The consumerization of cross-border payments has been driven by necessity. Migrant workers and families sending $200 remittances could not tolerate high costs or opaque delivery times. Mobile money platforms proved that payments did not have to be tethered to banks, while global players showed that exchange-rate transparency could be a selling point rather than a risk factor. As businesses globalize, the idea of payments halting for weekends or holidays looks increasingly anachronistic. Treasury platforms now integrate with wallet infrastructures that maintain balances in multiple currencies and release funds instantly, bypassing traditional cut-off windows. The result is a new operational flexibility. Treasurers can optimize for cost, speed or transparency depending on the transaction type. Supplier payments may benefit from low-cost local rails, while high-value transfers may still flow through SWIFT to ensure compliance with regulatory reporting. This portfolio approach, inspired by consumer FinTech, marks a departure from the one-size-fits-all model that defined cross-border payments for decades. Treasurers now find themselves at the frontier of FinTech adoption, piloting tools that only a few years ago were considered niche or even risky. Corporate treasurers now know that instant, transparent and low-cost international transfers are possible because they see them in the consumer space every day. The challenge is translating those possibilities into enterprise-grade solutions without losing the compliance rigor that corporations cannot compromise.
GenAI shifts from training to inference infrastructure; 75% compute demand by 2030 requires millisecond-critical latency optimization for production deployment
Inference infrastructure is emerging as a distinct layer in the generative AI stack, bridging compute and applications. Here are six things to be aware of: 1) The Shift From Training to Inference: AI’s spotlight has long been on training, with companies amassing data and building larger models. The real challenge now is inference: running those models in production, serving billions of queries and delivering instant results. 2) What Inference Really Means: Training is when a model learns from massive datasets on high-powered hardware. Inference is when a trained model is applied to new inputs in real time. It powers everything from ChatGPT prompts to fraud checks and search queries. This constant, real-time activity never stops. Keeping ChatGPT online alone reportedly costs OpenAI tens of millions of dollars per month. 3) The Scale of Demand: Generative AI has moved from research to mainstream use, creating billions of inference events daily. As of July 2025, OpenAI reported handling 2.5 billion prompts per day, including 330 million from U.S. users. Brookfield forecasts suggest that 75 percent of all AI compute demand will come from inference by 2030. 4) Why Infrastructure Matters: Unlike training, inference is the production phase. Latency, cost, scale, energy use and deployment location all determine whether an AI service works or fails. Optimized infrastructure spans computing, networking, software and deployment strategies to keep predictions reliable at scale. 5) Latency Is Business-Critical: Milliseconds make or break user experience. A delay can frustrate chatbot users, or worse, prevent a fraud detection system from stopping a fraudulent payment in time. Every millisecond counts when millions of customers are involved. 6) Cutting Costs With Optimization: Inference is a recurring operating expense, not a one-time investment. Providers rely on optimization techniques to lower costs without sacrificing accuracy: Batching: processing multiple requests at once; Caching: reusing frequent results; Speculative decoding: letting a smaller model draft quick answers before a larger one verifies them; Quantization: reducing numerical precision to cut compute and energy use.
Salt Security introduces MCP Protect and Agentic AI Governance controls integrated with CrowdStrike SIEM to secure proliferating agent-driven API interactions
With the rise of agentic AI, API exposure has proliferated. Agents fan out call paths and amplify traffic, effectively turning APIs into the enterprise “plumbing” of operations, according to Michael Callahan, chief marketing officer of Salt Security. This has created the “API fabric” — a complex, constantly moving mesh of connections that enterprises struggle to see, let alone secure. A large part of the API security conversation is on the role of MCP, an open standard championed by Anthropic PBC, and A2A, Google’s protocol for agent-to-agent interactions, according to Nicosia. Both sit atop existing APIs, acting as brokers to manage data retrieval and collaboration between agents. “For us, the visibility of the AIs and the MCPs … the protocols are so paramount because you can’t protect what you don’t know,” Nicosia said. “Having that visibility from either a zombie API or a zombie MCP protocol server, we give you that visibility. At least you’re aware of all of this proliferation that’s going on with the organization. And then how do you govern it? And then how do you protect against it?” Salt’s momentum has been bolstered by its close partnership with CrowdStrike Holdings Inc. The company is a Falcon Fund portfolio company and has integrated its API security solutions with CrowdStrike’s Falcon platform and next-generation security information and event management. Together, they provide customers with unified visibility across APIs and AI-driven workflows, Nicosia added.
McKinsey documents “great convergence” of traditional and alternative asset management driving $6-10.5 trillion through innovative semi-liquid products, evergreen funds and public-private model portfolios
A study by the consultancy McKinsey & Co noted that asset management achieved its “long-anticipated rebound in 2024 and 2025, but it arrived with more grit than grace.” Assets under management around the world reached a record $147 trillion by the end of June 2025. The report said one structural trend dominates: the “great convergence” between traditional and alternative asset management. “These two worlds are beginning to blend as public and private investing increasingly overlap, and as private capital managers penetrate deeper into wealth, defined contribution, and insurance channels. This convergence is showing up in dealmaking and partnerships across the public/private divide and through innovations such as semi-liquid products, evergreen funds, and public–private model portfolios,” the report said. Other forces add to money moving around: a reassertion of home country bias as investors rotate from global to local exposures, and the growth of active ETFs. The McKinsey report noted that after a period of rapid growth, private market investing – often mentioned in a wealth management context – is causing bloat and indigestion.
After peaking at nearly $1.7 trillion in 2021, global private markets’ fundraising slid to roughly $1.1 trillion in 2024 – a return to 2017 levels. The slowdown was broad, but most pronounced in private equity and real estate where exits stayed muted, the report said. “Private credit and infrastructure decelerated far less than private equity and real estate. Credit continues to benefit from the refinancing of sponsor portfolios as well as new areas of demand such as asset-backed finance and infrastructure lending. Infrastructure offers both inflation-protected, long-dated yields and exposure to a broadening range of ‘new economy’ assets, such as data centres,” it said. “Private wealth channels and secondaries have proved to be a bright spot in the industry. In private wealth, evergreen vehicles and semi-liquid fund structures have gained substantial traction among high-net-worth and affluent investors,” it said. In the US, these vehicles grew to $348 billion in AuM and attracted $64 billion in inflows in 2024. Secondaries are now a critical release valve, with global AuM above $700 billion and roughly $130 billion raised in 2024, it said. (With secondaries, an investor buys or sells pre-existing stakes in private markets from other investors.) “Together, flows from private wealth and secondaries are now injecting meaningful new capital into the ecosystem, backfilling an estimated 15 to 20 per cent of the annual fundraising shortfall compared to 2021,” it said.
Morgan Stanley’s E*Trade will launch cryptocurrency trading in 2026 through a partnership with digital asset infrastructure provider Zerohash
Morgan Stanley’s E*Trade will launch cryptocurrency trading in 2026 through a partnership with digital asset infrastructure provider Zerohash, underscoring Wall Street’s deepening push into digital assets amid a wave of supportive legislation from the Trump administration. E*Trade clients will be able to buy Bitcoin, Ether and Solana in the first half of 2026. the brokerage planned to add crypto trading next year. At the time, the initiative was still in early stages as E*Trade sought partnerships with infrastructure providers. E*Trade was acquired by Morgan Stanley in 2020 for $13 billion. At the time of the deal, the discount brokerage had more than 5.2 million users and offered a retail-focused platform for trading regulated financial securities, focused mainly on US residents. Zerohash will build a full wallet solution for E*Trade clients. Perhaps E*Trade’s biggest rival in the discount brokerage crypto space is Robinhood, which has rapidly expanded its footprint by offering crypto trading and, more recently, acquiring exchange Bitstamp in a $200 million deal.
Consumer Reports issues a fintech playbook can help organizations create sustainable financial products centered on financial well-being, user-centricity, and inclusivity
Consumer-advocacy organization Consumer Reports issued its “Fairness By Design Playbook” with expectations it can help fintechs build consumer trust. The new guide starts with the Consumer Reports thesis that what it calls fair design is not just beneficial to consumers, but can help organizations create sustainable financial products that consumers choose to use. It lists six components for fintechs and payments companies to address, including safety, privacy, transparency, support for financial well-being, and user-centricity, and inclusivity. Consumer Reports has used this framework as part of its evaluation of buy now, pay later services, mobile-banking apps, and digital wallets. In April, Consumer Reports, perhaps best known for its product-rating service, issued a report on digital wallets, including Apple Pay, Cash App, Google Pay, PayPal, Samsung Pay, and Venmo, which found differences in how each treats fraud monitoring and liability protection, among other elements. Digital-wallet backers scored a victory earlier this year when the U.S. Senate killed a Consumer Financial Protection Bureau rule aimed at mobile wallets and designed to bring more oversight to larger providers of digital money transfers. Consumer Reports also issued a report in 2023 on BNPL services.
Regions Bank’s sponsorship of Houston Dynamo FC to include financial education programming for consumers and business-to-business events and military appreciation by providing home improvements to a local veteran annually
Houston Dynamo FC and Regions Bank have announced a groundbreaking agreement, making Regions a proud sponsor of Houston Dynamo FC and Houston Dash. The collaboration – a first with a Major League Soccer club for the regional bank with presence across Texas, the South and Midwest – underscores the bank’s commitment in the Greater Houston community and to both organization’s commitment to growing the sport in Houston. As part of the sponsorship, the Dynamo’s premium east-side hospitality space inside Shell Energy Stadium will become the “Regions Bank Club” ahead of the 2026 season. The Regions Bank Club will provide attendees with a best-in-class matchday experience that reflects Houston’s culture, diversity, and energy. Jessica O’Neill, President of Business Operations for Houston Dynamo FC. “The Regions Bank Club will be a destination that elevates the fan experience inside Shell Energy Stadium and reflects our shared values of community and excellence. We look forward to celebrating the global game with Regions Bank customers and associates as we collectively deliver impactful programming throughout the Houston region.” “The essence of soccer promotes leadership and teamwork – qualities Regions Bank and our associates strive to embody every day in serving our clients and supporting our communities,” added Caroline Vérot Moore, Commercial Banking executive and Greater Houston market executive for Regions Bank. “Joining the Dynamo and Dash reflects Regions’ commitment to elevating Houston across a global stage. We’re grateful for this opportunity to be part of introducing new fans to the sport, creating an enhanced experience for longtime fans to more deeply enjoy it and serving our neighbors through a variety of community activations with the franchise.” Regions Bank will support the Dynamo’s ongoing commitment to military appreciation by providing home improvements to a local veteran annually through the “Rebuilding Together Project.” In addition to community engagement, Regions will work with the Dynamo on financial education programming for consumers and business-to-business events. The sponsorship also includes prominent brand exposure and activations during Dynamo and Dash home matches. Together, Houston Dynamo FC and Regions Bank will host a series of activations leading up to the 2026 season.
Intuit leverages custom-trained Financial LLMs that deliver 90% accuracy on transaction categorization, slashing latency by 50% compared to general-purpose LLMs
Intuit is announcing major GenOS enhancements that reveal how enterprises can build domain-specific AI systems that outperform general-purpose alternatives. The latest upgrades focus on three key areas: custom financial large language models, seamless expert-in-the-loop capabilities and advanced agent evaluation frameworks. The big breakthrough comes from Intuit’s new custom-trained Financial LLMs that deliver 90% accuracy on transaction categorization. That represents a marked improvement over previous models while slashing latency by 50% compared to general-purpose LLMs. For a platform already processing tens of millions of AI interactions, those efficiency gains translate into substantial cost savings and dramatically better user experiences. The key innovation lies in how Intuit approached the semantic understanding problem that plagues many enterprise AI implementations. Traditional machine learning models learn direct mappings between transactions and categories. Intuit’s Financial LLMs understand the contextual meaning behind financial terminology. The way Intuit’s Financial LLMs work is the system now actually learns what the user’s categories are because it has a better understanding of semantics. This semantic understanding
enables the models to handle personalized categorization systems. That’s a critical capability for enterprise deployments where different organizations have unique taxonomies and business rules. The training approach starts with transaction data from banks that’s been anonymized and scrubbed for personally identifiable information. Intuit then enhances the model through supervised fine-tuning and specialized guardrails built into the training process that improve semantic understanding. This methodical approach to domain-specific model training offers a template for other enterprises looking to build AI systems that outperform general-purpose alternatives in specialized domains. Beyond improving the accuracy of its Financial LLMs, Intuit is also significantly expanding its GenOS Evaluation Service within the Agent Starter Kit. While basic evaluation capabilities have existed since GenOS inception, the company is now making major investments in sophisticated frameworks that measure agent efficiency and decision quality under uncertainty. The enhanced evaluation service addresses a critical gap in enterprise AI deployments. Most companies focus on whether AI agents produce accurate results but ignore whether those results represent optimal decisions. Intuit’s GenOS evolution offers several lessons for enterprise AI teams.
Domain specialization beats generalization: Custom models trained on industry-specific data can significantly outperform general-purpose alternatives on specialized tasks. This happens despite requiring more upfront investment.
Evaluation frameworks are competitive advantages: Sophisticated measurement of AI agent efficiency and decision quality under uncertainty separates successful enterprise AI implementations from failed experiments.
Human-AI orchestration requires infrastructure: Seamless expert-in-the-loop capabilities demand purpose-built routing and handoff systems. Ad hoc human oversight isn’t sufficient.
Developer productivity compounds: Internal AI tooling investments create accelerating returns through improved developer velocity and code quality.
For enterprises looking to lead in AI adoption, Intuit’s approach suggests a clear strategy. The winning approach involves building specialized, domain-aware AI systems with sophisticated evaluation frameworks. Simply deploying general-purpose models isn’t enough.
Wells Fargo to deploy fintech’s unified trading technology across its fixed income operations, consolidating multiple venue APIs into a single normalized format
Wells Fargo has completed a sourcing agreement with TransFICC to deploy the London-based fintech company’s unified trading technology across its fixed income operations. The partnership integrates Wells Fargo’s trading systems with seven major electronic trading platforms through TransFICC’s One API service, including Tradeweb, Bloomberg, Octaura, GLMX, Aladdin, LTX and Investortools. Wells Fargo plans to connect additional venues in the future. TransFICC’s technology consolidates multiple venue APIs into a single normalized format, allowing banks to connect to various trading platforms without managing separate integrations for each. The system handles rates, credit, municipal bonds, mortgages and loan markets for Wells Fargo’s fixed income business. Wells Fargo joins a client roster that includes Citi, NatWest, NAB and Santander across TransFICC’s 17 sell-side and four buy-side institutional clients. The bank has been actively recruiting fixed income electronic trading talent, recently posting positions for lead software engineers specializing in low-latency trading platforms. Steve Toland, co-founder of TransFICC. “TransFICC will further ease Wells Fargo’s connectivity to European and Asian trading venues via our global network”.
FICO’s answer to AI risk is a foundation model that scores every output for accuracy and compliance
FICO announced the release of two foundation models. FICO Focused Language mainly deals with conversations and the language aspect of finance to determine fraud and process documentation for loans. On the other hand, FICO Focused Sequence works best for transaction analytics. Trust Score is a key component of what makes the two FICO models effective for heavily regulated industries, such as finance. The Trust Score serves as a guardrail that indicates how closely a response aligns with its training data. The Trust Score also takes into account context found in the data. So if the model is used to read through documentation about European financial instruments, the Trust Score can see if the response is relevant. A response with a high score indicates that it is accurate in terms of its data coverage and is not misleading. Responses with low scores may prompt the bank to review its data or refine how the model responds. FICO FLM works best on understanding the language used in transactions. It has two general use cases. The first is for compliance and communications. It understands the rules governing how financial institutions can and should provide information to customers and extract information from conversations. What is special about FLM is that since it monitors the back and forth between a bank and a person, it can detect if the customer is undergoing some financial hardship. The bank can tailor its approach to providing information to them, taking into account their economic position. The second use of FLM involves underwriting, which is the act of offering a loan or capital to an individual or a business. The model can take into account the person’s interactions with the bank and review loan documentation. FICO FSM deals with transaction data. “The architecture is different; it has something called a contrastive head and a supervised head,” Scott Zoldi, chief analytics officer at FICO said. “The contrastive head says, is this transaction in or out of pattern, while the supervised head says, is this change in behavior fraud or not. The supervised task knows the probability of fraud, the fact that she has hardship, and we have to intervene.”