Enterprise AI agents move from chat to process execution
Enterprise AI agents are software systems that combine large language models, workflow automation, and integrations with business applications to interpret unstructured requests, trigger governed processes, and complete specific tasks end to end inside existing enterprise environments. Rather than acting as generic chatbots, these agents are being embedded into real workflows across manufacturing and business operations to cut manual work, speed response times, and reduce data re-entry. The latest examples come from commercetools and Zenphi, which both focus on workflow automation but in different contexts. commercetools’ B2B Intake Agent targets complex order intake in manufacturing-style sales cycles, while Zenphi AI Studio supplies Google Workspace agents that live inside Google Chat and run defined workflows. Together they show how AI intake agents and Google Workspace agents are evolving from passive assistants into governed business process automation tools wired into CRM, service, HR, and finance systems.
commercetools and Mirion turn unstructured intake into B2B quotes
commercetools’ new B2B Intake Agent, co-developed with nuclear measurement and detection systems manufacturer Mirion Technologies, is built to translate messy customer requests into structured commerce actions. The agent reads emails, PDFs, spreadsheets, and other offline-style files, then turns them into actionable quotes and cart data instead of forcing sales and service teams to retype orders across systems. According to commercetools chief product officer Shiri Mosenzon-Erez, “We built this capability to apply AI where it creates immediate value inside real workflows.” The agent connects to CRM and service platforms such as Zendesk via APIs, so output can feed directly into existing sales and support processes. Mirion’s involvement grounded the design in real manufacturing needs, where complex products, strict requirements, and accuracy-sensitive workflows dominate. For Mirion, the promise is faster order intake, better responsiveness, and more time for teams to solve customer problems instead of decoding order formats.
Zenphi AI Studio brings governed Google Workspace agents to Chat
While commercetools focuses on B2B commerce, Zenphi AI Studio addresses day-to-day internal operations by turning Google Chat into a front end for workflow automation. Zenphi describes AI Studio as a no-code environment where teams build enterprise AI agents that submit leave requests, file reimbursements, start onboarding, report incidents, or assign tasks with reminders. The agents do more than answer questions: they execute Zenphi workflows step by step, with ownership, approvals, and exception paths defined in advance. CEO Vahid Taslimi highlights that AI Studio runs on the same Zenphi platform already executing over 1.5 million AI-powered tasks a month in production, with every action scoped and logged. Administrators define which workflows each agent can trigger, which data sources it may access, and which roles it can serve, making these Google Workspace agents suitable for HR, finance, legal, IT, and other sensitive business process automation scenarios.
From individual productivity tools to governed business process automation
Both commercetools and Zenphi show how enterprise AI agents are moving beyond personal productivity tools toward full business process automation. Most earlier AI features in office suites focused on drafting emails or summarising documents. By contrast, the B2B Intake Agent and Zenphi AI Studio bind AI tightly to structured workflows, audit trails, and permission models. commercetools targets AI intake agents that turn unstructured orders into CRM and commerce records without breaking manufacturer-specific rules. Zenphi targets Google Workspace agents that live in Google Chat but respect workflow logic and access controls configured on its platform. Together they signal a shift: AI interactions now routinely launch governed processes such as quoting, approvals, onboarding, or incident reporting. For enterprises that must balance speed with compliance, this approach makes AI part of production systems rather than sidecar tools, and it lays groundwork for broader agentic architectures in operations and manufacturing.
