Thrive Holdings’ efforts to modernize its accounting and IT services are entering a new phase as OpenAI acquires an equity stake in the company and prepares to deploy its own expertise within Thrive’s operations. In doing so, OpenAI is testing an AI-driven model that combines capital, sector expertise, and an embedded technology team.

Earlier this year, Thrive established a holding company to acquire and manage companies in the everyday services industry. The aim was to rebuild these companies with more efficient processes, new data practices, and the practical application of AI. With OpenAI’s deeper involvement, the idea turned into a real-time experiment in how traditional providers can update their work without relying solely on off-the-shelf tools.

Test cases for bringing AI into core operations

While many corporate conversations around AI tend to revolve around pilots and proofs of concept, Thrive is taking a different approach by acquiring companies outright and redesigning how they operate. Our two current businesses, Crete Professionals Alliance (accounting) and Shield Technology Partners (IT services), employ more than 1,000 people. Thrive committed $500 million to Crete and partnered with ZBS Partners to contribute more than $100 million to Shield.

For companies looking in from the outside, the appeal is clear. These industries have heavy workloads, manual labor, and tight margins. We also handle sensitive data and work under strict deadlines. AI systems deployed in that environment require domain context, training, and tuning to fit local processes, rather than general automation.

Crete has already started leveraging AI to reduce routine tasks such as data entry and early stage tax workflows. Shield plans to complete 10 acquisitions by the end of the year, providing Thrive with the foundation for its IT operations and redesigning them with new tools and methodologies.

What OpenAI gives you

OpenAI is under pressure to find real enterprise-scale use cases for its models. Investors value the company at about $500 billion, and its long-term commitments include about $1.4 trillion in infrastructure investments through 2033. To justify this number, companies are betting on spending big on tools that help them work faster and handle complex tasks at scale.

By acquiring a stake in Thrive Holdings, OpenAI gains things it couldn’t create on its own, including access to companies that can experience models in their daily work and specialist training on real-world operations. As Thrive’s company grows, OpenAI’s stake could grow, according to people familiar with the deal.

“We are excited to expand our partnership with OpenAI and incorporate their frontier models, products and services into areas where we see tremendous potential to benefit from innovation and adoption,” said Joshua Kushner, founder of Thrive Capital and Thrive Holdings.

This partnership also gives OpenAI an avenue to glean value from the engineering support it provides. Anuj Mehndiratta, a partner who oversees product and technology strategy at Thrive Holdings, said his team will develop custom models for Thrive companies and send researchers and engineers to the field.

What companies can learn from this approach

For many companies, the most difficult part of using AI is not the model, but redesigning existing work. Thrive’s strategy reflects a move toward deeper integration, where AI teams reside within the business units they support, rather than acting as external advisors.

This model allows companies to:

• Build tools that are shaped around real-world workflows rather than abstract use cases
• Train models on controlled, high-quality data
• Reduce the gap between engineering teams and frontline employees
• Direct feedback from staff speeds test changes

It also reveals the true cost of AI implementation. Custom work requires engineering time, domain knowledge, and long-term coordination between owners and model developers. Thrive and OpenAI’s partnership formalizes that collaboration in a way that could become more commonplace as companies seek results rather than demonstrations.

“Our partnership with Thrive Holdings is designed to demonstrate what’s possible when we rapidly deploy frontier AI research and deployment across an organization, revolutionizing the way companies work and engage with their customers,” said Brad Lightcap, COO of OpenAI.

broader competitive environment

The deal comes at a time when AI companies are trying to establish themselves within the accounts of major companies. Anthropic is reaching more businesses through its partnership with Microsoft. Google is attracting interest in its latest model, increasing its market value as companies explore new AI options. Meanwhile, OpenAI has acquired stakes in partners such as AMD and CoreWeave to support its long-term infrastructure needs.

OpenAI also expanded its reach this Monday, announcing another deal with Accenture. The company’s ChatGPT Enterprise product will be deployed to “tens of thousands” of Accenture employees, giving OpenAI a new path to large-scale enterprise use.

Possible blueprint

If Thrive companies demonstrate meaningful improvements in the way they operate, this model could influence how other companies think about AI transformation. Rather than layering tools on top of old processes, some companies may move toward deeper restructuring, guided by technical teams that understand both the model and the business.

For now, Thrive Holdings serves as a real-world case study of what that approach looks like when applied to an industry that rarely makes technology headlines but is fundamental to day-to-day business operations.

See: AI Business Realities – What Business Leaders Need to Know

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expos in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.

North American companies are now actively deploying agent AI systems that aim to reason, adapt, and act with complete autonomy.

Digitate’s three-year global program data shows that while adoption is universal across the board, regional maturation paths diverge. While North American companies are scaling towards full autonomy, European companies are prioritizing governance frameworks and data management to build long-term resilience.

From practicality to profitability

The story of enterprise automation has changed. In 2023, the main goals for most IT leaders were to reduce costs and streamline daily operations. By 2025, the focus will be further expanded. AI is no longer seen as just an operational utility, but as a benefit-producing capability.

Data supports this shift in perspective. According to the report, organizations in North America saw a median return on investment (ROI) of $175 million from implementation. Interestingly, this financial validation is not unique to the fast-moving North American market. European companies report comparable median ROIs of around $170 million, despite a more cautious and governance-focused approach.

This consistency suggests that although deployment strategies differ, such as a focus on risk management in Europe and a focus on speed in North America, the financial outcomes are similar. All organizations surveyed confirmed that they have leveraged an average of five different tools to implement AI within the past two years.

Generative AI remains the most widely deployed at 74%, but we are seeing a notable increase in “agent” capabilities. More than 40% of enterprises have deployed agents or agent-based AI, moving beyond static automation to systems that can manage goal-oriented workflows.

IT operational autonomy becomes a testing ground for agent AI

While marketing and customer service often dominate public discussions about AI, the IT sector itself is emerging as a key laboratory for these deployments. IT environments are inherently data-rich and structured, creating ideal conditions for models to learn while remaining sufficiently dynamic to require the adaptive inference that agentic AI systems promise.

This explains why 78% of respondents have deployed AI within IT operations, the highest percentage of any business function. Cloud visibility and cost optimization leads the adoption curve at 52%, followed closely by event management at 48%. In these scenarios, rather than alerting humans to problems, technology proactively interprets telemetry data to provide a unified view of spending across the hybrid environment.

Teams leveraging these tools report improved decision-making accuracy (44%) and efficiency (43%), allowing them to handle higher workloads without increasing escalations.

Cost and the human conundrum

Despite the optimism around ROI, the report highlights “cost and human challenges” that threaten to slow progress. This contradiction is straightforward. Businesses are deploying AI to reduce their dependence on labor and operational costs, but these very factors are acting as a major inhibitor to growth.

47% of respondents cited the continued need for human intervention as a major drawback. These agent AI systems are far from achieving the full autonomy of “set-it-and-forget” solutions and require continuous monitoring, adjustment, and exception management. At the same time, implementation costs were the second-highest concern at 42%, driven by costs associated with model retraining, integration, and cloud infrastructure.

The human resources needed to manage these costs are lacking. For 33% of organizations, a lack of technical skills is the primary barrier to future adoption. The demand for professionals who can develop, monitor, and manage these complex systems exceeds the current supply, creating a self-reinforcing loop in which investment increases operational capacity while simultaneously increasing human and financial dependence.

Gap between trust and awareness

Differences in perspective exist between executives and practitioners. Although 94% of all respondents expressed trust in AI, this trust is not evenly distributed. C-suite leaders are significantly more optimistic, with 61% categorizing AI as “very trustworthy” and viewing AI primarily as a financial tool.

Only 46% of non-executive practitioners share this high level of trust. Those close to the day-to-day operations of these models are more aware of the reliability issues, lack of transparency, and need for human oversight. This gap suggests that leadership is focused on long-term review and autonomy, while teams on the ground are grappling with practical delivery and governance challenges.

There are also different views on how these agents work. 61% of IT leaders view agent systems not as replacements but as collaborators that augment human capabilities. However, expectations for automation vary by industry. In retail and transportation, 67% believe agent AI will change the key tasks of their roles, while in manufacturing, the same percentage see these agents primarily as personal assistants.

Full agent AI autonomy is rapidly approaching

We expect the industry to rapidly move towards reducing human involvement in routine processes. Currently, 45% of organizations operate as semi-autonomous or fully autonomous enterprises. Projections suggest this number will rise to 74% by 2030.

This evolution means a change in the role of IT. As capabilities mature, IT is expected to move from being an operational enabler to an orchestrator role. In this model, IT functions manage “systems of systems” and enable various intelligent agents to interact correctly while humans focus on creativity, interpretation, and governance rather than execution.

Avi Bhagtani, CMO of Digitate, said, “Agentic AI bridges the gap between human ingenuity and autonomous intelligence, and heralds the dawn of IT as a strategic, profitable function.” “Companies have moved from experimenting with automation to extending AI for measurable impact.”

Moving to agent AI requires more than just software procurement. We need an organizational philosophy that balances automation and human augmentation. Policies alone are not enough. To ensure transparency and ethical oversight in all decision-making loops, governance must be integrated directly into system design. European organizations are currently leading the way in this area, prioritizing ethical deployment and strong monitoring frameworks as the basis for resilience.

Furthermore, the shortage of technical human resources cannot be solved by recruitment alone. Organizations should invest in upskilling their existing teams by combining operational expertise with data science and compliance literacy.

Finally, reliable autonomy depends on high quality data. Investments in data integration and observability platforms are necessary to provide agents with the context they need to act independently.

The era of experimental AI is over. The current stage is defined by the pursuit of autonomy, whose value comes not from novelty but from the ability to scale agent AI sustainably across the enterprise.

“As organizations balance autonomy and accountability, those that embed trust, transparency, and human engagement into their AI strategies will shape the future of digital business,” concludes Bagtani.

See also: How updates to the MCP specification improve security as your infrastructure grows

Want to learn more about AI and big data from industry leaders? Check out the AI ​​& Big Data Expos in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events including Cyber ​​Security Expo. Click here for more information.

AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.