Is a backlash brewing? Rapid innovation in AI coding and agents may force push for enterprise order and control

Artificial intelligence is proving to be a big bet for many companies across the enterprise landscape, and the gamblers are getting nervous.

A survey of 2,400 global employees and C-suite leaders released this week by the agentic AI platform Writer Inc. found that 79% of executives acknowledge struggling with AI-related issues such as lagging ROI, strategy gaps and internal power struggles. Nearly 40% of CEOs confessed to a “high or crippling amount of stress” when it came to deployment of their firm’s AI strategy.

Results from the Writer survey highlight the high-stakes environment that surrounds AI deployment, as enterprises deal with a dizzying array of new model releases, rising costs and growing security concerns. AI has created tension between information technology organizations and other lines of business, with more than half of the respondents reporting AI use as a “chaotic free-for-all.”

This unsettled climate may be setting the stage for a day of reckoning, a sentiment echoed by one industry leader whose company has been squarely in the middle of the agentic AI revolution.

“There are so many boards right now who have no idea how much of their data is hemorrhaging outside their organizations,” said Boomi LP Chief Executive Steve Lucas, who spoke with SiliconANGLE during the HumanX conference in San Francisco this week. “When chaos reigns, boards demand order.”

Seeking to harness AI

Lucas was an early voice in predicting the rise of agentic AI. He devoted a significant amount of his Boomi World keynote back in the spring of 2024 to describing how the use of AI agents would hasten the process of reimagining enterprise applications.

In recent weeks, Boomi has moved to provide more resources for trust and governance in the use of AI. The company’s recently unveiled Meta Hub aligns data standards across the enterprise to ensure that AI agents and humans can rely on consistent, trusted business logic and governance.

“There is going to be a backlash,” Lucas said. “We want to be the enterprise harness for AI.”

Achieving that measure of control may be a challenge given the pace of agentic automation, enabled by a growing array of sophisticated coding tools. Along with existing widely used tools such as Anthropic PBC’s Claude Code and OpenAI Group PBC’s Codex, new AI-based coding solutions have automated a wide range of programming tasks including the creation of new application features and an ability to debug existing ones.

Last week startup Cursor, which has raised more than $3 billion in funding from tech heavyweights such as Nvidia Corp. and Google LLC, unveiled a new version of its AI coding platform with a chatbot interface that employs multiple AI agents to complete user-specified tasks. Cursor’s growing popularity, with a reported 1 million daily users, underscores the rising influence of AI coding platforms in the enterprise, according to Cursor co-founder and CEO Michael Truell, who spoke during HumanX on Tuesday.

“About 65% to 70 % of enterprise code is now written by AI,” Truell said. “Coding agents are becoming critical infrastructure for companies. We’re on this long arc of a coding vendor reaching the same level of importance as your cloud.”

Coding and models stir debate

Despite the success of coding platforms such as Cursor, not all of the leading figures in the AI world believe that software development will or should be completely automated. Andrew Ng, who started and led Google Brain, offered a differing view during an appearance at the conference on Wednesday, taking issue with statements by some that human coding will become unnecessary.

“We’ll look back on that as the worst piece of advice ever,” Ng told the gathering. “I think everyone should learn to code.”

Contrarian opinions such as Ng’s are examples of the chaotic state of the AI world in 2026. This was further underscored by news from model provider Anthropic earlier this week.

On Tuesday, the company said it would release a preview of Mythos, the most powerful frontier model it has ever created. But the catch is that it would be available only to a small number of partners and cybersecurity researchers, based on an apparently powerful ability to identify vulnerabilities in proprietary and open-source software. The company said it was limiting release for “public safety and national security” purposes.

The move drew fire from some in the industry for not disclosing the unprecedented cybersecurity risks the model has apparently uncovered. However, the chairman of OpenAI went on record during an appearance at the conference this week to support Anthropic’s decision.

“Conceptually, I think it’s a really smart idea,” said Bret Taylor, former co-CEO of Salesforce Inc. and co-founder of Sierra Inc. “I think it’s a part of responsible, iterative deployment. This is the type of thing we need to know how to do as research labs.”

Confronting the hype cycle

In the larger view of the world, the debate continues among tech leaders over whether AI is overhyped and underperforming. In an appearance at HumanX on Tuesday, Amazon Web Services Inc. CEO Matt Garman expressed a lack of concern over talk of a “bubble” for AI.

“I’m very bullish that there is much more to come,” Garman told the gathering. “The last big bubble we had was the internet bubble. The last time I checked, the internet is pretty big.”

However, Databricks Inc. CEO Ali Ghodsi offered a different point of view. He has been outspoken in recent months about the skyrocketing valuations of AI startups, claiming that venture capitalists have privately told him they’re becoming exhausted by the current hype cycle.

“I think it’s still very highly bubbly,” Ghodsi said on Tuesday. “There’s a gap between artificial general intelligence we already have and the agency side of enterprise. [Agents] are missing the context. The next model release is not going to fix that.”

Though the next large language models may not bring AI fully to where it needs to be, the major model providers continue to expand their clout and market impact. Reports of an initial public offering later this year for OpenAI have placed its valuation in the neighborhood of $1 trillion. Anthropic’s own offering is currently projected to raise more than $60 billion on a $380 billion valuation.

Against this remarkable wealth remains a central question that was a focus of many hallway discussions at HumanX this week. Where exactly is all of this going and what will it mean for mere humans?

Much as he predicted the agentic revolution two years ago, Boomi’s Lucas has a pretty good idea of where the road leads, at least as far as the two largest model providers are concerned.

“I know the end game here,” Lucas told SiliconANGLE. “The end game for Anthropic and OpenAI is to create a digital version of every single one of us. It’s not just the system or the apps, it’s the humans themselves.”

Photos: Mark Albertson/SiliconANGLE