Cognizant has assigned more than 18,000 engineers to a newly established Gemini Frontier Capability Group as part of an expanded partnership with Google Cloud, the company announced Oct. 19. The specialists will assist enterprise customers in bringing AI agents into production environments. The expansion raises questions for smaller systems integrators and managed service providers about how to compete on industry knowledge and secure ongoing support work once deployment is complete.
The engineering priorities include modernizing legacy applications, enhancing software development through AI, automating intricate business processes, and bolstering cybersecurity defenses as organizations introduce autonomous agents. Planned enterprise offerings cover Cognizant's Catalog Support Agent for retail clients and a Healthcare Agentic Marketplace coordinated through its Neuro San platform, which uses an open-source framework enabling specialized AI agents to delegate tasks, communicate with one another, and connect with external tools. Manufacturing customers represent another target market, with proposed multi-agent order-to-cash workflows where coordinated agents would manage activities across business systems from receiving customer orders through payment collection. The group intends to use Gemini to support enterprise AI agents and deploy a standardized architecture based on Model Context Protocol for connecting them with business systems, incorporating Google Cloud's Model Armor to screen incoming prompts and generated responses for risks such as prompt injection and sensitive data exposure.
According to the report, Cognizant will build on its existing Frontier Certified Engineer delivery model, placing specialists alongside Google Cloud teams inside customer environments where engineers examine existing workflows, evaluate agent behavior, and help implement applications for production use. The approach offers a way to adapt AI systems to individual customer processes before deployment. More than 55,000 employees completed Build with Gemini, developing agents and applications and earning Gemini skills credentials, the company said. In July, Cognizant said it aimed to deploy Gemini Enterprise to 100,000 associates in 2026, with plans to expand to 200,000.
Cognizant's dedicated workforce could intensify competition for major enterprise AI contracts, pushing smaller systems integrators to concentrate on industries where they have proven AI integration experience and support their bids with customer references and measurable results. Competing with a global consultancy on engineering capacity alone is unlikely to be their strongest approach. Managed service providers with existing enterprise accounts may find opportunities for managed AI services after deployment, including monitoring agent performance and maintaining software connections, though providers should price recurring work separately from implementation and define escalation procedures when automated workflows fail. Security-focused providers may address ongoing AI agent security and governance as autonomous systems gain access to sensitive business information, verifying permissions, retaining activity logs, and establishing which actions require human approval.
For smaller partners, the path forward depends on securing the continuing role after an AI system goes live, which the report suggests may be the difference between a one-time implementation project and a recurring service relationship. Customer contracts could ultimately determine which providers retain responsibility once deployment is complete. Specialist consultancies could explore implementation or subcontracting relationships with larger integrators, though the announcement doesn't specify a reseller program or a route for smaller partners to participate, and partners considering these arrangements should negotiate customer access and technical handoffs before committing to projects. The threshold between deployment engineering and ongoing operations will likely define which firms capture value as enterprise AI scales beyond pilot projects.