As enterprise organizations continue exploring practical uses for artificial intelligence, one thing is becoming increasingly clear: the conversation is shifting from general AI experimentation to actionable, agent-driven workflows.

That was the central theme of the recent Quest PeopleSoft Agentic AI Community Meetup. Rather than delivering a formal presentation, the session took the form of an open roundtable discussion where PeopleSoft professionals shared what they are actively testing, where they remain cautious, and what technical and governance challenges still need to be solved.

The result was a candid and practical conversation focused less on hype and more on implementation realities.

Why Agentic AI Matters for PeopleSoft Teams

One of the strongest themes throughout the discussion was the distinction between generative AI and agentic AI.

While many organizations are already using tools like ChatGPT or Microsoft Copilot for productivity tasks, such as summarizing meetings, drafting emails, or creating presentations, participants emphasized that agentic AI represents something much more operational.

Jeff Smith, from Raymond James, described agents as the part of AI that “gets things done” inside the business. Instead of simply generating content, agentic AI systems can execute workflows, make recommendations, trigger actions, and interact directly with enterprise systems.

For PeopleSoft organizations, that opens the door to practical automation opportunities including:

• Voucher and invoice processing
• Applicant screening and hiring workflows
• Natural language data retrieval
• Recommendation engines inside PeopleSoft pages
• Intelligent workflow routing
• Data synthesis and reporting

The conversation made it clear that most organizations are still early in their journey, but many are actively experimenting with proof-of-concept implementations.

Moving Beyond OCR: AI-Powered Voucher Processing

One of the most practical examples shared during the meetup involved financial processing automation.

Jeff explained how he is exploring AI-driven voucher ingestion and invoice processing using Oracle technologies. Instead of relying solely on traditional OCR approaches, newer AI tools can recognize shapes, layouts, and contextual patterns more accurately.

The goal is straightforward: reduce manual entry, improve accuracy, and eliminate repetitive administrative work.

This is where many organizations appear to be focusing their first enterprise AI investments. Rather than attempting fully autonomous systems immediately, they are prioritizing targeted operational improvements that deliver measurable efficiency gains.

That pragmatic approach resonated with many attendees.

The Biggest Challenge: Data Security and Governance

If there was one topic that dominated the session, it was data protection.

Participants repeatedly returned to the same core question: how can organizations safely use large language models while ensuring sensitive enterprise data never leaves approved boundaries?

This concern is especially significant for heavily regulated industries including finance, healthcare, and higher education.

Attendee David Limb, City of Hope Cancer Center, raised one of the session’s most important technical discussions around public versus private LLM usage.

He described a growing architectural pattern many organizations are considering:

1. Use powerful public LLMs to generate code, SQL, or logic.
2. Keep sensitive enterprise data processing inside private or enterprise-controlled environments.
3. Prevent confidential PeopleSoft data from being exposed to public AI models.

The discussion highlighted a major industry concern: while vendors may promise that enterprise AI endpoints do not retain or train on customer data, auditors still require proof and governance controls.

Several participants noted that this is creating hesitation across organizations, even when technical safeguards exist.

The group also discussed how some enterprises are exploring:

• Private in-house LLM deployments
• Dedicated enterprise AI endpoints
• Azure enterprise AI connections
• OCI-hosted environments
• Segregated AI processing layers
• Controlled API gateways

The consensus was that governance and auditability may ultimately shape AI adoption just as much as technical capability.

How Organizations Are Integrating AI with PeopleSoft

Another valuable part of the conversation centered on implementation patterns.

Attendees discussed several emerging approaches for integrating AI into existing PeopleSoft environments without directly exposing core systems.

API-Based Integration Layers

Many organizations are experimenting with REST APIs and middleware layers that sit between PeopleSoft and AI services.

This architecture allows organizations to:

• Protect API keys and credentials
• Apply governance controls
• Filter sensitive data
• Log AI interactions
• Route requests through approved security frameworks

Participants noted that this approach is currently one of the most common starting points for AI experimentation.

RAG and Data Lake Architectures

Arul TT, from North Dakota University System, shared how his team is testing retrieval-augmented generation (RAG) approaches.

In this model, PeopleSoft data is exposed through controlled database views, synchronized into a data lake, and then used as a trusted knowledge layer for AI chat experiences.

This allows organizations to build conversational AI tools without directly exposing entire transactional systems.

The approach reflects a broader enterprise trend toward curated AI-ready data environments.

Using PeopleCode and AI Together

One of the most technically interesting discussions came when Arul demonstrated a proof of concept involving PeopleSoft recruiting workflows.

In the example, applicant data and resumes were evaluated using AI recommendations, and the resulting decisions were written back into PeopleSoft workflows.

Notably, the integration was handled through the application layer using PeopleSoft’s Integration Broker and some limited PeopleCode rather than direct database manipulation.

That distinction mattered.

Participants emphasized that preserving PeopleSoft business logic, application integrity, and transactional controls remains critical even when AI is involved.

The recruiting example illustrated how agentic AI could eventually support:

• Candidate ranking
• Workflow status updates
• Intelligent recommendations
• Hiring disposition automation
• Decision support inside PeopleSoft

The implementation remains in proof-of-concept status, but attendees viewed it as one of the clearest examples of practical PeopleSoft AI integration discussed during the session.

AI Adoption Still Feels Fragmented

A recurring theme throughout the meetup was fragmentation.

Several participants acknowledged that AI initiatives are emerging independently across departments, with little centralized strategy.

Some teams are experimenting with Microsoft Copilot. Others are testing LangChain or LangGraph frameworks. Some are waiting for Oracle-delivered capabilities. Others are building custom integrations internally.

This decentralized experimentation phase feels familiar to many longtime IT professionals.

Attendees compared today’s AI landscape to earlier technology transitions involving cloud computing, distributed systems, and web application architectures. Organizations are still determining which platforms, governance models, and integration strategies will ultimately become standard.

The Road Ahead for PeopleSoft and Agentic AI

Despite the uncertainty, the overall tone of the meetup remained optimistic.

The participants were not debating whether AI would influence enterprise applications. Instead, they were discussing how to implement it responsibly, securely, and in ways that provide measurable operational value.

That distinction is important.

The session showed that the PeopleSoft community is moving beyond theoretical AI conversations and beginning to focus on:

• Secure enterprise integration
• Workflow automation
• AI-assisted decision making
• Governance and auditability
• Practical business use cases
• Long-term operational efficiency

As more organizations experiment with AI-enabled workflows inside PeopleSoft, conversations like this one will likely become increasingly valuable.

The biggest takeaway from the meetup may be that successful agentic AI adoption will depend less on flashy demos and more on thoughtful architecture, governance, and collaboration between business, IT, and security teams.

Want more?

Explore more content and resources to help you get the most from your Oracle investment:

• Visit the Quest Learn Library for blogs, how-to demos, and on-demand sessions.
• Connect with peers in one of our Quest PeopleSoft Community User Groups to swap stories, ask questions, and share tips with other users facing the same challenges.

Not a Quest member yet? Join today and tap into the ultimate Oracle customer network.