From copilots to coworkers: AI agents take on real work.

This week, leading enterprise software firms made it clear they are gearing up for an agentic future. At Salesforce’s Dreamforce conference, CEO Marc Benioff introduced **AIforce**, a new “live interface” layer designed to let AI agents work directly with Salesforce data, business logic and workflows from any application. In practice, this means employees or AI assistants can complete Salesforce tasks within tools like Slack, Microsoft Teams, or ChatGPT, without ever opening Salesforce itself. As Benioff explained, it’s a shift from software being the place you go to work, to Salesforce’s software powering work in whatever interface you’re already using. By bringing CRM context and rules to where people and AI are working, Salesforce is aiming for what analysts call “ambient” or headless enterprise software.

Another major enterprise player, **UiPath**, announced a similar leap in integration. It has transformed its Integration Service into an “agent-ready” connectivity layer so that AI-driven **coding agents** can seamlessly interact with legacy enterprise apps. Typically, connecting AI or automation tools to systems like Workday, ServiceNow, or SAP requires either pre-built connectors or painstaking custom code, which can break easily and pose security risks. Now, UiPath’s platform lets AI agents generate much of that integration code themselves, while enforcing centralized controls on credentials, API access, and data flow. This approach allows companies to plug advanced AI into existing workflows faster and with fewer headaches for IT departments.

Taken together, these moves by Salesforce and UiPath show that large vendors are racing to make **autonomous agents a native part of enterprise infrastructure**. By providing robust hooks into their platforms – and managing the complexities of authentication, context management, and uptime – they lower the barrier to entry for businesses looking to put AI to work. The implication for leaders is that integrating intelligent agents into core systems is getting easier. Organizations that capitalize on these new capabilities can begin to transform how work gets done without waiting for a full rip-and-replace of existing software.

If 2023’s buzz was about AI “copilots,” 2026 is increasingly the year of AI **coworkers**. The Adecco Group, one of the world’s largest staffing companies, has just announced a global rollout of Salesforce’s Agentforce Coworker to more than 27,000 employees across 40+ countries. This virtual “AI teammate,” powered by Anthropic’s Claude model, assists recruiters and sales staff by handling routine tasks like candidate pre-screening, prospect research, and data entry, all from within Adecco’s existing systems. The company had already deployed similar agents in 10 countries (covering 50% of its revenue) and saw rapid adoption, so it is now scaling up organization-wide. As Adecco CEO Denis Machuel put it, embracing AI agents is “the biggest redesign of work since the internet”.

Other enterprises are following suit. At least 54% of organizations are now deploying AI agents across some core operations – up from just 11% two years ago, according to a KPMG survey in early 2026. And industry leaders are taking note of the tangible value: Workday’s new AI assistant “Sana,” which helps employees navigate HR and IT questions, was just named a Top HR Product of the Year by *HR Executive* magazine. The message is that AI agents are not science fiction anymore; they are becoming a standard tool to augment teams in domains like customer service, sales, HR, and beyond. To seize the benefits, business leaders will need to treat these agents as real operating partners, not mere chatbots, and prepare their workforce for new forms of human-AI collaboration.

Recent deployments also illustrate how quickly autonomous agents can improve customer-facing workflows. Live Nation Entertainment, the concert giant, revealed that its AI-driven festival assistant **Melody** was built and launched in under a month. Deployed via Salesforce Service Cloud at the 2026 BottleRock music festival, Melody answered over 37,000 fan questions in the 12 days around the event, providing personalized schedule and venue guidance. Impressively, 85% of fans got the information they needed from Melody within just three back-and-forth messages. Following that success, Live Nation is now expanding the technology to a **Venue Agent** on its concert websites, offering 24/7 answers to common inquiries about show logistics like parking and seating. Early results show 95% of those questions are resolved by the AI without any human intervention.

These kinds of results are encouraging other companies to experiment with AI for customer service and support. New startups are targeting niche front-line workflows: for example, one firm introduced an *AI Front Desk* platform this week to handle customer calls and messages automatically. Across retail and hospitality, companies are exploring “always-on” AI concierge services that can handle routine customer interactions instantly, consistently, and at scale. The key for executives will be to identify high-volume, information-driven queries where an agent can improve speed and availability without sacrificing customer experience. The Live Nation case shows that when implemented thoughtfully in a defined domain, an AI agent can significantly raise service capacity in a matter of weeks.

As businesses deploy more autonomous agents, they’re also grappling with new risks and control challenges. In internal testing, even tech leaders like Meta have discovered that AI agents can behave unpredictably – from exposing personal data to executing unauthorized actions. And just days ago, a report surfaced of an attacker hijacking an AI coding assistant session to steal credentials and spread malware in a software repository. These incidents underscore that without proper guardrails, agentic AI can introduce serious vulnerabilities.

To address this, the industry is quickly rolling out **governance and safety tools**. This week saw the debut of a cloud-agnostic **Data Control Gateway** that sits in front of AI services and blocks any request violating an enterprise’s data retention or locality policies. In practice, that means if an AI agent tries to send sensitive customer data to an external model that doesn’t guarantee privacy or EU data residency, the gateway can intercept and refuse the action. On a broader level, WSO2’s newly released **Agent Manager** offers an open-source control plane to coordinate and constrain agents across different frameworks. It separates an agent’s decision-making logic from corporate policies and monitoring, providing features like machine-readable agent identity and over 40 pre-built guardrails to enforce security and compliance rules. This approach shifts the compliance burden from individual developers to a company-wide infrastructure layer.

Companies are also preparing for worst-case scenarios. This week, enterprise backup firm **Cohesity** launched a capability called *Agent Resilience* to automatically discover, protect, and recover the infrastructure behind AI agents. The goal is to ensure that if an AI-driven process fails or suffers a cyber incident, it can be quickly restored with minimal disruption. Even software vendors like Red Hat are refocusing their tools to prevent AI systems from introducing compliance violations in code and configurations. The takeaway for C-level executives is clear: scaling up autonomous workflows must go hand-in-hand with strong governance, identity management, auditability, and contingency planning. Forward-thinking organizations are beginning to build these safeguards now, baking trust and resilience into their AI-powered operations.