Big banks back AI while keeping core software in place

JPMorgan Chase and Anthropic have signalled that large organisations still expect to run established enterprise software alongside newer AI tools, despite investor concerns that AI agents could undermine subscription-based business models.

In a private webcast with investment clients, leaders from JPMorgan's internal IT unit said the bank had no immediate plans to use AI to replace traditional enterprise applications. They also indicated the bank expected to increase spending with some existing providers that are adding AI features, according to accounts posted on X and confirmed by a bank spokesperson.

The comments land at a tense moment for listed enterprise software groups. Investors have questioned whether generative AI and task-based "agents" could reduce the need for large numbers of paid user seats across systems used for customer management, HR, service management and content creation.

Shares in several large software providers have fallen sharply over the past year, including Salesforce, Workday, HubSpot, ServiceNow and Adobe.

Bank spending

JPMorgan has one of the biggest technology budgets in corporate IT. The bank spends about USD $18 billion a year on technology and employs more than 300,000 people. That scale gives it a strong incentive to seek efficiency gains from automation, including through AI tools.

Even so, the bank's IT leaders did not signal a near-term move away from existing enterprise application providers. The tone contrasted with a common market narrative that AI agents will quickly reduce the number of paid users required for major software suites.

The disclosure also offers a snapshot of how a large buyer frames AI adoption. Many suppliers now bundle AI functions into existing products. Some also introduce add-on pricing for features such as generative writing, summarisation, workflow assistance and natural language interfaces.

Anthropic's tools

Anthropic, which develops the Claude model and related tools, also uses established enterprise applications internally, according to publicly visible hiring and partner material.

A job posting for a Head of IT Regulatory Compliance listed experience with Workday, Salesforce and NetSuite among examples for successful candidates.

Anthropic also agreed to let Intercom, a customer service provider, describe Anthropic's choice of vendor rather than building its own customer service agent. "Anthropic could have easily chosen to build an in-house AI agent to power its customer service...the team saw a greater opportunity to partner with a company with deep expertise in AI-first customer service so they could remain laser focused on their goal of building safe, reliable, and transformative AI for humanity," said Intercom's published account.

The relationship also reflects a broader pattern in the AI supply chain. Intercom uses Anthropic's AI in its product. Several enterprise application providers, including Microsoft, Canva, Salesforce and Figma, use Anthropic models for some AI functions.

Anthropic has also expanded Claude's features for working with Excel spreadsheets, a common file format across enterprise workflows and reporting. Microsoft, which competes in AI tools for knowledge workers, also owns Excel as part of its Office suite.

Employee sentiment

Some commentary from within Anthropic has also underscored the continued role of familiar enterprise systems. An Anthropic design lead for Claude posted on X: "worst part about working at a company that keeps getting more successful is that gradual switch to enterprise-grade internal tools that no one likes using (see: workday)." The employee later appeared to delete the post.

The remark echoed a long-standing theme among users of large HR and finance platforms. Customers often criticise user experience and workflow friction. Many still keep such systems in place for compliance, local rules, security expectations and the operational cost of replacement.

Workday has positioned AI as an additive development rather than a replacement risk. "AI is a tailwind for us, it's absolutely not a headwind," said Carl Eschenbach, Chief Executive Officer, Workday. "We are uniquely positioned to be one of the AI winners in the enterprise because of our incumbency, and lastly, because of the trust we get from our customers," said Eschenbach.

Pressure points

The outlook does not remove competitive risks for enterprise application providers. Many have seen slower growth compared with some database, infrastructure and cloud firms that benefit from rising AI-related compute demand. They also face open questions on how to price AI features and automation in ways that protect revenue while matching customer expectations.

AI chat tools could also shift how employees interact with enterprise systems. Anthropic has shown ways for Claude to access workplace applications such as Figma, Canva and Asana from within the chatbot interface. That raises the prospect of AI tools becoming a primary front end for work systems, even if the underlying enterprise applications remain in place.

Microsoft has also discussed a future in which AI agents perform tasks inside enterprise systems. That could reshape the commercial model if software companies sell subscriptions tied to AI "workers" rather than human users.

Outside large incumbents, startups and internal engineering teams continue to argue that AI can generate bespoke applications and reduce dependency on standard SaaS contracts. Some executives counter that developer effort often yields higher returns when focused on core products and revenue drivers. "It's foolish to think that I would spend any time saving a few bucks at the opportunity cost to the core business," said Saumil Mehta, Global President, Ticketmaster.