Pondering Pega’s approach to agentic AI and workflow automation
During its Q1 2025 earnings call, Pega Founder and CEO Alan Trefler outlined a bold vision for harnessing agentic AI to transform enterprise workflows.
Pega is widely recognized for serving large, complex B2C and business-to-citizen organizations, particularly in regulated sectors like financial services, healthcare, insurance, and government. Its strength lies in orchestrating sophisticated customer-facing workflows across fragmented systems and applications.
Over its four decades, Pega has reinvented itself multiple times. Pega Blueprint, introduced in early 2024, marks the latest iteration, harnessing genAI for process design. It allows users to describe jobs-to-be-done or business processes in natural language and let the AI generate optimized workflows, data models, and user experiences, leveraging best practices.
Pega Blueprint can ingest Business Process Model and Notation (.bpmn), Data Definition Language (.ddl), OpenAPI YAML (.yaml) files, and even system logs—using process mining to reverse-engineer how work actually gets done. This foundation has enabled Pega to build a rich repository of workflows, capturing the collective knowledge and experience of its clients and partners.
By late 2024, enterprises and system integrators had already created over 60,000 blueprints—early evidence of strong interest in the approach.
Trefler highlighted two critical shortcomings in today’s agentic AI solutions, especially for the complex, regulated enterprises Pega serves.
First, he challenged the overreliance on prompts and free-text instructions, arguing they demand specialized prompt engineers and often ignore the wealth of industry best practices already available.
Second, he pointed to the lack of process predictability and governance—non-negotiables in highly regulated environments.
Trefler instead advocates for a hybrid model—anchoring processes in Blueprint-generated workflows, with AI agents executing defined steps to deliver both automation and control.
The hype around agentic AI has led many providers to over-index on autonomy. In contrast, the hybrid approach is gaining momentum, and for good reason. It provides a safer path to adopting AI agents, ensuring both control and predictability.