Stay in control amid Microsoft and OpenAI’s agentic land grab

OpenAI and Microsoft are making a land grab for your use of agentic AI by embedding intelligent, autonomous agents directly into the enterprise technology stack. This presents opportunities to boost efficiency and productivity but also introduces challenges around governance, workforce readiness, and vendor dependency.

• OpenAI’s ChatGPT Scheduled Tasks introduces the ability to automate routine processes such as scheduling meetings or sending reports directly within ChatGPT. It integrates with third-party apps such as Outlook, Slack, and others.
• Microsoft Copilot’s enhanced agentic capabilities, announced in January by CEO Satya Nadella, enable enterprises to build intelligent agents that automate cross-platform tasks and provide real-time insights across tools such as Teams, Power Platform, and Azure.

These developments embed ’agentic’ capabilities—AI systems that act autonomously—into widely used platforms, offering cost-effective alternatives to traditional automation solutions.

The rise of AI—and its ready application to enterprise workflows in the form of agents—is inspiring most enterprise leaders to reconsider the mix of services and technology they need to get work done (in line with our 2030 Tech-Services vision). Our most recent (2024) Pulse Survey (see Exhibit 1) shows that almost three-quarters of enterprise leaders plan to renegotiate their IT services contracts, and nearly two-thirds want to renegotiate software and SaaS deals.

Exhibit 1: Companies are ready to shift what they need to spend on getting work done

Sample: N=605 enterprise leaders
Source: HFS Pulse, 2024

Embedding puts agentic in the hands of every employee

The rise of embedded agentic solutions is changing the way enterprises should approach work automation. OpenAI and Microsoft deliver AI automation directly within tools employees already use, eliminating the need for standalone automation solutions, such as robotic process automation (RPA). Non-technical users can deploy agents to handle repetitive workflows without requiring custom development, empowering more employees to experiment with AI.

By embedding agentic AI into their platforms, Microsoft and OpenAI have thrown down the gauntlet to traditional automation providers such as UiPath and Automation Anywhere. Enterprises that rely on these standalone solutions must now assess whether embedded capabilities offer a more cost-effective alternative.

Act now to avoid an enterprise-wide free-for-all

The ease of deploying agentic AI that OpenAI and Microsoft offer increases the risk of uncoordinated adoption, potentially leading to fragmented processes and compliance issues. Enterprises must prioritize governance frameworks to regulate how and where these tools are used.

Establish governance frameworks: Define policies deploying and integrating AI to ensure compliance, security, and accountability. Identify data usage boundaries and establish oversight mechanisms.

Start with controlled pilots: Focus on high-impact, manageable use cases such as IT service automation, customer support, or routine internal workflows. Test agentic tools in isolated environments before scaling broadly.

Assess automation investments: Evaluate whether existing RPA or custom automation solutions remain relevant as embedded AI becomes available. Explore opportunities to consolidate or augment these investments.

Upskill employees: Invest in AI literacy and training programs to prepare employees for collaboration with agentic tools.

Balance vendor relationships: While leveraging Microsoft or OpenAI’s agentic capabilities, maintain flexibility by diversifying platforms where possible. Avoid over-reliance on a single vendor’s ecosystem.

Enterprise-wide vs. domain-specific—the battle lines are being drawn

While Microsoft and OpenAI focus on embedding agentic AI directly into existing tools such as Teams, Power Platform, and ChatGPT—leveraging their ecosystems to deliver user-friendly, cross-platform agents—examples such as ServiceNow take a slightly different approach. The Now Platform AI has agentic AI capabilities to streamline enterprise workflows such as IT service management, HR, and operations. Its AI-driven agents are primarily process-specific and tied to predefined workflows, often requiring configurations to address particular enterprise needs. While this approach delivers precision for specific use cases, it lacks the broad accessibility of Microsoft and OpenAI’s tools, which aim to democratize automation for non-technical users across multiple business functions.

While Salesforce’s Einstein GPT placed it among vendors focused on domain-specific agentic solutions (integrating AI directly with Salesforce’s CRM data), its recently announced Agentforce 2.0 pitches it into the ‘enterprise-wide’ battleground. With Agentforce, Salesforce wants to be the operating system for your digital workforce—complete with simplified agent creation using an ‘Agent Builder’ to automatically generate ‘digital workers’ from natural language descriptions. Read more about our take on Agentforce here: From physical to digital labor: Will Agentforce 2.0 replace traditional jobs and outsourcing?

Exhibit 2: A mix of approaches to agentic AI is likely to be required by most enterprises

Source: HFS Research, 2025

A bottom-up approach leaves end users to figure out the value

Microsoft and OpenAI’s embedded, platform-agnostic approaches may be more scalable across enterprise ecosystems, enabling broader adoption by knowledge workers without requiring heavy customization or technical expertise.

However (see Exhibit 2), the domain-specific approach may be a better fit where controlled adoption aligned to known needs for specific outcomes, defined from the top, is a priority.

The Microsoft/OpenAI approach is more typically ‘tech’—building a technology and putting it in the hands of ‘bottom-up’ users to figure out how to create value with it. Both have their uses in today’s enterprise.

The Bottom Line: Be alert to control issues, stay focused on outcomes, and expect to have to mix and match your agentic solutions.

Microsoft and OpenAI’s latest developments are reshaping the enterprise automation market by embedding intelligent agents into existing platforms. This accessibility will accelerate adoption, reduce costs, and challenge traditional automation providers. However, enterprises must manage the risks of fragmented workflows, data security issues, and vendor dependency.

Choosing between the land-grabbing ‘bottom-up’ approach to finding value vs. the targeted top-down domain-specific approach depends on whether your organization’s immediate priorities are ecosystem-wide enablement or domain-specific transformation. Given that most large enterprises need both, be prepared to experiment with a mix of approaches.