Another highly anticipated IPOs is happening today. Cryptocurrency trading platform Bullish is expected to make its debut on the stock market.
Already, its stocks initial public offering price has been revised up multiple times, suggesting a healthy appetite for shares. However, only time will tell if its IPO will end up going the way of Circle or eToro, two other crypto-related companies that went public in recent months.
Heres what you need to know about Bullishs IPO.
What is Bullish?
Founded in 2021, Bullish is a cryptocurrency exchange that focuses on spot and derivatives trading. Its CEO is Thomas Farley, who was previously the president of the NYSE Group, which operates the New York Stock Exchange.
Bullish received an early investment from VC heavyweight Peter Thiel.
The company says it has more than 275 employees. For its most recent quarter, which was its Q1 2025 period, Bullish says it had $647 billion in total spot volume, according to its Form F-1 filing with the Securities and Exchange Commission (SEC).
Additionally, Bullish says it had $111 billion in total perpetual futures volume for the quarter.
The company said its average daily volume totals around $2.6 billion, and that trading volume for Q1 2025 was up 78% versus the same quarter a year earlier.
However, Bullish isnt only a crypto exchange. In 2023, the company purchased the crypto media outlet CoinDesk from the Digital Currency Group. CoinDesk, according to Bullish, now has 48 million unique visitors annually. It also receives more than 600 million monthly API calls to its CoinDesk Data offerings.
Bullish IPO share price rises yet again
One already notable thing about Bullishs IPO is that ever since the company announced its intention to go public, its estimated IPO share price has continually risen.
As Fast Company reported yesterday, Bullish initially said its shares would be offered for $28 to $31 each. But on Monday, the company upped its estimated initial share price to between $32 and $33 each.
That estimated initial share price was again revised upwards later in the day yesterday, to its final IPO offering price of $37 per share.
In addition to its increased IPO price, Bullish also raised the number of shares it offered in its IPO. Initially, Bullish said it would issue 20.3 million shares in its IPO. Its final share issue was nearly 50% higher, at 30 million shares.
In its Form F-1 filed with the SEC, Bullish says its plans to use the proceeds from its IPO for general corporate and working capital purposes, including funding potential future acquisitions.
When is Bullishs IPO?
Bullish priced its shares yesterday and is expected to begin trading today: Wednesday, August 13, 2025.
What is Bullishs stock ticker?
Bullishs stock will trade under the ticker BLSH.
What exchange will Bullish shares trade on?
Bullish shares will trade on the New York Stock Exchange (NYSE).
What is the IPO share price of BLSH?
Bullishs IPO price is $37 per share.
As previously noted, thats a significant increase from the $28 to $31 price range that Bullish shares were forecast initially to begin trading at. Its also higher than BLSHs revised expected range of between $32 and $33 each.
The fact that BLSH shares have been continually revised upwards suggests there is strong interest in the stock.
How many BLSH shares are available in its IPO?
In the companys latest press release on the matter, Bullish announced that it is offering 30,000,000 ordinary shares in its IPO.
How much did Bullish raise in its IPO?
With 30 million shares sold at $37 each, it means that Bullish raised approximately $1.11 billion from its IPO.
How much is Bullish worth?
Bullish is now valued at around $5.4 billion, according to Reuters.
How have other crypto IPOs done this year?
Bullishs IPO will be closely watched today as its performance will be used to judge investor confidence in other potential cryptocurrency-related IPOs.
In general, investors have been more bullish (no pun intended) on cryptocurrencies and crypto-related companies since President Donald Trump took office.
Trumps administration is seen as being much more friendly to the crypto industry, and it has passed significant laws giving clarity to the regulations surrounding the industry.
When it comes to IPOs, this crypto-friendly climate has perhaps benefited no other company more than Circle Internet Group (NYSE: CRCL), an issuer of stablecoins.
Since CRCL shares went public in early June, they have surged more than 400% as of yesterdays close (at one point, they were up more than 750%).
However, while Circle shares have surged, another crypto company has seen less stellar results since its IPO.
In May, crypto trading platform eToro Group Ltd. (Nasdaq: ETOR) went public. But according to Yahoo Finance data, as of yesterdays market close, ETOR shares have lost 2.4% of their value since its IPO.
Many on Wall Street will be eagerly watching Bullishs stock price over the next several weeks to see if it will follow in the footsteps of Circle or eToroor somewhere in between.
The nations electronic vehicle charging network is getting more reliableand Tesla Superchargers are leading that advancement.
To get more people into electric vehicles, we need a robust EV charging network. That doesnt just mean having accessible chargers; it means public chargers must be reliable, offer a good customer experience, and can’t be too expensive.
For years, public chargers have been plagued with reliability issues. A 2024 study found that one in five EV charging attempts at public stations fail.
Imagine if you go to a traditional gas station and 2 out of 10 times the pumps are out of order, study lead Omar Asensio told Harvard Business School at the time. Consumers would revolt.
President Donald Trump has also thrown the state of the EV charging network into question. In February, he paused funding for the National Electric Vehicle Infrastructure (NEVI) program, cutting off millions of dollars for EV chargers. (This week, Trump allowed NEVIs $5 billion in funding to keep flowing to states, with revised guidelines.)
Still, despite that pause, the EV charging network has been growing. And theres been progress with the state of those public EV chargers, according to J.D. Power, which today released findings from its 2025 U.S. Electric Vehicle Experience Public Charging Study.
‘Non-charging visits’ are declining
The top takeaway is that the EV charging network is the most reliable its been in years. Just 14% of EV owners say they visited a charger without successfully charging their vehicle.
A non-charging visit, as J.D. Power calls it, could occur because of wait times at public chargers, but the majority of such visits happen because the charger is out of service or not working properly.
Thats down five percentage points from 2024, and marks the lowest level in four years. It also matches the all-time low since J.D. Power began tracking these visits in early 2021.
Tesla Superchargers have the lowest percentage of failed charging visits, at just 4%. Other charging companies including Electrify America (6%), Red E (10%), and EVgo (12%) were all below that average for failed visits as well.
Tesla chargers: Satisfaction is high but declining
Tesla also leads the EV charging companies in terms of customer satisfactionbut that satisfaction has dipped.
On a 1,000-point scale, Tesla Superchargers have a customer satisfaction with a score of 709the top of the J.D. Power study, but also a 22-point drop from last year.
Other EV charging networks, including the Mercedes-Benz Charging Network, Rivian Adventure Network, and Ford Charge collectively earned a satisfaction score of 709 as well, but these werent ranked in the study because of their limited footprint.
J.D. Power measures satisfaction across multiple factors, like speed, the physical condition of charging station, things to do while charging, safety, cost, and how easy it is to pay.
‘Very easy to use’if you drive a Tesla
Teslas drop in satisfaction seems to come mostly from non-Tesla owners.
Tesla has facilitated an experience for its owners by creating an optimal technical environment that makes the charging process very easy to use and complete payments, says Brent Gruber, executive director of the EV practice at J.D. Power. That process isnt quite as streamlined for non-Tesla owners.
Non-Tesla owners using the Superchargers are particularly less satisfied than Tesla owners with the cost of charging; non-Tesla drivers are often charged a higher per-kilowatt-hour rate.
Tesla has faced a challenging year, with sales plunging as EV drivers look to distance themselves from CEO Elon Musk. Tesla owners have also been trading in their EVs at all-time high levels.
Charging costs are a concern for EV drivers across vehicle and charging types, though. Satisfaction with charging costs for both Level 2 and DC fast chargers dropped to 459 (down 16 points) and 430 (down 16 points), respectively, on a 1,000-point scale.
Some of that may be because new fast charging networks kept their prices low as they built out their presence, and because free charging was sometimes a perk for new EV purchases.
Electricity rates have also been rising across the country, which affects EV charging prices.
With artificial intelligence progressing at a rate that can generously be described as unprecedented, many students considering college are wondering what a traditional degree may have to offer in these rapidly changing times.
In fact, AI skills and background education were top of the list of factors taken into account as LinkedIn assembled its first-ever list of top colleges, which was released on August 12.
While the list also prioritizes other categories culled from its dataincluding hiring rates, job mobility, and alumni networksthe special breakout category that many will surely be most interested in is the one that shows the highest percentages of grads entering the workforce in AI-focused jobs.
Many grads are probably concerned about the ability of traditional four-year programs to equip them for this new job landscape, which is understandable as most professionals are currently unclear about how it will impact their companies futures.
According to LinkedIn, 81% of C-suite executives are favoring job candidates that are comfortable with AI tools.
Which schools are most AI focused?
The California Institute of Technology (Caltech) comes in at the top of two lists, including the highest percentage of recent graduates going into AI fields, and the highest percentage of new grads listing AI literary and engineering skills on their LinkedIn profiles.
LinkedIn also found that Caltech offers a range of AI-focused classes and overall programs, including AI4Science initiative, which the website describes as an initiative aimed at bringing AI researchers with experts from other disciplines to push modern AI tools into every area of science and engineering.
The Massachusetts Institute of Technology (MIT) came in second in both categories, with University of California-Berkeley, Stanford, Carnegie Mellon University, Harvey Mudd, and University of California-San Diego also appearing on both lists.
Schools with the highest percentage of recent grads going into AI occupations
California Institute of Technology
Massachusetts Institute of Technology
Carnegie Mellon University
Harvey Mudd College
University of California-Berkeley
Stanford University
University of Rochester
University of California-San Diego
University of Chicago
Harvard University
Schools with the highest percentage of recent grads adding AI-related skills on LinkedIn profiles
California Institute of Technology
Massachusetts Institute of Technology
Harvey Mudd College
Stanford University
Carnegie Mellon University
University of California-Berkeley
University of California-San Diego
Brown University
Georgia Institute of Technology
Princeton University
While these two breakout categories are particularly relevant for those considering tech fields, the larger list looks at more traditional considerations for students, including job placement, recruiter demand, career success, network strength, and knowledge breadth. They also specify notable skills listed by grads from the top schools.
For the overall list, Princeton took home the top spot, with Duke and the University of Pennsylvania taking the second and third spots. The entire list can be found on LinkedIns website.
Ten years from now, it will be clear that the primary ways we use generative AI circa 2025rapidly crafting content based on simple instructions and open-ended interactionswere merely building blocks of a technology that will increasingly be built into far more impactful forms.
The real economic effect will come as different modes of generative AI are combined with traditional software logic to drive expensive activities like project management, medical diagnosis, and insurance claims processing in increasingly automated ways.
In my consulting work helping the worlds largest companies design and implement AI solutions, Im finding that most organizations are still struggling to get substantial value from generative AI applications. As impressive and satisfying as they are, their inherent unpredictability makes it difficult to integrate into the kind of highly standardized business processes that drive the economy.
Agentic vs. Interpretive
Agentic AI, which has been getting tremendous attention in recent months for its potential to accomplish business tasks with little human guidance, has similar limitations. Agents are evolving to assist with singular tasks such as building websites quickly, but their workflows and outputs will remain too variable for large organizations with high-volume processes that need to be predictable and reliable.
However, the same enormous AI models that power todays best-known AI tools are increasingly being deployed in another, more economically transformative way, which I call interpretive AI. And that is whats likely to be the real driver of the AI revolution over the long term.
Unlike generative and agentic AI, interpretive AI lets computers understand messy, complex, and unstructured information and interpret it in predictable, defined ways. Using much of the same IT infrastructure, the emerging technology can power large organizations complex processes without requiring human intervention at each step.
Use cases
Some interpretive AI applications are already in use. For example, doctors are saving significant time by using interpretive AI tools to listen to conversations with patients and fill in information on their electronic health record interfaces to track care and facilitate billing. In the near future, the technology could determine fault in auto accidents based on police reports written in any of thousands of different formats, or process video recorded from a laptop screen as someone edits a presentation to provide teammates with an automated update on work completed. The applications are wide-ranging and span all manner of industries.
Based on estimates for areas such as coding and marketing where generative AI is most applicable, interpretive AI could unlock 20% to 40% productivity gains for the half of GDP that comes from large corporations. First, though, they must commit to developing a comprehensive, long-term strategy involving multiple business functions and careful experimentation, and change entrenched processes and work culture norms that slow its adoption. Done right, the obstacles are surmountableand the payoff could be massive.
A different application of generative AI models
One of the most basic drivers of economic growth is the ongoing effort to standardize and scale up a particular process, making it faster, cheaper, and more reliable. Think of factory assembly lines enabling mass production, or the internets codification of computer communication protocols for use across disparate networks.
Generative AI has been, on the whole, disappointing when it comes to automation. For example, many firms have tried to use generative AI chatbots to reduce the time their human resources staff spends answering employees questions about internal policies. However, the open-ended output from such systems requires human review, rendering the labor savings modest at best. The technology seems to inherit much of the unpredictability of humans along with its ability to mimic their creative and reasoning skills.
Agentic AI promises to do complicated work autonomously, with smart AI agents developing and executing plans for achieving goals step-by-step, on the fly. But again, even when agents become smart enough to help a typical knowledge worker be more productive, their outputs will be quite variable.
Enter interpretive AI. For the first time, computers can usefully process the meaning of human language, with all its nuance and unspoken context, thanks to the unprecedentedly large models developed by firms like Open AI and Google. Interpretive AI is the mechanism for using the models to exploit this revolutionary advance.
Until now, computers ability to capture, store, aggregate, summarize, and evaluate a large organizations activities were limited to those that were easy to quantify with data. Interpretive AI can quickly and precisely execute these functions for many other important activities, at a vast scale and at minimal marginal cost. For instance, no longer will businesses need manual processes to monitor and manage levels of activity and progress in knowledge-worker tasks such as coding a feature into a software solution or developing a set of customer-specific outreach strategies, which usually require dedicated middle management staff to collect information.
Companies can make productivity gains by using interpretive AI for a range of other previously hard-to-measure employee issues as well, including the tone and quality of their interactions with customers, their cultural norms in the workplace, and their compliance with office policies and behavioral expectations.
Transforming the management of knowledge work
The use of interpretive AI will enable the widespread transformations that unlock newly efficient ways of working at large organizations (which are responsible for organizing and producing most of the worlds goods and services). It will dramatically reduce the need for extensive, costly, slow-moving, and unenjoyable middle management work to coordinate complex interrelated programs of activities across teams and disciplines.
Even better, it can efficiently understand operationally vital but opaque aspects of how work happens, such as the decades worth of legacy code and data that make even minor technology process changes time-consuming and challenging for any long-lived enterprise.
Of course, interpretive AI is not mutually exclusive with generative and agentic AIagain, its simply a different way to use the powerful models that power those technologies. A decidedly unsexy way, certainly, but for businesses looking for ways to maximize the economic impact of AI over the next ew years, its just the unsexy they need.
A new study from Cornell University goes against the grain of popular thought, arguing that left-handed people aren’t necessarily more creative than their right-handed counterparts after all.
It’s research that hits close to home for this writer. From an early age, I’ve worn my left-handedness as a badge of pride. As a kid, I always felt different from the other students in class, because I had to use a left-handed desk. Back then, I also had to use special scissors in home economics, bat on the “wrong” side of the plate at softball . . . the list goes on.
But despite the minor inconveniences, it was a label I readily embraced because I was told I was “special” (only 10% of the population is left-handed) and, perhaps most of all, because I knew I was in good company.
Who wouldn’t want to be a member of a club that includes Michelangelo, Leonardo da Vinci, Aristotle, one of the Beatles, Bill Gates, Nikola Tesla, Marie Curie, Babe Ruth, Bart Simpson, Oprah, and Jerry Seinfeld? In fact, five out of the last eight presidents have been left-handed: Gerald Ford, Ronald Reagan, George H.W. Bush, Bill Clinton, and Barack Obama. (President Trump is a rightie.)
To this day, I still make a mental note of who is and is not a lefty. Picasso and John Lennon aren’t, but Paul McCartney is. So is my best friend, Gaby, my editor, Connie, and my boss, Christopher. It’s a secret club we lefties share, believing there is something just a little special, a little more creative about us.
That’s why the new research from Cornell stopped me in my tracks.
The science of creativity
In Handedness and Creativity: Facts and Fictions, published in the Psychonomic Bulletin & Review, researchers argue that while there’s a plausible link between creativity and handedness based on theories that look at the neural basis of creativity, they found no evidence that left- or mixed-handed individuals are more creative than right-handers. In fact, they even found right-handers scored statistically higher on one standard test of divergent thinking (the alternate-uses test).
“The data do not support any advantage in creative thinking for lefties, said the studys senior author, Daniel Casasanto, associate professor of psychology at Cornell.
And while the Cornell researchers acknowledge that left- and mixed-handers may be overrepresented in art and music, they argue that southpaws are underrepresented in other creative professions, like architecture.
When determining which professions constitute creative fields, researchers drew on data from nearly 12,000 individuals in more than 770 professions, which were ranked by the creativity each requires. By combining originality and inductive reasoning, they concluded that physicists and mathematicians rank alongside fine artists as having the most creative jobs. Using this criteriaand considering the full range of professionsthe researchers found that left-handers are underrepresented in fields that require the most creativity.
The focus on these two creative professions where lefties are overrepresented, art and music, is a really common and tempting statistical error that humans make all the time, Casasanto said. People generalized that there are all these left-handed artists and musicians, so lefties must be more creative. But if you do an unbiased survey of lots of professions, then this apparent lefty superiority disappears.
Casasanto did agree, however, that there are scientific reasons to believe that left-handed people would have an edge in creativity when it comes to “divergent thinking”the ability to explore many possible solutions to a problem in a short time and make unexpected connectionswhich is supported more by the brains right hemisphere. But again, the study revealed that handedness makes little difference in the three most common laboratory tests of its link to divergent thinking; if anything, righties have a small advantage on some tests.
Finally, researchers conducted their meta-analysis by crunching the data from nearly 1,000 relevant scientific papers published since 1900. Most were weeded out because they did not report data in a standardized way, or included only righties (the norm in studies seeking homogeneous samples), leaving just 17 studies reporting nearly 50 effect sizes. This may be why the newest study came to a different conclusion than what is held in popular belief or prior scientific literature.
A new study from Cornell University goes against the grain of popular thought, arguing that left-handed people aren’t necessarily more creative than their right-handed counterparts after all.
It’s research that hits close to home for this writer. From an early age, I’ve worn my left-handedness as a badge of pride. As a kid, I always felt different from the other students in class, because I had to use a left-handed desk. Back then, I also had to use special scissors in home economics, bat on the “wrong” side of the plate at softball . . . the list goes on.
But despite the minor inconveniences, it was a label I readily embraced because I was told I was “special” (only 10% of the population is left-handed) and, perhaps most of all, because I knew I was in good company.
Who wouldn’t want to be a member of a club that includes Michelangelo, Leonardo da Vinci, Aristotle, one of the Beatles, Bill Gates, Nikola Tesla, Marie Curie, Babe Ruth, Bart Simpson, Oprah, and Jerry Seinfeld? In fact, five out of the last eight presidents have been left-handed: Gerald Ford, Ronald Reagan, George H.W. Bush, Bill Clinton, and Barack Obama. (President Trump is a rightie.)
To this day, I still make a mental note of who is and is not a lefty. Picasso and John Lennon aren’t, but Paul McCartney is. So is my best friend, Gaby, my editor, Connie, and my boss, Christopher. It’s a secret club we lefties share, believing there is something just a little special, a little more creative about us.
That’s why the new research from Cornell stopped me in my tracks.
The science of creativity
In Handedness and Creativity: Facts and Fictions, published in the Psychonomic Bulletin & Review, researchers argue that while there’s a plausible link between creativity and handedness based on theories that look at the neural basis of creativity, they found no evidence that left- or mixed-handed individuals are more creative than right-handers. In fact, they even found right-handers scored statistically higher on one standard test of divergent thinking (the alternate-uses test).
“The data do not support any advantage in creative thinking for lefties, said the studys senior author, Daniel Casasanto, associate professor of psychology at Cornell.
And while the Cornell researchers acknowledge that left- and mixed-handers may be overrepresented in art and music, they argue that southpaws are underrepresented in other creative professions, like architecture.
When determining which professions constitute creative fields, researchers drew on data from nearly 12,000 individuals in more than 770 professions, which were ranked by the creativity each requires. By combining originality and inductive reasoning, they concluded that physicists and mathematicians rank alongside fine artists as having the most creative jobs. Using this criteriaand considering the full range of professionsthe researchers found that left-handers are underrepresented in fields that require the most creativity.
The focus on these two creative professions where lefties are overrepresented, art and music, is a really common and tempting statistical error that humans make all the time, Casasanto said. People generalized that there are all these left-handed artists and musicians, so lefties must be more creative. But if you do an unbiased survey of lots of professions, then this apparent lefty superiority disappears.
Casasanto did agree, however, that there are scientific reasons to believe that left-handed people would have an edge in creativity when it comes to “divergent thinking”the ability to explore many possible solutions to a problem in a short time and make unexpected connectionswhich is supported more by the brains right hemisphere. But again, the study revealed that handedness makes little difference in the three most common laboratory tests of its link to divergent thinking; if anything, righties have a small advantage on some tests.
Finally, researchers conducted their meta-analysis by crunching the data from nearly 1,000 relevant scientific papers published since 1900. Most were weeded out because they did not report data in a standardized way, or included only righties (the norm in studies seeking homogeneous samples), leaving just 17 studies reporting nearly 50 effect sizes. This may be why the newest study came to a different conclusion than what is held in popular belief or prior scientific literature.
If youve ever been a patient waitingdays, sometimes more than a weekfor treatment approval, or a clinician stuck chasing it, you know what prior authorization feels like. Patients sit in limbo, anxiety growing as care stalls. Nurses and physicians trade hours of patient time for phone calls, faxes, and glitchy portals. Everyone waits, some in pain, while the people on both sides of the system lose faith in it a little more each day.
This isnt a minor inconvenience. According to the American Medical Associations (AMA) 2024 Prior Authorization Physician Survey, 93% of physicians report that prior authorization delays access to care, and 94% say it negatively affects patient outcomes. Physicians handle an average of nearly 40 requests each week, spending 13 hours of their time on the process. Nearly 9 out of 10 share that its a contributing factor to burnout.
Weve been down this road for many years. In 2018, health insurer and provider groups signed a consensus statement promising significant improvement of prior authorization. In 2023, the AMA reported that two major insurers pledged to reduce the number of services needing prior authorization. The promises added upso why hasnt the burden eased?
Last month, AHIP (a national trade association representing the health insurance industry) and major payers rolled out six reforms standardizing electronic submissions, speeding decisions, improving transparency, and preserving continuity of care when members switch plans. More than 50 leading health plans, encompassing 257 million Americans, signed on to be a part of this reform with all commitments delivered by 2027. Thats important. But for someone awaiting chemotherapy or a nurse on the ward, 2027 feels like forever away.
No need to wait
Real reform doesnt need to wait. Heres whats already happening inside health plans that have embraced agentbased AI systemstechnology is being put in place that’s designed not just to speed up forms but to fundamentally change how prior authorization and other core operations get done. These systems dont replace people. They work alongside nurses, case managers, and administrators, handling the repetitive, document-heavy work so humans can focus on clinical decisions and patient care.
Tangible results
This AI is just beginning to be adopted by plans; the transformation is measurable:
Turnaround times are slashed by more than 50%.
76% of authorizations are handled automatically.
Eighteen minutes are saved per prior authorization request, which for an average-sized health plan unlocks tens of thousands of hours each month, that enables clinical care teams to shift from administrative work to greater focus on patients.
AI can handle the tedious parts of the processsorting through PDFs, faxes, and clinical notesin seconds instead of hours, while keeping nurses and physicians involved for the decisions that require human judgment.
Paradigm shift
This isnt just a tweak to the old processits a shift that allows entire operations teams to work differently. Patients get quicker answers and fewer anxious phone calls. Providers get back more time to spend helping patients. For the people working behind the scenes, it means moving past the repetitive paperwork grindhours spent sorting through forms, faxes, and filesand focusing on work that actually supports better care.
AHIPs own language makes the case: these reforms are meant to provide faster access to evidence-based care, simplify workflows for providers, and preserve care continuity when people switch insurers. That should be the minimum standardnot a distant promise.
Health plans need to act nowscaling existing AI deployments and embracing process redesignso they can deliver three concrete outcomes today:
Patients can start the treatments they need sooner, without the constant backandforth or long waits for approval.
Doctors and nurses get back valuable time, so they can spend more of their day with patients instead of buried in forms and phone calls.
People can keep their care moving, even if they switch insurance plans midtreatment, without having to start the approval process all over again.
Within reach
Health plans dont need to wait for 2027. The technology exists today to deliver meaningful prior authorization reform. And lets be honest: another round of press releases wont change outcomes. But scaling the results and impact that AI is delivering to prior authorization todaythats reform in motion. And its within reach.
For patients trapped in limbo and clinicians stretched thinner every day, reform cant arrive soon enough. The question isnt whether we can make prior authorization faster, simpler, and less needed. Its when health plans will actfor the sake of all those involved in our health system, we must choose urgency.
Over the past five years, advances in AI models data processing and reasoning capabilities have driven enterprise and industrial developers to pursue larger models and more ambitious benchmarks. Now, with agentic AI emerging as the successor to generative AI, demand for smarter, more nuanced agents is growing. Yet too often smart AI is measured by model size or the volume of its training data.
Data analytics and artificial intelligence company Databricks argues that todays AI arms race misses a crucial point: In production, what matters most is not what a model knows, but how it performs when stakeholders rely on it. Jonathan Frankle, chief AI scientist at Databricks, emphasizes that real-world trust and return on investment come from how AI models behave in production, not from how much information they contain.
Unlike traditional software, AI models generate probabilistic outputs rather than deterministic ones. The only thing you can measure about an AI system is how it behaves. You cant look inside it. Theres no equivalent to source code, Frankle tells Fast Company. He contends that while public benchmarks are useful for gauging general capability, enterprises often over-index on them.
What matters far more, he says, is rigorous evaluation on business-specific data to measure quality, refine outputs, and guide reinforcement learning strategies. Today, people often deploy agents by writing a prompt, trying a couple of inputs, checking their vibes, and deploying. We would never do that in softwareand we shouldnt do it in AI, either, he says.
Frankle explains that for AI agents, evaluations replace many traditional engineering artifacts, i.e., the discussion, the design document, the unit tests, and the integration tests. Theres no equivalent to a code review because theres no code behind an agent, and prompts arent code. That, he argues, is precisely why evaluations matter and should be the foundation of responsible AI deployment.
The shift from focusing on belief to emphasizing behavior is the foundation of two major innovations by Databricks this year: Test-Time Adaptive Optimization (TAO) and Agent Bricks. Together, these technologies seek to make behavioral evaluation the first step in enterprise AI, rather than an afterthought.
AI behavior matters more than raw knowledge
Traditional AI evaluation often relies on benchmark scores and labeled datasets derived from academic exercises. While those metrics have value, they rarely reflect the contextual, domain-specific decisions businesses face. In production, agents may need to generate structured query language (SQL) in a companys proprietary dialect, accurately interpret regulatory documents, or extract highly specific fields from messy, unstructured data.
Naveen Rao, vice president of AI at Databricks, says these are fundamentally behavioral challenges, requiring iterative feedback, domain-aware scoring, and continuous tuning, not simply more baseline knowledge.
Generic knowledge might be useful to consumers, but not necessarily to enterprises. Enterprises need differentiation; they must leverage their assets to compete effectively, he tells Fast Company. Interaction and feedback are critical to understanding what is important to a user group and when to present it. Whats more, there are certain ways information needs to be formatted depending on the context. All of this requires bespoke tuning, either in the form of context engineering or actually modifying the weights of the neural network.
In either case, he says, a robust reinforcement learning harness is essential, paired with a user interface to capture feedback effectively. That is the promise of TAO, the Databricks research teams model fine-tuning method: improving performance using inputs enterprises already generate, and scaling quality through compute power rather than costly data labeling and annotation.
While most companies treat evaluation as an afterthought at the end of the pipeline, Databricks makes it central to the process. TAO uses test-time compute to generate multiple responses, scores them with automated or custom judges, and feeds those scores into reinforcement learning updates to fine-tune the base model. The result is a tuned model that delivers the same inference cost as the originalwith heavy compute applied only once during tuning, not on every query.
The hard part is getting AI models to do well at your specific task, using the knowledge and data you have, within your cost and speed envelope. Thats the shift from general intelligence to data intelligence, Frankle says. TAO can help tune inexpensive, open-source models to be surprisingly powerful using a type of data weve found to be common in the enterprise.
According to a Databricks blog, TAO improved open-source Llama variants, with tuned models scoring significantly higher on enterprise benchmarks such as FinanceBench, DB Enterprise Arena, and BIRD-SQL. The company claims the method brought Llama models within range of proprietary systems like GPT-4o and o3-mini on tasks such as document Q&A and SQL generation, while keeping inference costs low. In a broader multitask run using 175,000 prompts, TAO boosted Llama 3.3 70B performance by about 2.4 points and Llama 3.1 70B by roughly 4.0 points, narrowing the gap with contemporary large models.
To complement its model fine-tuning technique, Databricks has introduced Agent Bricks, an agentic AI-powered feature within its Data Intelligence Platform. It enables enterprises to customize AI agents with their own data, adjust neural network weights, and build custom judges to enforce domain-specific rules. The product aims to automate much of agent development: Teams define an agents purpose and connect data sources, and Agent Bricks generates evaluation datasets, creates judges, and tests optimization methods.
Customers can choose to optimize for maximum quality or lower cost, enabling faster iteration with human oversight and fewer manual tweaks.
Databricks latest research techniques, including TAO and Agent Learning from Human Feedback (ALHF), power Agent Bricks. Some use cases call for proprietary models, and when thats the case, it connects them securely to your enterprise data and applies techniques like retrieval and structured output to maximize quality. But in many scenarios, a fine-tuned open model may outperform at a lower cost, Rao says.He adds that Agent Bricks is designed so domain expertsregardless of coding abilitycan actively shape and improve AI agents. Subject matter experts can review agent responses with simple thumbs-up or thumbs-down feedback, while technical users can analyze results in depth and provide detailed guidance. This ensures that AI agents reflect enterprise goals, domain knowledge, and evolving expectations, Rao says, noting that early customers saw rapid gains.
AstraZeneca processed more than 400,000 clinical trial documents and extracted structured data in less than an hour with Agent Bricks. Likewise, the feature enabled Flo Health to double its medical-accuracy metric compared with commercial large language models while maintaining strict privacy and safety. Their approach blends Flos specialized health expertise and data with Agent Bricks, which leverages synthetic data and tailored evaluation to deliver reliable, cost-effective AI health support at scaleuniquely positioning us to advance womens health, Rao explains.
From benchmarks to business data
The shift toward behavior-first evaluation is pragmatic but not a cure-all. Skeptics warn that automated evaluations and tuning can just as easily reinforce bias, lock in flawed outputs, or allow performance to drift unnoticed.
In some domains we truly have automatic verification that we can trust, like theorem proving in formal systems. In other domains, human judgment is still crucial, says Phillip Isola, associate professor and principal investigator at MITs Computer Science & Artificial Intelligence Laboratory. If we use an AI as the critic for self-improvement, and if the AI is wrong, the system could go off the rails.
Isola points out that while self-improving AI systems are generating excitement, they also carry heightened safety and security risks. They are less constrained, lacking direct supervision, and can develop strategies that might be unexpected and have negative side effects, he says, also warning that companies may game benchmarks by overfitting to them. The key is to keep updating evaluations every year so were always testing models on new problems they havent already memorized.
Databricks acknowledges the risks. Frankle stresses the difference between bypassing human labeling and bypassing human oversight, noting that TAO is simply a fine-tuning technique fed by data enterprises already have. In sensitive applications, he says, safeguards remain essential and no agent should be deployed without rigorous performance evaluation.
Other experts note that greater efficiency doesnt automatically improve AI model alignment, and theres no clear way to measure AI model alignment currently. For a well-defined task where an agent takes action, you could add human feedback, but for a more creative or open-ended task, is it clear how to improve alignment? Mechanistic interpretability isnt strong enough yet, says Matt Zeiler, CEO of Clarifai.
Zeiler argues that the industrys reliance on a mix of general and specific benchmarks needs to evolve. While these tests condense many complex factors into a few simple numbers, models with similar scores dont always feel equally good in use.
That feeling isnt captured in todays benchmarks, but either well figure out how to measure it, or well just accept it as a subjective aspect of human preference; some people will simply like some models more than others, he says.
If the results from Databricks hold, enterprises may rethink their AI strategy, prioritizing feedback loops, evaluation pipelines, and governance over sheer model size or massive labeled datasets, and treating AI as a system that evolves with use rather than a onetime product.
We believe the future of AI lies not in bigger models, but in adaptive, agentic systems that learn and reason over enterprise data, Rao says. This is where infrastructure and intelligence blur: You need orchestration, data connectivity, evaluation, and optimization working together.
In 2022, Diarrha N’Diaye-Mbaye had achieved a lifelong dream: Ami Colé, her three-year-old beauty brand, was on the shelves of Sephora. In the wake of George Floyd’s murder in 2020, she’d received a wave of support from venture capitalists and retailers. But by this year, much of that interest had dried up. In mid-July, N’Diaye-Mbaye abruptly announced she would be shuttering her fledgling brand because she could not find enough capital to stay afloat.
The news sent shock waves through the beauty industry, but it’s an increasingly familiar story for venture-backed Black-owned brandsparticularly those that scaled with the help of major retailers who went all-in on DEI after 2020’s racial reckoning.
Tina Wells recently shut down Wndr Ln, the luggage brand she launched in partnership with Target, after the retailer cancelled all future orders. Thirteen Lune, a diversity-focused online retailer, went through insolvency proceedings last December. Many other black-owned beauty brands have closed in the wake of Trump’s election, including Beauty Bakerie, Ceylon, and Koils by Nature.
Black founders are now trying to figure out what went wrong. For many, the answer is that investors and retailers like Target quickly launched diversity, equity, and inclusion (DEI) programs without long-term strategies to help Black-owned brands scale and find success. Ultimately, DEI was often perceived as a moral endeavor, rather than smart business. So it’s not that surprising that so many of them are now struggling.
“DEI was synonymous with altruism, rather than strategy,” says Marcus Collins, a professor of marketing at the University of Michigan and the author of For The Culture. “They saw serving Black people as a good thing to do rather than seeing them as consumers with unbelievable buying power.”
The Rise and Fall of Ami Colé
At Sephora’s annual beauty festival last September, crowds gathered in a pavilion featuring the hottest up-and-coming brands to grace the retailer’s shelves. Ami Colé’s booth was designed to look like a Harlem hair salon, complete with a bright orange swivel chair and African-inspired baskets.
It was a proud moment for founder Diarrha N’Diaye-Mbaye, who named her company after her mother, a Senegalese immigrant who opened a hair salon in New York. As she described in The Cut, she began building Ami Colé in 2019, but it wasn’t until 2021 that retailers and investors began returning her calls. As the Black Lives Matter uprisings spread across the country, the business community tried to address systemic racism by launching diversity, equity, and inclusion (DEI) programs.
Target vowed to invest $2 billion in at least 500 Black-owned businesses; Walmart poured $100 million into a racial equity center; Sephora took a pledge to devote 15% of its shelf space to Black-owned brands. For months, Black founders received an influx of cash and interest from retailers: N’Diaye-Mbaye herself raised $1 million to launch her brand. But five years later, as the Trump administration wages war against DEI, the mood in the country has shifted, and support for Black entrepreneurs is drying up.
Now N’Diaye-Mbaye does not have enough capital to keep her brand going, and she’s far from alone. According to Crunchbase, Black-owned beauty brands raised $16 million in 2024, a sharp decline from $73 million in 2022. This withdrawal of support for Black-owned brands is happening across product categories, and a wave of startups has quietly closed in recent months. Some of these brands’ founders are now in a worse position than they were before they received the DEI support; they’re dealing with debt, unsold inventory, and other liabilities.
DEI Programs Had No Long Term Vision
Karen Young, founder of the beauty brand Oui the People, predicted this wave of closures. In July 2024, she posted a TikTok video about Black founders she knew who were struggling to get the funding they needed to keep their businesses afloat, as support evaporated.
While many brand founders dream of getting picked up by a national retailer, Young knows firsthand how expensive this can be. To launch Oui the People at Sephora, she had to buy enormous quantities of inventory, pay for displays and product samples, and pour a lot of money into marketing to get on consumers’ radars.
This is consistent with other reporting I’ve done about how brands can spend upwards of $100,000 on in-store fixtures at Sephora, and must also provide product testers and samples. To get a spot on a seasonal display at the store, entrance can cost $250,000. And after all of that, Sephora takes a 65% cut in sales.
To pay for all of this, Young raised $8 million in venture capital, led by New Age Capital. “The first thing Sephora’s merchants asked me was whether I had funding,” says Young. “You need capital to get off the ground. It’s only when you scale that you have a path to profitability.”
Young did all of this work in 2019, before the DEI programs began popping up. This turned out to be a blessing, she says, because Sephora and her investors worked with her to come up with a plan for Oui the People to find its place in the market and achieve scale.
In contrast, after Floyd’s murder, many companies pumped money into Black-owned brands without any sort of long-term strategy. “DEI can’t come without infrastructure,” she says. “Retailers brought in these very small Black-owned businesses across their stores, then just stopped there.”
This is what happened to Ami Colé. (Sephora declined to comment; Ami Colé did not respond to our request for comment.) As N’Diaye-Mbaye writes in The Cut, she used her $1 million to launch at Sephora, but struggled to compete with brands that had access to far more capital. She eventually raised $2 million more from venture capital firms like G9 Ventures and Greycroft, but without further ongoing investment she sees no path to success.
“Diarrha performed miracles on the capital she raised, more than comparable brands in her category, like Kosas and Saie,” says Young. “She created amassive shade range and cultivated a loyal customer base. But she hasn’t had the same access to capital as comparable brands.”
Collins, the professor, says the sudden withdrawal of support for Black-owned brands is devastating for foundersand not just financially. “These entrepreneurs had hope,” he says. “They named their companies after their parents because they wanted to build a family legacy. And overnight, the rug was pulled out from under them.”
Doomed to Fail
Wells of Wndr Ln believes retailers treat Black-owned brands more poorly when they are brought in through DEI programs. She’s seen this firsthand across her two decade career in which she has owned a consulting business, launched her own brands, and also written books for both children and adults. “I’ve worked with 400 clients over the last two decades of my career,” she says. “Only two have come through DEI initiatives, and both were awful.”
One of those experiences happened with Target, a retailer she has worked with closely for six years. In 2019, Target asked Wells to write a series of children’s books featuring a Black female lead charactersomething they felt was missing on their shelves. (Wells had previously published successful middle grade books.) This led to a bestselling series called the Zee Files, which was exclusively sold at Target.
Later, Wells wrote a business book called The Elevation Approach, and Target invited her to create a line of coordinating home office products. In each case, Target poured substantial marketing dollars into the launches, which led to their success. “Target did it because it was good for business,” says Wells. “They felt the Black customer was worth cultivating and investing in.”
Then, in the aftermath of Floyd’s murder, Target launched an internal committee called REACH, focused on improving racial equity throughout the company, including bringing on 500 new Black-owned brands. In 2021, REACH reached out to Wells, asking whether she would be interested in launching a luggage brand at Target. Given all of her positive experiences at Target thus far, Wells said yes and began developing Wndr Ln (pronounced Wonder Lane), a line of colorful suitcases and overnighters. She used her own money to manufacture the products.
When Wndr Ln debuted in August 2023, Wells noticed a lack of support from Target compared to her previous projects. Target did not invest much in marketing, nor were products prominently displayed in store. (A Target spokesperson confirmed it carried the Wonr Ln collection, but says the company’s policy is not to comment on vendor relationships.)
Shortly after the launch, Target cancelled all future orders. Wells never received a clear explanation about why this happened, but at the time, Target’s sales were in decline partly because of a massive consumer boycott over its Pride collection. Whatever the reason, Wells was left in a bind. “If you’re producing product for a single retailer and they cancel future orders, your business is dead,” says Wells. “I was left with massive liability I am still dealing with today.” (She cannot comment on the financial details of the end of this Target partnership for legal reasons.)
For Wells, the problem with DEI programs is that they don’t often focus on driving profit and revenue. DEI is often seen as a moral issue, rather than an opportunity to bring in Black entrepreneurs who can target the valuable Black consumer.
As a result, with a recession looming and the Trump administration attacking DEI, it is easy for brands to abandon Black-owned brands. “America is a capitalistic society,” says Wells. “The goal of Fortune500 companies is to increase shareholder value. The minute anything is not in alignment with that goal is not going to have long-term success.”
Marcus Collins emphasizes that launching DEI programs were designed to address real problems. Historically, American companies have ignored the needs of Black and brown customers. Yet, there is abundant research showing that catering to diverse consumers is good for business. “The thing that gets in the ways is racism, so let’s just call it what it is,” says Collins. “Companies don’t cater to Black people because they don’t think Black people matter.”
Going forward, Karen Young says that companies should focus on partnering with Black founders because they have better insight into the diverse consumers they’re seeking. And importantly, the business community needs to give these entrepreneurs access to the resources they require to succeed, including access to capital from banks and VCs. “These are baseline resources that other founders get,” says Young. “We’re just asking for the same treatment.”
But in some ways, setting up DEI programs may not even be necessary in the years to come, Wells argues. It will soon be abundantly clear that brands will lose out financially if they don’t cater to Black and brown consumers, as white people become the minority in the U.S. by the 2040s.
“America is becoming more diverse every day,” says Wells. “If you don’t want to serve your Black and brown customers, don’t worry. Someone else will come along to do it. That’s how capitalism works.”
In the filtered water space, there is one company that has dominated brand awareness for decades. Water pitchers and filtration devices from Brita can be found in so many millions of homes and offices around the world that the term market saturation is more than just a pun.
But there’s another water filtration company that, despite lower kitchen visibility, is actually a bigger player in the clean-water game. Culligan, founded in 1936 as a water softening and filtration service company, became known for its white-glove service.
[Photo: Culligan]
Often installed in basements or storage closets, Culligan’s equipment was as utilitarian as a water heater or furnace. Once the system was installed in a home or office, its users hardly gave it another thought, or look. “It was the technician that was actually working with the product,” says Kathy Chi Thurber, Culligan’s new global president of consumer products. “The products didn’t have to be beautiful, but the technicians had to be able to talk about our history, our capabilities, our research, and innovation.”
Now, as a private 15,000-person company that pulled more than $3 billion in revenue in 2023, Culligan is embarking on a total brand and strategy overhaul. And aggressively so. Within the past five years, Culligan has acquired 362 companies in the clean-water industry, from local water purifiers to filter companies to component manufacturers. It’s positioning itself as a dominant player in a world where water safety and water scarcity are of increasing concern.
[Photo: Culligan]
Out of the basement and into your kitchen
One priority is to start competing more directly in the consumer space, bringing its equipment out of the basement and into the hands of water drinkers everywhere. “We’d never really given an eye to the consumer, and that has 100% changed,” says Chris Quatrochi, chief product and technology officer at Culligan International.
To venture into the Brita-dominated consumer market, Culligan turned to the industrial design firm Ammunition Group. Known best for its work designing Beats by Dre headphones and products for companies like Polaroid, Square, and Lyft, Ammunition was tasked with helping Culligan develop products that appeal to regular consumers. It also updated the brand to tell those consumers that Culligan is not the box-in-the-basement brand they may have known in the past.
“Our portfolio has not been the greatest from a, I would say, beauty perspective,” Quatrochi says. “If you really want to show that you are leading edge from a water-quality perspective, you have to have a product that demonstrates that.”
Ammunition started by applying its deep product design background to creating a water filtration pitcher that embodies this new company focus. Building on its 2020 acquisition of the water filter maker ZeroWater, Culligans ZeroWater Technology line of three handheld pitchers and two countertop dispensers is the companys first foray into the consumer space.
[Photo: Culligan]
Designing a better water pitcher
Ammunition’s design focused primarily on the ways people actually use filtered water pitchers. “One of the constraints is putting it in your refrigerator,” says industrial designer Robert Brunner, Ammunition’s founder. Research into the market showed that more than 70% of water pitcher users, particularly those in the U.S. and Western Europe, keep their pitchers in the refrigerator, often in the door of the appliance.
At the same time, most of the pitchers on the market don’t actually fit into a fridge door all that well. Their rectangular shape and bulging handle tend to take up a lot of space, and need more room around them to be moved in and out.
[Photo: Culligan]
Ammunition rethoght that form factor to better fit inside the refrigerator door, using a rounded square shape for the pitcher that allows it to fit more like a carton of milk. The pitcher also has an innovative open-ended handle that cuts down on its overall bulk and allows more stuff to fit in the refrigerator door’s shelves alongside it, while also being more ergonomically comfortable to carry and hold.
“Figuring out how to have that single connection point for that handle so it’d be mechanically robust and reliableit was actually a fair amount of engineering effort to make sure that could work when it’s getting filled up with water,” Brunner says. “The handle is extremely important, because when this thing is full, it’s quite heavy, and you have to be able to manipulate it, carry it, pour it. We wanted to maintain this simplicity.”
The design team also thought about the spout shape and the challenge of pouring water for people with dexterity and mobility issues. That led to considerations about one of the key parts of using a water pitcher: refilling it. Ammunition designed a sliding lid that makes holding the pitcher under the tap and refilling it easier.
The lids circular shape became a recurring theme in the design of the pitcher line, as well as the broader work Ammunition is doing across Culligan’s other product and service categories. “The circular element is really the most natural shape to route water from one place to another, pipes being the most obvious example,” says Christopher Kuh, vice president of Ammunition’s industrial design studio. “It’s really an important and core element.”
Another differentiating factor is the built-in water-quality meter. Measuring total dissolved solids (TDS) at the scale of parts per million, the digital meter slots into the pitchers and the countertop dispensers to give users a clear readout of how well the filter is functioningand when it’s time to replace it.
“The TDS meter actually is going to start to read a value above zero at some point in time, which gives you a clear indication of the end of filter life,” Kuh says. In a clever turn, the meter can be removed from the pitcher or dispenser to dip into, say, a glass of water direct from the tap to see just how much the filtration system is doing.
[Photo: Culligan]
A bigger rebrand moment
These design moves were informed by deep user research Culligan has conducted over the past three years. Thurber says Ammunition was game for putting its design prototypes in the hands of users from the very early stages and taking their feedback to inform new iterations of the designs before landing on a final product that looks and feels different from what’s already out there.
“We all know who the major competitor is that has, like, 60% to 70% market share,” Thurber says. “It would be very hard to break through if we were not serious about what we wanted to do, and if we were not game-changing in our design and our functionality.”
But this doesn’t mean Culligan is abandoning the more utilitarian water products that have kept it in business for nearly a century. Instead, Ammunition’s design approach for the pitcher is being extended throughout Culligan’s product offerings, including the industrial-scale water softeners and filtration systems that still live in basements and utility closets, as well as the company’s large and growing business in office water coolers. Some of those redesigned products will be coming online in the next year.
