interactions are secure is trickier.
Investment in cleaner data
In mid-October, Visa launched its Trusted Agent Protocol (TAP), a framework that would allow AI agents to share and access data that would ensure it can protect against fraud and bot activity. This enables merchants to avoid blocking legitimate transactions and degrading user experience, says Birwadker.
For now, TAP applies only to the Visa network. But having established it across their payments system, the massive payment processing giant intends on broadening its use. Enabling agents to safely and securely act on a consumers behalf requires an open ecosystem-wide approach and we will look to extend Trusted Agent Protocol to be compatible with other payment networks and methods in future phases, says Birwadker.
The behind-the-scenes transfer of data is where most within the e-commerce sector are rushing to catch up to what they predict is coming with the advent of agentic AI shoppers, says Robin Anderson, head of product management at Tribe Payments, a global paytech company. Were seeing investment in cleaner data, faster checkouts, stronger fraud controls and tighter integrations between systems, he says. This is because an AI agent will make a buying decision in seconds, and if theres frictiona payment fails, a price isnt clearthe sales gone. Anderson believes the arrival of AI shopping agents is going to change e-commerce in quite a fundamental way.
An agent-to-agent future
The future of shopping is agent-to-agent, agrees Bernadette Nixon, CEO of Algolia, an AI search company. The transaction will happen on the back end, she says. It won’t be a series of blue links. It won’t be a product listing page or a product detail page. It’ll be the transaction. And for that reason, it needs to be seamless. That requires accurate datawhich means public data scraping wont suffice. Just scraping brands or retailers websites doesn’t yield the necessary information to provide a good user experience, she says, because they don’t have accurate pricing. They don’t have accurate inventory.
Protocols and the companies behind them are therefore crucial. Visa is far from the only company in the space: online payments company Stripe has its own Agentic Commerce Protocol, an open standard developed in conjunction with OpenAI.
It all opens up new opportunities for businesses, says Daniel Ruhman, CEO and co-founder of Brazilian fintech Cumbuca, where early AI agent adoption has run ahead of other countries.
You could ask ChatGPT or Claude to find me a handbag, navigate checkout pages, and have your agent handle the payment for you, all with your consent, he says. Thats standard, but agentic AI could go further. Through this, agents can even access your financial data to offer spending insights or advice, he says, what we call agentic open finance, where an AI agent connects to your bank accountwith your permissionto help you understand and manage your money.
The Federal Reserves influence on the economy is immense, and often misunderstood. President of the San Francisco Fed Mary Daly gives an exclusive, firsthand look into the central banks daily decision-making, explaining how the Feds policies, at both the regional and national level, ripple through society. From housing prices to immigrations impact on labor, Daly weighs the major factors shaping the U.S. economy. As political and market pressures mount, she reflects on what it means to lead with discipline and data, and what every business leader can learn from the Feds balancing act.
This is an abridged transcript of an interview from Rapid Response, hosted by the former editor-in-chief of Fast Company Bob Safian. From the team behind the Masters of Scale podcast, Rapid Response features candid conversations with todays top business leaders navigating real-time challenges. Subscribe to Rapid Response wherever you get your podcasts to ensure you never miss an episode.
You run one of the Fed’s 12 regional banks. Your district covers nine Western states, plus Guam, American Samoa, The Mariana Islands. Can you briefly describe the role of your office, and how it relates to the Fed overall? When we hear Fed Chair Jerome Powell announcing a change in interest rates, are you feeding into that? How does all this work?
In 1913, when the Fed was formed, there was a decision that we shouldn’t be Washington-centered. That having a presence in Washington with the Board of Governors was important, but having 12 regional reserve banks was equally important so that we could balance out the decisions about the economy across the country, not just in DC.
So I lead one of the 12 reserve banks, and those reserve banks do feed into monetary policy. We go to each and every FOMC meeting. We are rotating on votes, but we’re always participating. We’re thinking about how our districts with the lived experience in the economy is and how that matters when we make monetary policy.
Monetary policy, the misnomer is it’s all about numbers and markets, but it’s actually about people and lived economy experiences throughout the nation. And so that’s the role of reserve banks, in addition to managing all the operational duties that our teams have, including making sure you have cash when you need it, that your bank can get it and distribute it, making sure the banking system is safe and sound.
You’ve said there’s no politics in the Fed. You’re not funded by the federal government, so a shutdown doesn’t affect you, but everybody tries to influence you guys, policymakers, the White House, investors. How do you keep that politics and that pressure out of what you do?
The founding of the Federal Reserve 1913 had two elements to it that I think have been durable over time and led the way for central banking across the globe. First was that you had to have a regional voice and the second was that you had to be independent. Because monetary policy is made for the longer run and the decisions we make on where to put cash depots and how to distribute our supervision, that’s all got to be done no matter what administration is in place.
So to be durable, especially on the monetary policy part, Congress said let’s make these individuals independent of political persuasion and really thinking about the goals we gave them, and in our case, it’s price stability and full employment, making sure inflation is at 2%, making sure that the economy is not producing lots of unemployment or running so hot, so un-sustainably that inflation should go up. So those are the goals we have.
You asked how do you maintain that? How do you not get influenced? Ultimately, who we work for is the American people. Of course, individuals have points of view and we have to consider those because otherwise we’d just be in an echo chamber. But there’s a difference in listening to understand and listening to be persuaded.
And when the President tries to remove a Fed Governor, as President Trump has done with Lisa Cook, how distracting is that from
It’s really not distracting from the task at hand. Let me just speak about myself. We’re fiduciary stewards of public trust and public responsibilities, and so that’s where I have to attend. Now, I think about not just what’s right in front of me, but ensuring that the American people have a stable and healthy economy over the long term and that the independence of the Fed is preserved not just for the next two months or two weeks, but in fact over the time period, passing that baton to the next generation of leaders.
There’s been so much disruption this year in 2025. Are there particular economic indicators that you are most focused on right now?
So I think about it as a three-legged stool. So the first component is the public data, the things we get from the government, the things that we get on a regular basis. They’re very, very important, but they’re only one part of our overall data collection. We also get data from the private sector. One of the more critical components of that three-legged stool, which is underappreciated in my opinion, is the time that the reserve bank Presidents in particular spend talking to people, to CEOs, small, medium and large businesses, to community members, civic leaders, unions and workers thinking about not just what were they doing last week, but what are they doing going forward.
So right now, I’m very focused on that third leg. And the reason is because when you get to a point where the economy is changing, you have to rely on people who are telling you not what they were doing last week, but what they are doing next week, the next month, the next quarter and ultimately, the next year. And we take that valuable information back to the FOMC meeting. It’s really a robust process and one that I think is critical at these moments.
Obviously, the economy is always changing to some extent, but it certainly feels like we’re at a certain kind of inflection point. I know you’ve rated the sentiment of your region as cautiously optimistic, which is a little incongruous with an economy that seems like it’s moving to something we’re not quite sure where it’s going to go. Can you address that disconnect and maybe explain how and where you see the economy moving?
Absolutely. So there is quite a bit of uncertainty still, not as high as earlier in the year. The uncertainty really spiked after April 2nd, after Liberation Day. There was just so much uncertainty people didn’t know if they were going to be able to buy their smartphone or if they should buy it right away or if they should wait.
Consumers were uncertain. Businesses were uncertain about what’s this going to mean. But at this point, I think those things have settled, and the economy weathered that fairly well. The unemployment rate has gone up a little bit, but not that much. Inflation has gradually come down except in the tariffed sectors.
So the only places where you see prices rising are in the ones direcly affected by tariffs. And so people think of that increasingly as a one time price level adjustment and then they’ll be okay.
Another thing that I think is important is pick a basket of goods that you like to purchase. Put them in a cart at one of your favorite online retailers and then check what has happened to that basket of goods over time. And what you’re seeing is that while certain items have gone up, other ones are being deeply discounted, so people feel like they’re not losing the kind of ground they lost in the big inflation rise after the pandemic. So I think that gives people some confidence.
Recession indicators were quite high and rising earlier in the year and now they’re not really predicting it at all. Consumer sentiment has gone back up after falling, business sentiment has gone back up after falling. So I think that’s where I get the cautious optimism. I was at University of Utah a couple weeks ago and the students are optimistic, and I was really encouraged by that because that generation is like a bellwether. They see that if they learn these new skills, AI, et cetera, they can really make a dent in the economy.
So what is the new economy going to look like? The truth is no one knows, but we do know what the elements will be. Certainly, artificial intelligence is making its way. Is it going to be transformative? Is this going to be the new steam engine or electricity? I don’t know. But it is making a contribution to people’s ability to do things faster, better, cheaper and hopefully, will also make a contribution to us doing things that we never imagined were possible.
With no end in sight for the political standoff that shuttered the federal government, funding for some key programs is drying up.
More than 40 million Americans may not see their food stamps issued next month, as the government shutdown extends into its third full week. Some states have begun warning their residents of the looming threat to the Supplemental Nutrition Assistance Program, better known as SNAP.
Because Republicans in Washington D.C., failed to pass a federal budget, causing the federal government shutdown, November 2025 SNAP benefits cannot be paid, a notification on Pennsylvanias SNAP info page reads.
New York Governor Kathy Hochul demanded that the federal government release funds for SNAP recipients, accusing the Trump administration of deliberately enacting a cruel, senseless and politically motivated punishment that could be avoided.
Im outraged that Washington Republicans are deliberately withholding federal funding from millions of New Yorkers who rely on SNAP to put food on the table, Hochul said in a press release, highlighting the three million New Yorkers who stand to be affected by a SNAP shortfall.
Ronald Ward, the acting associate administrator of SNAP, warned states in a letter that the program was only funded through October, Axios reported earlier this month. That budget shortfall could leave 42 million people without the benefits they rely on, beginning in November. The letter cautioned states to hold off on distributing funds to SNAP recipients EBT cards until further notice.
Blame game
The federal shutdown has turned into a heated blame game, even compared to past shutdown standoffs. At the end of September, Democrats refused to support a bill to fund the federal government, seizing on the rare opportunity for political leverage to demand an extension to the tax credits that reduce the cost of health insurance for millions of Americans. Democrats have also called for Republicans to roll back Medicaid cuts from the Big Beautiful Bill that passed in July.
Because Republicans cant hit the 60 vote threshold needed to fund the government without Democrats, the shutdown is a stalemate unless one side backs down.
The Trump administration has taken extraordinary measures to associate the shutdown with his political opposition, even ordering airports to play a video of Secretary of Homeland Security Kristi Noem blaming Democrats for shutdown-related travel delays. Many airports refused to air the video, citing policies against displaying political content.
Noem isnt the only member of Trumps cabinet to spread that messaging. Democrats are putting free healthcare for illegal aliens and their political agenda ahead of food security for American families, Agriculture Secretary Brooke Rollins said on X, blaming what she referred to as the Democrat shutdown.
During the shutdown, some government websites are displaying unusually partisan messages. The USDAs website is currently topped by a banner noting that it wont be updated and blaming the Radical Left Democrat shutdown. President Trump has made it clear he wants to keep the government open and support those who feed, fuel, and clothe the American people, the message reads.
Selective funding
Most of the federal government is shuttered in light of the political standoff, but the Trump administration is finding ways to fund its own political priorities.
Trump ordered the Pentagon and the White House to use all available funds to pay active-duty members of the military, avoiding the political fallout of servicemembers missing paychecks. The White House also opted to fund the Special Supplemental Nutrition Program for Women, Infants and Children, better known as WIC, using money collected from tariffs. “The Trump White House will not allow impoverished mothers and their babies to go hungry because of the Democrats’ political games, White House Press Secretary Karoline Leavitt told Axios.
The Democrats are so cruel in their continual votes to shut down the government that they forced the WIC program for the most vulnerable women and children to run out this week.Thankfully, President Trump and the White House have identified a creative solution to transfer https://t.co/tj9Xt7f4yQ— Karoline Leavitt (@PressSec) October 7, 2025
Trump went around Congress to allocate those funds, but Congress also has the ability to selectively dole out cash for programs that would otherwise have their funding cut off during a federal shutdown. Still, the food stamps program may not be a priority for Republicans given the partys willingness to slash SNAP dramatically to fund tax cuts and defense spending in the massive bill that passed this summer.
The federal shutdowns short-term hit to SNAP could be devastating for Americans who rely on the program to put food on the table, but lasting changes to the program mean fewer Americans will be eligible for food assistance when the spigot of federal funds does eventually open back up.
Hannah Elsakr says that Adobe’s top clientsthe owners of some of the most protected, most valuable brands and IP in the worldhad a stark message for the company regarding Firefly, its generative AI engine: They wanted more, and they wanted better.
“They told us they actually needed models that understood all their products, all their brands, their creative direction,” says Elsakr, Adobe’s VP of GenAI New Business Ventures. “They have characters, they have particular motion styles, and they needed us to train on that.”
Fireflywhich uses prompts to create assets across all Adobes vector, bitmap, and motion appscouldn’t do this because it doesnt understand brands at the IP level. “We consider that a feature, not a bug,” Alsark tells me. Firefly is a generic engine, but Adobes top clients need to create millions of assets for dozens of different platforms and marketing media, all of them conforming to their own strict IP rules and brand books.
This is why the company has built a new consultancy arm for Fortune 2000 companies to develop bespoke AI models that could craft hundreds of thousands of images, illustrations, and videos that conform strictly to their IP and creative guidelines. Its name: Adobe AI Foundry.
[Image: Adobe]
The million-asset problem
The move comes as a direct response to a mathematical nightmare facing every major brand. Alsark calls it the “combinatorics math problem”. A company with just eight products that wants to market them across 15 channels in 35 languages with a few refreshes a year is already looking at creating half a million individual assets. “With social, we know we’re doing probably three refreshes a week,” she explains. “So the real numbers are in the millions and millions and millions.”
Before now, that scale was simply impossible. Time, budgets, and human resources are all finite. “The only unlock here is responsible AI,” Alsark insists. This crushing demand from the attention economy is precisely why Adobes biggest clients, companies like The Home Depot and Walt Disney Imagineering, came to them. They needed an industrial-scale solution, but one that respected their most valuable assettheir intellectual property.
Adobes public Firefly model was a start, but it was designed to be IP-agnostic. While brands applauded this safety-first approach, they needed an AI that could learn their worldtheir characters, their color palettes, their unique aesthetic. Last year’s self-serve “Custom Models” were a step in that direction, allowing companies to train the AI on a single concept, like a specific style or shape. But clients wanted more. They wanted the whole kingdom, not just one castle.
[Image: Adobe]
A bespoke AI partnership
Adobe AI Foundry isn’t a piece of software you buy off the shelf; it’s a deep, consultative partnership that feels more like hiring a boutique division of AI experts from Accenture or IBM than licensing a tool. Adobe targets the Fortune 2000, a customer base where it already has deep roots through its Creative Cloud and marketing software suites.
“You actually get a team of allocated experts from Adobe,” Alsark says, listing “applied scientists, engineers, [and] creative workflow experts.” These teams work directly with a client to build a unique generative AI model from the ground up, trained exclusively on that company’s private data and assets.
The process is intensive and can take a couple of months just to get the first results. It begins with use-case discovery, where Adobes team identifies the core business problem, whether its creating seasonal ad campaigns for a retailer like Amazon Fresh or generating limitless variations of a hero image for a beauty brand like MAC.
Then comes the heavy lifting. Adobes engineers surgically reopen their foundational Firefly models to retrain them on the clients proprietary IP, a process involving billions of parameters. After months of training and untold GPU cycles and consumed watts, the final model inherits all the world knowledge of the base Firefly engine but is then overfitted to speak the client’s brand language fluently and exclusively. The output is locked down; it belongs to the client and will never be mixed with another company’s data.
The trust factor
The entire proposition hinges on two things Adobe believes it alone can offer at this scale: commercial safety and seamless integration. The company has been careful to build its foundational models on licensed Adobe Stock data, shielding its clients from the copyright nightmares that have plagued other AI models. Foundry extends that protection by creating a secure vault for a brands own IP.
This focus on safety and its existing enterprise presence is Adobes strategic moat against competitors like Canva, which is also aggressively pursuing the corporate market. When I bring up the competition, Alsark doesn’t seem to be concerned. She claims that clients came to Adobe after experimenting with everything from startups to hyperscalers because they trust Adobe to understand the entire creativity landscape.
“They are already deeply into our creative tools. We’re in their marketing stack, and we are enterprise-grade,” she says. The ability to plug a custom, brand-aware AI model directly into the Photoshop, Illustrator, and After Effects workflows that a company&8217;s creative teams already live in is a killer advantage. Molly Battin, the CMO at The Home Depot, an early Foundry customer, calls it “an exciting step forward in embracing cutting-edge technologies to deepen customer engagement.”
For an old guy like me, tired of seeing Silicon Valley promising revolutions every other week, this feels different. The initial AI craze, as I called it in my chat with Alsark, is still about public-facing tools that felt like creative toys. For most of us, anyway. Adobe AI Foundry represents the next, far more serious phase: AI is being forged into a bespoke, industrial-grade weapon for the worlds biggest brands.
Its no longer about a single person creating a wild image or helping with a creative roadblock; its about a corporation generating a million on-brand assets before lunch. Its less of a craze and more of a quiet, brutally efficient corporate takeover of the creative process.
Stocks are climbing on Wall Street Monday and pulling near their records following last weeks roller-coaster ride.
The S&P 500 rose 1% and got back within 0.4% of its all-time high set earlier this month. The Dow Jones Industrial Average was up 358 points, or 0.8%, and the Nasdaq composite was 1.4% higher just before noon Eastern time.
Cleveland-Cliffs helped lead the way with a jump of 24% after the steel companys CEO, Lourenco Goncalves, said it would provide details soon about a potential deal with a major global steel producer that could mean bigger profits. He also said Cleveland-Cliffs has potentially found rare earths at sites in Michigan and Minnesota.
Such materials have thrust into the global spotlight after China put curbs on the export of its own rare earths, a move that President Donald Trump earlier this month characterized as hostile. Trumps ensuing threat of higher tariffs on China triggered big swings for Wall Street, but the concerns eased a bit after Trump said such high tax rates are unsustainable.
Another source of worry for Wall Street, from the banking industry, also appears to be easing. Stocks of smaller and midsize banks climbed Monday, recovering some of their losses after a couple raised alarm bells last week by warning about potentially bad loans theyve made.
The disclosures had raised questions about whether the growing list of problems is just a collection of one-offs or a signal of something larger threatening the entire industry.
Zions Bancorporation rose 2.5% following its 5.1% drop last week. It will report its latest quarterly earnings after trading ends for the day, and scrutiny will be high after it said its charging off $50 million of loans where it found apparent misrepresentations and contractual defaults by the borrowers.
This will be a heavier week for corporate earnings reports generally. Big names delivering their latest results will include Coca-Cola on Tuesday, Tesla on Wednesday, and Procter & Gamble on Friday.
The pressure is on companies to show that their profits are growing because they need to justify the big gains their stock prices have made. The S&P 500 is still near its all-time high, which was set earlier this month following a torrid 35% run from a low in April.
Delivering bigger profits is one of the easiest ways for companies to quiet criticism that stock prices have gone too high. The other is for stock prices to fall.
Corporate profit reports have also taken on more importance because theyre offering windows into the strength of the U.S. economy when the U.S. governments shutdown has delayed many important economic updates.
Thats making the job of the Federal Reserve more difficult, as it tries to decide whether high inflation or the slowing job market is the bigger problem for the economy. Fed officials have indicated theyre likely to cut interest rates a few more times through next year in order to give the economy a boost. But that could be a mistake if inflation worsens, because low interest rates can push prices even higher.
On Friday, the U.S. government will issue an update for inflation during September. The report was supposed to arrive earlier in month, and the Social Security Administration needs the numbers to calculate cost-of-living adjustments for beneficiaries.
But the government said, No other releases will be rescheduled or produced until the resumption of regular government services.
In the bond market, Treasury yields held relatively steady. The yield on the 10-year Treasury eased to 3.99%, from 4.02% late Friday.
Treasury yields have been falling recently, and lower yields help make stock prices look less expensive by encouraging some investors to buy stocks when they otherwise would have bought bonds.
On Wall Street, Amazons stock held up despite a widespread outage for its cloud computing service that caused disruption for internet users around the world early Monday. Amazons stock rose 0.8%.
In stock markets abroad, indexes rose across much of Europe and Asia.
Japans Nikkei 225 jumped 3.4%, after its governing Liberal Democrats found a new coalition partner, securing support for its leader Sanae Takaichi to become the countrys first female prime minister. Investors expect Takaichi to push for low interest rates, higher government spending, and other policies that could help the market.
Indexes rose 2.4% in Hong Kong and 0.6% in Shanghai after China reported its economy grew at a 4.8% annual pace in the last quarter, supported by relatively strong exports as companies increased shipments markets other than the U.S.
Still, it was the slowest pace in a year. The worlds second-largest economy is still struggling to emerge from a prolonged downturn in its property market and to encourage consumers and businesses to spend more.
By Stan Choe, AP business writer
AP Business Writers David McHugh and Elaine Kurtenbach contributed.
President Donald Trump and Australian Prime Minister Anthony Albanese signed a critical-minerals deal at the White House on Monday as the U.S. eyes the continents rich rare-earth resources when China is imposing tougher rules on exporting its own critical minerals abroad.
The two leaders described the agreement as an $8.5 billion deal between the allies. Trump said it had been negotiated over several months.
Todays agreement on critical minerals and rare earths is just taking the U.S. and Australia’s relationship to the next level,” Albanese added.
This month, Beijing announced that it will require foreign companies to get approval from the Chinese government to export magnets containing even trace amounts of rare-earth materials that originated from China or were produced with Chinese technology. Trump’s Republican administration says this gives China broad power over the global economy by controlling the tech supply chain.
Australia is really, really going to be helpful in the effort to take the global economy and make it less risky, less exposed to the kind of rare earth extortion that were seeing from the Chinese, Kevin Hassett, the director of the White Houses National Economic Council, told reporters on Monday morning ahead of Trumps meeting with Albanese.
Hassett noted that Australia has one of the best mining economies in the world, while praising its refiners and its abundance of rare earth resources. Among the Australian officials accompanying Albanese are ministers overseeing resources and industry and science, and Australia has dozens of critical minerals sought by the U.S.
The prime minister’s visit comes just before Trump is planning to meet with Chinese President Xi Jinping in South Korea later this month.
For Albanese’s part, the prime minister said ahead of his visit that the two leaders will have a chance to deepen their countries’ ties on trade and defense. Another expected topic of discussion is AUKUS, a security pact with Australia, the U.S., and the United Kingdom that was signed during U.S. President Joe Bidens Democratic administration.
Trump has not indicated publicly whether he would want to keep AUKUS intact, and the Pentagon is reviewing the agreement.
Australia and the United States have stood shoulder to shoulder in every major conflict for over a century, Albanese said ahead of the meeting. I look forward to a positive and constructive meeting with President Trump at the White House.
The center-left Albanese was reelected in May and suggested shortly after his win that his party increased its majority by not modeling itself on Trumpism.
Australians have chosen to face global challenges the Australian waylooking after each other while building for the future, Albanese told supporters during his victory speech.
By Seung Min Kim and Aamer Madhani, Associated Press
