How multi-agent AI is redefining enterprise software

image via TNW | The heart of techThe Next Web

September 9, 2026, 1:25 PM

• •The article frames multi-agent AI as a way to make enterprise software interpret information, coordinate decisions, and initiate tasks rather than just store data.
• •It says specialized agents can split work across functions such as retrieval, analysis, policy validation, and execution.
• •Swaroop Borukar argues that orchestration, observability, secure interfaces, and permissions are essential for reliability in enterprise deployments.
• •The piece stresses that high-risk actions should include safeguards such as audit trails, escalation paths, and human approval thresholds.
• •It highlights infrastructure challenges including chains of model calls, tool use, latency management, and compute cost at enterprise scale.
• •Examples include sales systems that proactively coordinate follow-ups and IT systems that investigate incidents and prepare fixes for approval.

The article argues that multi-agent AI could shift enterprise software from static systems of record toward systems that help coordinate and execute business workflows. It cites Workday product manager Swaroop Borukar, who says specialized agents with clear roles, permissions, and orchestration are better suited to complex enterprise environments than a single general-purpose assistant. The piece emphasizes that adoption depends on guardrails such as auditability, human oversight, secure data access, and mechanisms for handling errors and escalation. It also notes that latency, capacity, and compute costs will be major constraints as organizations run large numbers of agentic workflows across existing business systems.

Read original article