Genpact has launched Genpact Onyx, an enterprise platform designed to build, deploy and govern AI agents inside complex business operations. The system uses the company’s Process Intelligence Core, a collection of operational knowledge developed through decades of managing enterprise workflows, to give agents context about how work is actually performed. Unveiled at Genpact’s Agentic Operations Day in New York on November 12 and announced as available on November 16, 2026, Onyx is intended to move AI agents beyond isolated tasks and into governed, accountable business processes.
Genpact wants process knowledge to be the foundation for enterprise agents
The launch reflects a growing challenge for organizations moving AI agents from pilots into production: agents need more than access to a large language model and a set of instructions. They also need to understand the business rules, exceptions, approvals and handoffs that shape real operational work.
Genpact is positioning Onyx around that requirement. Its Process Intelligence Core is intended to encode the company’s accumulated experience across finance, risk, supply chain, banking, insurance and procurement. That knowledge includes predefined workflows and agents, but also the less predictable elements of enterprise processes: unusual cases, exceptions and decisions that typically depend on experienced employees.
The distinction matters because business operations rarely follow a perfectly linear sequence. An invoice may not match a purchase order. A payment may require additional checks. A supplier record may be incomplete. In these situations, an agent needs to know when it can proceed, when it should ask for more information and when a human decision is necessary.
Genpact says Onyx can be embedded into existing enterprise technology environments, connect with customers’ data and AI foundations, and support a choice of models through its partner ecosystem. The company is therefore pitching the platform as an operating layer for agents rather than a replacement for an organization’s existing applications or AI infrastructure.
A learning loop and graduated autonomy
One of Onyx’s central features is a feedback mechanism that sends reviewed exceptions and expert decisions back into the Process Intelligence Core. Genpact’s stated aim is to make the platform more effective as it encounters additional cases, allowing future workflows to be handled more quickly and with less manual intervention.
The company also describes an approach it calls “earned autonomy.” Agents initially recommend actions for human experts to review and approve. As they demonstrate reliability, they can take responsibility for a greater share of the decision-making process.
This model is designed to avoid the assumption that every agent should be fully autonomous from the beginning. In finance and procurement, for example, a system might initially identify invoice discrepancies and prepare a proposed resolution. A person could approve the action until the organization has enough evidence to delegate more decisions within defined limits.
Governance is another part of the platform’s architecture. Genpact says agent actions are controlled, auditable and traceable to the reasoning behind a decision. The company also cites built-in guardrails, responsible AI controls, security and data isolation. Those features are important for enterprises that need to establish which agent performed an action, what information it used and why a workflow proceeded.
Accounts payable is the first packaged application
Genpact Onyx will underpin the company’s own agentic products, beginning with the Genpact Accounts Payable Suite, as well as custom solutions developed for customers.
Accounts payable is a practical starting point for this strategy. The function involves recurring, document-heavy tasks such as invoice capture, purchase-order matching, exception handling, approvals and payment preparation. These processes offer opportunities for automation, but they also involve financial controls and exceptions that make unrestricted autonomy risky.
A process-aware agent could handle routine cases while routing mismatches or unusual transactions to a finance professional. The potential benefit is not simply faster invoice processing; it is reducing manual effort without removing oversight from decisions that affect cash and supplier relationships.
Genpact has also been expanding its agentic finance portfolio. In October 2026, it announced an Agentic Accounts Receivable Suite focused on cash flow and working capital, providing a broader context for Onyx as a shared foundation for finance-related automation.
The governance problem is becoming harder to ignore
The timing is significant as enterprises scale the number of agents operating across departments. Gartner forecast in April 2026 that an average Fortune 500 company could have more than 150,000 agents in use by 2028, compared with fewer than 15 in 2025. The analyst firm also reported that only 13% of organizations believed they had the right AI agent governance in place. gartner.com
That gap raises questions about agent identity, permissions, access to enterprise data, monitoring and accountability. Gartner has separately warned that weak AI governance could contribute to business-critical service disruptions for 40% of infrastructure and operations organizations scaling AI by 2028.
Onyx’s emphasis on process context and traceability puts it within this broader market shift. It also places Genpact in competition with enterprise automation and agent platforms from companies such as Microsoft, Salesforce, ServiceNow, UiPath and IBM. These vendors approach the market from different starting points, including productivity software, CRM, IT service management, robotic process automation and enterprise AI infrastructure.
Genpact’s argument is that operational knowledge is a differentiator: companies need agents that understand how work gets done, not simply agents that can perform isolated actions.
From platform launch to measurable outcomes
The announcement establishes Onyx as the foundation for Genpact’s agentic solutions, but it does not disclose customer adoption figures, independent performance benchmarks or quantified productivity improvements attributable to the platform. Client representatives from Advantage Solutions, Wesco, USA Today and Ryan Specialty spoke at the launch event about their use of Genpact’s agentic solutions, but the announcement does not provide detailed case-study metrics. Nov 16, 2026
That leaves an important test for the platform: whether its process intelligence and governance capabilities translate into measurable improvements in production environments. Enterprises will need evidence around exception resolution, accuracy, operating costs, processing times and the frequency of human intervention.
For Genpact, Onyx represents a move to turn its business-process expertise into a reusable technology foundation. The longer-term opportunity is to combine agents’ execution capacity with human judgment and operational accountability. Whether that model scales will depend on how consistently the platform can deliver results across different clients, systems and highly variable business processes.
Market Landscape
Enterprise agent platforms are shifting from simple task automation toward systems that can coordinate multiple agents, access business data, follow organizational policies and operate within existing workflows.
Genpact Onyx enters a market that includes several overlapping categories:
- Enterprise AI platforms: Microsoft Copilot Studio, IBM watsonx and Salesforce Agentforce.
- Workflow and process automation: UiPath, ServiceNow and automation platforms focused on enterprise execution.
- Process intelligence: Tools that analyze workflows, identify bottlenecks and provide operational context.
- Agent governance: Systems for monitoring agent behavior, managing permissions, auditing actions and controlling autonomy.
Genpact’s positioning is distinct in that it combines a technology platform with process knowledge gained from operating enterprise business functions. Its competitive advantage will depend on how effectively that knowledge can be reused across customers without sacrificing customization, security or transparency.
Top Insights
- Genpact Onyx is designed to build, run and govern enterprise AI agents using operational knowledge codified through decades of business-process experience.
- Its Process Intelligence Core captures workflows, exceptions and human decisions across finance, risk, supply chain, banking, insurance and procurement.
- The platform introduces earned autonomy, allowing agents to progress from recommending actions to taking greater responsibility as reliability is established.
- The Genpact Accounts Payable Suite is the first named proprietary solution built on the platform.
- Genpact has not yet disclosed independent performance benchmarks or quantified customer outcomes specifically attributable to Onyx.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
