How Lemon Learning built a smart learning assistant with Rakam AI

Introduction

Lemon Learning helps businesses make training more accessible by integrating learning directly into the tools employees use. But their ambition went further: to personalize that experience in real time.

To do that, they needed a recommendation engine powered by AI, one that could understand user behavior, suggest relevant training content, and do it all within the tight boundaries of client expectations and regulatory compliance.

With Rakam, Lemon Learning developed an AI system that recommends content contextually, responsibly, and transparently, turning a useful tool into a truly intelligent assistant.

The Challenge

Lemon Learning’s platform already offered contextual learning, embedded inside applications like Salesforce. But as their client base expanded, so did the complexity of user needs.

They wanted to guide users toward the most relevant content without making the system feel intrusive or random. The AI couldn’t simply "guess", it needed to recommend with purpose, while respecting sensitive data environments.

They faced three key challenges:

• Content had to match real-time user behavior, not just static rules.

• Clients needed transparency and traceability to accept AI-based suggestions.

• The AI system had to be auditable and bias-aware, especially for enterprise clients in regulated industries.

"It wasn’t enough to deliver recommendations. We had to show why each one made sense, and make clients trust the process."

The Vision

Lemon Learning wanted a recommendation assistant that felt intelligent but also controlled. The assistant would need to be fully embedded into Salesforce, while being trusted by large enterprise customers.

The goals were clear:

• Make content discovery effortless for users

• Align every suggestion with user context and behavior

• Respect the traceability, security, and audit needs of enterprise systems

• Ensure the model could be explained, and improved over time

This meant rethinking how AI recommendations work inside training systems, not as a black box, but as an integrated, transparent layer.

The Solution

Rakam worked closely with Lemon Learning to design a pedagogical AI assistant that learns from user interactions and suggests training content directly in their working environment.

1. A Personalized Learning Engine

The system was designed to match training content with what users are doing in real time, clicks, modules, roles, and tool usage, all while respecting data boundaries.

• Contextual triggers prompt recommendations at just the right moment

• Profiles adapt over time based on behavior

• Suggestions align with existing content and objectives, never outside material

"The goal was to reduce noise, not add more content. Every suggestion had to matter."

2. Built-In Compliance and Explainability

In high-sensitivity environments, explainability is non-negotiable. Rakam helped Lemon Learning structure an AI engine that clients can audit, trace, and understand.

• The logic behind each recommendation is logged and accessible

• Bias detection modules run checks during training

• A full audit trail shows how and why decisions are made

"We couldn’t ask clients to trust the system, we had to show them why it deserved trust."

3. Fully Integrated with Salesforce

The assistant lives within Salesforce as a native component, making it easier for companies to deploy and manage without a steep learning curve.

• No need for external tools or dashboards

• Seamless interface with Salesforce’s permission and role systems

• Embedded directly into daily workflows

This gave clients a sense of continuity while introducing advanced AI capabilities without disrupting the user experience.

The Outcome

The new recommendation engine became more than a product feature, it became a key part of Lemon Learning’s value proposition.

• Enterprise clients now benefit from tailored learning journeys without manual curation

• Product teams can fine-tune AI suggestions without sacrificing transparency

• Compliance teams have clear visibility into how decisions are made

"This wasn’t just a new capability, it was a new layer of trust between us and our clients."

Today, the system not only drives content engagement but helps companies deliver just-in-time training in ways that feel both smart and safe.