Pegasystems Inc. (NASDAQ: PEGA) used its flagship PegaWorld conference in Las Vegas to unveil a major upgrade to its core automation platform. In the upcoming Pega Infinity 26 release, the company will embed native support for the open Model Context Protocol (MCP) standard, a framework that lets external, third‑party AI agents discover, invoke, and complete individual steps of Pega‑driven business processes. By exposing workflow logic through an MCP‑compliant server, Pega aims to give enterprises a reliable “plug‑and‑play” bridge to agents built on leading large‑language‑model (LLM) services—Anthropic Claude, Google Gemini, OpenAI, AWS AgentCore, and others—without sacrificing the governance, auditability, and cost controls that mission‑critical applications demand. The announcement positions Pega’s new capabilities as a direct response to growing market anxiety around AI‑driven automation, where hallucinations, compliance gaps, and runaway token expenses have stalled many projects. With MCP, Pega promises predictable outcomes, predictable costs, and a unified orchestration layer that can scale across the “millions of Pega‑powered workflows” already in production worldwide.
New MCP‑Enabled Capabilities Unveiled at PegaWorld
During a keynote on Monday, June 8, Pega’s chief product officer Kerim Akgonul walked the audience through three concrete features that will ship as part of Infinity 26:
- MCP server capabilities – A built‑in server that publishes Pega workflows as MCP endpoints. This eliminates the need for custom adapters or point‑to‑point integrations, allowing any authorized AI agent to query the catalog of available processes, retrieve the required input schema, and receive step‑by‑step execution instructions.
- Agentic assignment agent – An AI‑driven outreach component that automatically contacts the appropriate employee or customer when a workflow step requires additional data, a decision, or an approval. The agent can reach out via email, chat, or telephony, embedding the request directly into the user’s preferred communication channel and feeding the response back into the process without manual hand‑off.
- Document agent – A specialized agent that ingests complex documents (PDFs, images, scanned forms), parses them, categorizes content, scores relevance, and triages items for downstream handling. A new chat interface lets users ask natural‑language questions about the document’s contents and receive instant insights, turning static files into interactive knowledge sources.
All three capabilities are being demonstrated live on the Las Vegas conference floor and will be available to existing Pega customers as part of the standard Infinity 26 update, slated for general availability in Q3. The open‑source nature of MCP means that any future AI service that adopts the protocol can immediately interoperate with Pega’s orchestration engine, future‑proofing the investment.
How MCP Extends Pega’s Business Orchestration Platform
Pega’s Business Orchestration and Automation Technology (BOAT) platform already provides a robust, step‑wise execution engine that enforces business rules, audit trails, and service‑level guarantees. By layering MCP on top of BOAT, Pega transforms each workflow step into a hand‑off point for an external AI agent. The agent receives a concise, machine‑readable description of the task, performs the work (for example, generating a response, extracting data, or making a recommendation), and returns the result to the BOAT engine, which then records the interaction, enforces policy constraints, and moves the case forward.
This approach contrasts sharply with “other approaches” that require agents to re‑reason the entire process at every iteration—a practice that inflates token consumption and introduces nondeterministic outcomes. Pega’s architecture keeps the process logic centralized, while the AI agent focuses solely on the micro‑task it is best suited for. The company also highlighted its Pega Blueprint AI™ workflow design agent, which automatically scans existing processes to pinpoint steps where an AI‑driven micro‑service could add efficiency, and its orchestration layer that audits every agent interaction across the lifecycle, preserving a complete, tamper‑evident log for compliance teams.
By exposing a standardized MCP interface, Pega opens the “growing universe of agents” to millions of existing workflows—from dispute resolution and claims processing to customer onboarding and field service. Enterprises can now reuse a single, vetted AI agent across multiple processes, dramatically reducing the time and expense of building bespoke adapters for each use case.
Relevance for Enterprise Buyers
For CIOs, CTOs, and other technology leaders, the announcement tackles two of the most pressing pain points in today’s agentic AI deployments: cost volatility and regulatory risk. Gartner estimates that over 40 % of agentic AI projects will be canceled by the end of 2027 because of escalating token costs, unclear business value, and insufficient controls. Pega’s MCP‑enabled orchestration directly addresses these concerns by:
- Predictable cost management – Because the BOAT engine governs how many tokens each step may consume and can enforce caps, organizations avoid surprise bills that have plagued many LLM‑centric pilots.
- Governance and auditability – Every agent call is logged, versioned, and tied to the underlying business rule set, giving compliance officers the evidence they need to satisfy internal policies and external regulations.
The pre‑built agentic assignment and document agents illustrate how MCP can be applied to common enterprise scenarios. In a claims‑processing workflow, for instance, the assignment agent can automatically solicit missing documentation from a claimant, while the document agent extracts key fields from uploaded PDFs and surfaces them to the adjuster in real time. By eliminating manual hand‑offs and custom code, firms can shorten time‑to‑value, reduce development overhead, and lower the risk of integration errors that often derail AI initiatives.
Key Takeaways
- Pega Infinity 26, expected in Q3, will include MCP server capabilities that let authorized third‑party AI agents execute Pega workflows.
- New pre‑built agents (agentic assignment and document agents) are designed to automate outreach and document processing within mission‑critical processes.
- Gartner predicts more than 40 % of agentic AI projects will be canceled by 2027; Pega positions its MCP support as a way to mitigate cost and compliance risks.
TechInsyte's Take
Pega’s MCP integration offers a concrete path for enterprises to harness external AI agents without sacrificing the governance that mission‑critical applications require. The approach hinges on the upcoming Infinity 26 release, so buyers should monitor the actual performance and audit capabilities once the product is in production. Organizations evaluating AI agents should assess whether MCP‑enabled orchestration aligns with their existing process architecture and compliance frameworks.
Source: Businesswire
