Capital One Financial has reportedly entered a “new era” after completing its acquisition of Discover Financial Services. With the acquisition, Capital One grew in size and added a debit and credit card network, which could “supercharge” its banking and card businesses. The Discover network could enable Capital One to make more money from debit card payments than competitors that are not both a card issuer and network; use the incremental interchange revenue to boost its bottom line or fund debit card rewards to attract new customers; and fund more investment and enhance rewards and deals to keep expanding its credit business. When Capital One announced in February 2024 that it planned to acquire Discover for $35.3 billion, it said the transaction would create a global payments platform with 70 million merchant acceptance points in more than 200 countries and territories. “Our acquisition of Discover is a singular opportunity to bring together two very successful companies with complementary capabilities and franchises, and to build a payments network that can compete with the largest payments networks and payments companies,” Richard Fairbank, founder, chairman and CEO of Capital One, said. “Through this combination, we’re creating a company that is exceptionally well-positioned to create significant value for consumers, small businesses, merchants and shareholders as technology continues to transform the payments and banking marketplace.”
Kumo’s ‘relational foundation model’ingests raw database tables and lets the network discover the most predictive signals on its own without the need for manual effort to deliver “zero shot” capabilities on structured data
Stanford professor and Kumo AI co-founder Jure Leskovec argues that his company’s tool, a relational foundation model (RFM), is a new kind of pre-trained AI that brings the “zero-shot” capabilities of large language models (LLMs) to structured databases. Kumo’s approach, “relational deep learning,” sidesteps the manual process with two key insights. First, it automatically represents any relational database as a single, interconnected graph. For example, if the database has a “users” table to record customer information and an “orders” table to record customer purchases, every row in the users table becomes a user node, every row in an orders table becomes an order node, and so on. These nodes are then automatically connected using the database’s existing relationships, such as foreign keys, creating a rich map of the entire dataset with no manual effort. Second, Kumo generalized the transformer architecture, the engine behind LLMs, to learn directly from this graph representation. Transformers excel at understanding sequences of tokens by using an “attention mechanism” to weigh the importance of different tokens in relation to each other. Kumo’s RFM applies this same attention mechanism to the graph, allowing it to learn complex patterns and relationships across multiple tables simultaneously. Leskovec compares this leap to the evolution of computer vision. RFM ingests raw database tables and lets the network discover the most predictive signals on its own without the need for manual effort. The result is a pre-trained foundation model that can perform predictive tasks on a new database instantly, what’s known as “zero-shot.” The RFM can serve as a predictive engine for these agents. Kumo’s work points to a future where enterprise AI is split into two complementary domains: LLMs for handling retrospective knowledge in unstructured text, and RFMs for predictive forecasting on structured data. By eliminating the feature engineering bottleneck, the RFM promises to put powerful ML tools into the hands of more enterprises, drastically reducing the time and cost to get from data to decision.
The five seismic shifts that will reshape communications and marketing- Attention Economy, Creator Economy, Stakeholder Economy, Experience Economy, Intelligence Economy
The next five years will transform communications more dramatically than the previous five decades. Here are 5X: five seismic shifts that are reshaping the future of our profession. The Attention Economy: Earned renewal. The strategy: Companies can’t buy attention the way they used to — they must earn it by creating ideas that move through culture, not just media. The impact: Winning ideas spread because they let people express their identity and values. Breakthrough concepts don’t just reach communities — they form them. The Creator Economy: Influence redefined: The new reality: Brands must co-create with those who shape culture, measuring actual influence over vanity metrics. Why it’s important: Creators increasingly drive earned strategies, capturing attention that attracts traditional media coverage and fuels broader cultural conversation. The Stakeholder Economy: Business meets culture. The Experience Economy: Digital, integrated: What this means: Audiences, who increasingly live their waking hours connected, expect personalized, participatory experiences across physical and digital touchpoints. Moments must be felt, not just seen. The breakdown: The companies that succeed in this AI generation will look nothing like those that succeeded in the last one. If you haven’t changed your playbook, you’re already behind.
AI is setting the idea of four-day work into action by allowing firms to share the gains of improved technology and giving workers some of their time back with no change in pay
The idea of four-day work week is gaining traction among proponents of the four-day work week, and at least one software startup CEO has moved his company to a 32-hour week — with no change in pay — because of AI. Sen. Bernie Sanders (I-Vt.) is a four-day work week booster, having introduced a 32-hour workweek bill last year, though such a proposal is unlikely to get far in Congress. Instead of firing people, proponents argue that firms share the gains of improved technology by giving workers some of their time back. Instead of fearing AI will replace them, workers welcome its advancements and figure out creative ways to leverage the tech. Economist Juliet Schor, lead researcher for 4DWG, a global nonprofit that’s piloted shorter work weeks with 245 organizations in the U.S., U.K. and elsewhere says “The ability of large language models like ChatGPT to wipe out millions of good-paying positions means we need to be intentional about how we adjust to that technology. Reducing hours per job is a powerful way to keep more people employed.” Smaller firms can more easily implement a big change like a four-day week — larger companies are likely to have a harder time making it happen, experts say. But reducing work hours to make sure a lot of people don’t lose their jobs when technology advances isn’t a new idea. Shortening work hours as a way to reduce unemployment was one of the arguments wielded by advocates for five-day work weeks back in the early 20th century. (That used to be a wild idea, too.) Roger Kirkness, the CEO of a small software startup called Convictional moved the company to a four-day workweek — without reducing anyone’s pay. Kirkness tells that using AI accelerates writing code but it doesn’t speed up everything — teams still need to be able to think creatively to solve problems and get real work done. The four-day work week is meant to keep everyone fresh, with enough time to recharge so they can do deep-thinking.
Report reveals new gen AI models typically spend three weeks at the top of the usage charts and then drop off as new open-source rivals emerge with frontier models depreciating on a 6–12 month timescale
Generative AI has entered the mainstream faster than any previous new technology. But the tech industry hasn’t yet figured out the best ways to build AI products — and fierce competition, along with rapid advances, means nobody stays at the top of the heap for long. Those are some of the top-line findings of a new report on the state of AI foundation models from Innovation Endeavors, the venture capital firm co-founded in 2010 by former Google CEO and chairman Eric Schmidt. The report says that 1 in 8 workers globally now uses AI on a monthly basis. 90% of that growth took place in the last six months. New models regularly topple technical benchmarks, but the cost of training them is also ballooning. “The average duration of human task a model can reliably do is doubling every seven months,” per the report. New models typically spend three weeks at the top of the usage charts and then drop off as newcomers emerge and open-source rivals absorb and commoditize their advances. Frontier models “depreciate on a 6–12 month timescale,” the report says. AI is “fundamentally disrupting” software development and collapsing the distinctions between programming, product management and design, the report also finds. Therapy, life organization and learning” lead the list of general-interest AI use cases.
New campaign spotlights neurodivergent talent as a token effort, but to reposition it as essential to the future of the creative industry
New global campaign by Havas, Beyond the Brief, launched live at the 2025 Cannes Lions International Festival of Creativity, was designed not to spotlight neurodivergent talent as a token effort, but to reposition it as essential to the future of the creative industry. Helmed by Donna Murphy, Global CEO of Havas Creative and Health Networks, the campaign builds on the foundation of Neuroverse: Powered by Havas, aiming to shift how the industry identifies and cultivates creative potential radically. The campaign’s anchor panel was named “Neurodivergent Minds: They Don’t Need Advertising. Advertising Needs Them.” Days before the launch, unbranded digital teaser displays appeared along the Croisette, posing provocative questions like, ‘What if the future of creativity doesn’t look like the past, and never did?’ Each display featured a QR code that led to a dedicated microsite with a full agenda of neurodiversity-focused programming, including featured panels, downloadable insights, and events hosted beyond Havas. The campaign also highlighted that neurodivergent people are more than just creative contributors; they’re also a critical consumer demographic. A session at the Havas Café titled “The New Creative Alchemy: Neurodivergent Minds & AI as Industry Catalysts” offered a roadmap for how brands and agencies can better support neurodivergent talent and create products and campaigns that resonate with their experiences. “With Beyond the Brief, we’re looking to amplify these voices and challenge the industry to rethink the systems in place,” Murphy emphasized. By debuting the initiative at the largest creative festival in the world, she made it clear that this wasn’t about corporate social responsibility; it was about redesigning the very ecosystem of creativity. Too often, neurodivergent candidates are excluded not because they lack.
Embedded Finance 2.0: SaaS platforms are chasing banking charters, adopting multi-tenant model for splitting deposits and baking compliance features such as automated KYC and ledger-level reporting into the API call
Embedded Finance 2.0 is where the “Pay” button graduates into a full on-platform treasury desk and the question morphs from Can we process a payment? to Should we hold your cash? First, the deep-pocketed few will likely chase full-fat charters. Intuit already controls an OCC-granted industrial loan company, and rumor has it Shopify is exploring a Canadian Schedule I license so it can plug directly into FedNow and CAD settlement rails. Owning the license erases sponsor fees, provides direct central-bank access and turns Fed holidays into just another dashboard metric. It also drags CEOs into capital-ratio land, where quarterly stress tests replace flashy conference-stage keynotes. Second, mid-tier platforms are furiously diversifying sponsors. Stripe quietly maintains half a dozen partner banks across continents; Adyen splits deposits between its European and U.S. charters so funds never cross jurisdictions. A multi-tenant model may mollify supervisors who now demand credible “if-this-bank-fails” exit plans. It also means engineering teams must reconcile half a dozen core-banking APIs before morning coffee—a non-trivial tax on velocity. Third, a new generation of BaaS providers such as Unit, Treasury Prime, and Griffin, is selling compliance as the actual product: automated KYC, ledger-level reporting, FDIC-ready dashboards baked right into the API call. The value prop is clear—“Let us worry about Section 314(b) so you can focus on restaurant software”—but only holds if the RegTechs stay ahead of evolving rules. If they lag, their SaaS clients inherit the audit headache anyway. If Embedded Finance 1.0 merely added a Pay button, 2.0 aims to become the balance sheet. The platforms that survive will treat compliance as a feature, not a cost center. That means real-time ledgers auditors can query in a single GET request, FedNow and SEPA Instant wired in at the kernel, and credit models transparent enough that supervisors nod before shareholders cheer. SaaS founders used to brag about daily active users; tomorrow they might brag about their Liquidity Coverage Ratio. The irony would be delicious if it weren’t so expensive. Once embedded finance volumes crest half-a-trillion dollars, systemic stability demands oversight inside the platforms where money truly lives. The badge on your business debit card might one day read Shopify, Toast or Xero—but the compliance brain under the hood will need to think, grizzled and cautious, like JPMorgan.
Citi’s report warns the more interconnected, always-on environment from real-time payment systems and sophisticated AI-driven fraud poses significant threats
As instant payments become the norm, financial institutions could be at risk of falling behind with increasingly sophisticated AI-driven fraud posing significant threats, Citi, warns in a new report. In the report, it details how Asia and emerging markets are leading the way when it comes to the 24/7 availability of seamless instant transactions. To many, this will come as no surprise, with real-time systems like Pix in Brazil, UPI in India, or M-Pesa in Kenya, driving economic development. Despite the enhancements to the speed of payments across the globe, and the promise of boosting global GDP by over $280billion in 2028, a more interconnected, always-on environment can amplify the risks of fraud. Citi warns that in 2024, consumers globally lost an estimated $1trillion to scams across all payment methods. Citi says that by embracing innovation, prioritising cybersecurity, and fostering collaboration across the financial ecosystem, businesses and financial institutions can start to unlock the full potential of real-time payments and build a more inclusive, efficient, and dynamic financial future. By enabling faster settlements, improved working capital management, and enhanced customer experiences, real-time cross-border payments are unlocking new opportunities for businesses operating globally.
Early Warning claims Zelle’s scale of >150 million enrolled user accounts, the ability to do away with collecting or storing sensitive account and routing information and fraud-free transaction rate of 99.95% make it a viable solution for all federal disbursements and receipts
Early Warning Services submitted a comment letter today responding to the U.S. Department of the Treasury’s Request for Information regarding the transition to electronic payments for all federal disbursements and receipts. Early Warning’s comment letter detailed the critical benefits it can provide Treasury through Zelle, including: Scale: Zelle reaches more than 150 million enrolled user accounts, which is equal to more than the combined population of California, Texas and Florida. Security: Zelle removes a risk vector that paper checks present. There is no need to collect or store sensitive account and routing information because Zelle is already embedded within U.S. bank accounts. Safe: Today, more than 99.95 percent of all Zelle transactions are completed without any report of scam or fraud. Treasury checks are 16 times more likely to be reported lost or stolen, returned undeliverable, or altered than digital payments. Speed: Reliance on paper checks slows access to funds and burdens recipients, especially those with limited mobility or transportation. With disbursements using Zelle, funds can be delivered in minutes to an enrolled recipient’s verified bank account.
Rocket’s new Preferred Pricing, is offering clients who finance through Rocket Mortgage and purchase with a Redfin agent or buy a Redfin-listed home a one percentage point interest rate reduction for the first year of their loan or a lender credit at closing of up to $6,000
Rocket Companies, the Detroit-based homeownership platform, has announced the completion of its acquisition of Redfin, uniting the most-visited real estate brokerage website with America’s largest mortgage lender. “I’ve used Redfin every day for the last 20 years. It helped me find and fall in love with my first home, completely changing how I thought about real estate,” said Varun Krishna, Rocket Companies CEO. “The Redfin team is best-in-class in building a product experience focused on simplicity. It was a perfect fit for Rocket’s vision of what the homeownership experience should be.” Alongside the acquisition, the companies introduced Rocket Preferred Pricing, offering clients who finance through Rocket Mortgage and purchase with a Redfin agent or buy a Redfin-listed home a one percentage point interest rate reduction for the first year of their loan or a lender credit at closing of up to $6,000. This offer applies to qualified buyers using conventional, FHA, or VA loans. Rocket Mortgage and Redfin plan to roll out additional offerings for buyers, agents, and brokers in the coming months. Redfin has also unveiled a refreshed brand identity as “Redfin Powered by Rocket,” aligning with the unified experience. “The gulf between the American Dream of homeownership and reality has never been wider,” said Redfin CEO Glenn Kelman. “The reason Rocket and Redfin came together was to bridge that gap, so that the people who spend their days dreaming on Redfin.com can easily use Rocket financing to own their dream.”