PeopleFluent, as part of the Learning Technologies Group, is committed to embedding AI features into PeopleFluent Learning (PFL) to provide customers with tools to improve efficiency and productivity.  As the AI tools have improved, it also requires a deliberate approach to ensure that deliverability and security are top priorities.

After the original plan for 26.10, there were security breaches, such as what occurred with the Canvas Learning Management System, as well as recent concerns over agentic AI tools from Open AI, Anthropic, and Meta showing potential security risks.  These prompted a review of the PFL security around embedding AI.

As a result, we expanded the requirements, resulting in a two-phase approach across the 26.10 and 27.02 releases. 

Phase I - Core AI Infrastructure & Targeted Features

The first phase, which will be released in version 26.10, is concentrating on the core infrastructure.  This is broken into three feature areas.

AI Setup

Within the application there will be the functionality to set up PFL for AI functionality.  This includes:

• A new system configuration to enable AI features.
• New role-level access permissions to allow access to the AI features.
• The ability for customers to set up their AI Models within the application as part of a ‘bring your own AI’ model.

AI Orchestrator

The AI Orchestrator serves as the central control layer that manages incoming requests from the PFL AI-supported features, applies authentication, enforces user permissions, and routes requests to the appropriate AI services. It ensures consistent governance, scalability, and auditability across all AI interactions. This approach reduces risk, centralizes control, and enables flexible integration with multiple AI providers. 

Spring AI

Spring AI provides the foundational framework for integrating large language models (LLMs) into the LMS, standardizing how prompts are constructed, executed, and managed across providers. It abstracts provider-specific complexity, allowing the platform to support multiple LLMs with minimal engineering effort. This results in faster development cycles, easier vendor flexibility, and reduced long-term technical debt.

This infrastructure will support some targeted features that administrators will be able to use.  This list is not guaranteed, as the work on 26.10 continues.  However, the following features are among the first:

• Language Bundle Translations – This will provide AI assistance when creating a language bundle for a course, job profile, competency, etc.  AI will translate the appropriate fields from the source language into the desired language for the bundle.
• Course Metadata – Allow AI to generate course metadata like a better description, or a catalog image.
• Custom Language String Translation – This will allow administrators to use AI to help translate custom language strings for the UI as part of the Search/Customize Language Strings functionality.

These features, while useful, are intended to illustrate the use of AI within PFL.  They will prove out the infrastructure is in place to support further development of AI features in Phase 2 and beyond.

Phase 2 - Advanced Tools and Socrates

For 27.02 there will be more functionality added within the infrastructure and more features added for users.

There are two additional areas added to the infrastructure.  These are not needed to support the specific functionality in Phase 1, but are needed to assist with more open-ended functionality available with Socrates. These are:

• Spring AI Advisors – Advisors are modular components within Spring AI that intercept and enhance requests and responses, enabling capabilities such as security filtering, prompt augmentation, logging, and context injection. They ensure that every AI interaction is aligned with business rules, user entitlements, and compliance requirements. This layer improves response quality, enforces governance, and provides extensibility without modifying core logic.
• Spring AI Tools – Tools are structured integrations that allow the AI to securely access and interact with LMS data, APIs, and external systems to perform actions or retrieve real-time information. They extend the AI beyond conversational responses into actionable workflows and data-driven insights. This enables more accurate, context-aware outputs while maintaining strict control over data access and system behavior.

Given only so many tools can be created in a given release, AI functionality won’t be able to do anything users request. This is part of what the AI Orchestrator will limit. More tools will be added within future releases based on the themes of those releases.

Socrates (AI Assistant)

Socrates is a conversational, chat-style AI assistant embedded within the LMS that enables users to ask questions, explore content, and receive contextual guidance in real time.  Examples of possible use cases customers have suggested could include:

• A built-in AI admin assistant inside PFL that can help reduce time and effort on repetitive admin tasks.
• Leveraging AI to generate quizzes or assessments based on prompts, learning objectives, or questionnaires would save time and add consistency.
• A chatbot like functionality that can do things in the LMS, like planning a new classroom course and all that goes with it, enrolling users or setting up auto-enroll settings, moving users from one session to another and asking questionson data in the platform to generate and overview or report.
• Using AI to make training processes more effective—and not tied to the opinions of any single individual.

We will also explore other targeted features such as:

• AI Assisted Search - Use the AI tool for natural language processing so users can search using conversational queries such as: “Show beginner React courses under 2 hours” rather than pure keyword searches.
• AI Search Summaries - Similar to modern search engines, we will add AI-generated search summaries as part of the search result presentation.
• Rustici Generator - Integrating with Rustici Generator will retrieve content from within courses, providing the ability to enhance the search functionality as well as the Course Metadata feature.

Future Releases

With these changes, the future release themes will be shifted from the original plan:

• 27.06 - June 2027 - Exam Enhancements
• 27.10 - October 2027 - Course Administration Enhancements
• 28.02 - February 2028 - Compliance Enhancements

As we develop features for these releases, we will explore ways that AI will support or improve the features. 

In 2027, we are planning to upgrade the version of Java for the application from Java 17 to Java 21. The primary driver of this change is for improved performance, but also to extend the functionality into the 2030s before Java will need another upgrade.

Visit the Product Roadmap for more information about the planned AI features.