Within a rapid 72-hour span this month, two leading artificial intelligence companies, Anthropic and OpenAI, launched significant initiatives signaling a strategic pivot from model development to enterprise deployment. Both firms unveiled dedicated arms for enterprise services, announced major partnerships within the financial services sector, and released agent tooling specifically designed to streamline Wall Street workflows. This coordinated surge underscores a unified message: the next evolutionary phase of frontier AI is not solely about groundbreaking models, but critically about their practical, widespread implementation and adoption by businesses.
For developers and enterprises alike, the implications of this accelerated push toward deployment are still unfolding, promising to reshape how AI is integrated into core business operations and create new opportunities and challenges within the tech ecosystem.
The Strategic Land Grab for Enterprise AI Services
The past week has witnessed a concerted effort by AI giants to capture the burgeoning enterprise AI market, particularly in the complex and highly regulated financial services sector. This strategic maneuver highlights a growing recognition that the true value of advanced AI lies not just in its theoretical capabilities, but in its ability to be effectively deployed and utilized within real-world business contexts.
Anthropic, bolstered by substantial investment from firms like Blackstone, Hellman & Friedman, General Atlantic, Apollo, Goldman Sachs, and Sequoia Capital, has established a new services firm targeting mid-sized enterprises. These organizations, often overlooked by larger consulting and systems integration firms, include community banks, regional health systems, and mid-market manufacturers. Anthropic’s approach involves embedding Applied AI engineers directly with clients, working alongside the firm’s dedicated engineering staff. This hands-on model focuses on workflow discovery, the development of custom Claude-powered solutions, and long-term client support, aiming to bridge the gap between advanced AI capabilities and the practical needs of these businesses.
In parallel, OpenAI has launched its own enterprise deployment initiative, branded as "DeployCo." This venture operates at the higher end of the market, focusing on large enterprises and employing a similar forward-deployed engineering model. OpenAI’s strategic acquisition of Tomoro, an applied AI consulting firm, immediately provides DeployCo with an estimated 150 experienced Forward Deployed Engineers (FDEs). Backed by over $4 billion in initial investment and strategic partnerships with major consultancies like McKinsey, Bain & Company, and Capgemini, DeployCo is positioned to tackle the most complex AI integration challenges for global corporations.
Both companies are operating under a shared thesis: the "deployment gap"—the widening chasm between the capabilities of frontier AI models and the actual number of AI-powered solutions shipped by enterprises—represents the next significant revenue opportunity. This convergence of strategy and execution within the same week underscores the urgency and immense potential perceived in this market segment.
Brad Shimmin, an analyst at The Futurum Group, commented on the surprising scale of this opportunity, stating, "I would say there is a tremendous, somewhat surprising opportunity here. Even within highly regulated industries, where inaccuracies are not tolerated, generative and agentic AI promises to reinvent the way both consumers and financial professionals work with data."
Jason Cutler, SVP of Anthropic Consulting and Engineering at Caylent, an AWS Premier Partner that has launched a dedicated Anthropic practice, observed the shift firsthand. "A year ago, there was a lot of concern—is Claude going to take the work of services partners like us?" Cutler told The New Stack. "And just in the last week, we’ve seen Anthropic investing in a services company, OpenAI investing in a services company, Google hiring FDEs. We know there’s a need."
Cutler further elaborated on the phases of enterprise AI maturity he observes in the field: training and enablement, operating model and governance, and transformation. He noted that many organizations are still predominantly stuck in the first phase, where AI is introduced into existing processes without fundamentally reimagining them. "You can bring AI into an existing process, but you’re not getting the full advantage of AI yet. You really have to recreate the process alongside AI and your employees to really benefit," he explained.
Finance Emerges as the Proving Ground for Enterprise AI
The financial services sector has quickly become a focal point for both Anthropic and OpenAI’s enterprise AI strategies. This focus is driven by the industry’s inherent data intensity, complex regulatory environment, and a clear demand for efficiency gains and enhanced analytical capabilities.
On May 4th, PwC and OpenAI announced a significant collaboration aimed at developing AI agents tailored for the core operating functions of a Chief Financial Officer’s office. This initiative encompasses areas such as financial planning, forecasting, reporting, procurement, payments, treasury, tax, and the month-end accounting close. OpenAI is leveraging its own finance organization as "customer zero," developing a procurement agent internally before extending these patterns to enterprise clients. Early successes reported by OpenAI include processing five times as many contracts with the same headcount using Codex and managing over 200 investor interactions during a recent fundraise with an IR-GPT tool.
The following day, Anthropic released ten pre-built agent templates designed to address some of the most labor-intensive workflows in finance. These templates target critical functions like pitch building, Know Your Customer (KYC) screening, month-end close, general ledger reconciliation, earnings review, and underwriting. Available as plugins within Claude Cowork and Claude Code, and as cookbooks for Claude Managed Agents, these tools are enhanced by new data connectors from leading providers such as Dun & Bradstreet, Verisk, SS&C Intralinks, Third Bridge, and Moody’s. These connectors provide agents with access to live, industry-specific data essential for financial professionals. Anthropic reports that its Claude Opus 4.7 model currently leads Vals AI’s Finance Agent benchmark with a score of 64.37%.
Sanjay Subramanian, PwC Partner and US & Global Anthropic Alliance Lead, highlighted the compelling value proposition for regulated industries like financial services. "Financial services companies are looking at documents," Subramanian explained to The New Stack. "Intelligent document processing with both structured and unstructured data—that’s been one of the great use cases."
He cited a specific engagement with an insurance company where an underwriting cycle was dramatically compressed from ten weeks to just ten days. This transformation was achieved through a phased deployment: initial backtesting against historical outcomes, followed by co-delivery with human oversight, and ultimately, agents providing first-pass deliverables for underwriters to review at key checkpoints. Subramanian emphasized that liability remains with the human reviewer, stating, "The human still has to review it before it goes to the next level. We’re not changing that process. I think it’s too early to change that process."
Understanding the Limitations: What Breaks in AI Deployment
While the potential for AI in finance is immense, not all use cases have proven equally successful. Sanjay Subramanian identified a pattern in AI deployment failures, pointing to challenges with high-variance, unpredictable input. "A supply chain company where they’ve got lots of parts that need to be fixed—if those parts are so diverse, the questions are so diverse, there’s less precision around that outcome," he noted.
Conversely, successful applications tend to be deterministic and back-testable, such as in ticketing systems, underwriting processes, and document review against established policies. Open-ended customer service scenarios, where the range of potential questions is effectively unbounded, have proven more challenging.
"The quality of these models is going up and up. The ability for companies to deploy them is not keeping up. That gap is increasing," Subramanian observed, emphasizing the organizational hurdles that often impede AI adoption. Chief Information Officers (CIOs), accustomed to cost containment, often resist the upfront investment required to rebuild legacy infrastructure and processes to fully leverage AI. "That’s probably one of the toughest things to get people comfortable with—that reinvention."
Jason Cutler echoed this sentiment from a developer’s perspective, highlighting that the governance conversation has become paramount. "When we see things like PHI and credit card authorizations and sensitive information, we need to make sure we’re setting that up correctly on the foundational layer, so that customers feel safe that by leveraging AI, they’re not out of bounds from what they have to do from a compliance standard," he stated.
The Evolving Role of Junior Developers in the AI Era
Both Cutler and Subramanian addressed the question of whether tools like Claude Code will displace junior developers or offer them new avenues for growth. They largely pushed back against the notion of widespread displacement, albeit with nuanced perspectives.
"In some cases, junior developers seem to be catching on even faster," Cutler observed. "In the age of AI, some people are catching up very, very fast, and their own curiosity is lending itself to leveraging the tools effectively." Caylent has developed a "Playbook Catalyst" initiative to systematically capture how developers are utilizing Claude Code across their organizations, identify best practices, and use this knowledge to drive broader team enablement.
Subramanian framed the impact in terms of a "baseline shift," where AI redefines normative standards for quality and time-to-value. "What it’s going to do is re-baseline what normative is—what quality is, what the expectation of time to value is. But it also means that if you’re a new developer, you’re able to learn quicker, you’re able to test things out quicker. We can create automated packages to review your code, to coach you." He suggested that AI can effectively substitute for senior developer mentorship, which is often inaccessible to junior developers, accelerating their learning curve.
On the front of COBOL modernization—a critical concern for financial institutions reliant on legacy systems—Subramanian described a dynamic mirroring broader trends. Senior developers who were initially skeptical of Claude Code are finding that it enhances, rather than diminishes, their capacity. "They’re not spending time in meetings where developers working under them need to ask questions," he explained. "They’re actually able to reduce time to value in their deployment capabilities, and allow senior developers to spend time really building new capabilities as they transform from an old code base to a new code base."
Navigating the "Fox in the Henhouse" Debate
The strategic partnerships announced by Anthropic and OpenAI have not been without controversy. Venture capitalist Chamath Palihapitiya issued a stark warning on X following the DeployCo announcement: "If you are running a consulting business and you are deploying Anthropic or OpenAI directly into your organization (I’m looking at you, PwC and Accenture) you are letting the fox into the henhouse."
Palihapitiya’s argument centers on the potential for AI labs to simultaneously fund and build competing services while leveraging client data to enhance their own offerings. "OpenAI and Anthropic are openly funding and starting competitors to you while also using your usage to drive more success for them. This is not a failure on their part but a failure on your part."
This concern is particularly relevant given PwC’s formal partnerships with both Anthropic and OpenAI. PwC is part of the Claude Partner Network alongside Accenture and Deloitte, and is simultaneously co-developing finance agents with OpenAI’s internal finance organization. Anthropic’s new services firm targets a market segment just below PwC’s traditional client base, while OpenAI’s DeployCo, with investment from McKinsey and Bain & Company, directly competes in the large enterprise space where PwC operates.
However, Jason Cutler of Caylent expressed a different perspective, viewing DeployCo’s launch as a validation rather than a threat. "It shows that private equity companies know that this work is going to need to get done. I think it actually validates Caylent going out and creating this practice, because the business is out there today," Cutler told The New Stack.
The ultimate impact of these competitive dynamics—whether Palihapitiya’s warning proves prescient or Cutler’s optimism is well-founded—may hinge on the scalability of the AI labs’ own deployment arms. For developers actively engaged in building AI solutions within the financial services sector, the immediate reality is that the foundational infrastructure is being established, agent templates are shipping, and certifications are being developed. The work, undeniably, is present.
