Overview

Learning recommendations are an AI-driven, personalized system within Talent Management (TM), designed to help employees identify the most valuable learning opportunities for their unique career paths. This system combines individual employee data with organizational context to surface a ranked list of courses that effectively close skill gaps and advance career goals.

The main goal is to provide a complete platform for an employee's career development needs. Learning is linked to specific career results, like improving in a current job or getting ready for a new one.

Tailored learning pathways

The learning recommendation engine operates through a sophisticated, multi-phase process that leverages extensive data and intelligent algorithms.

• Data-driven personalization:
  
• Employee data: Recommendations begin with understanding the employee's current role, prior job experience, existing skills, and skill proficiency. Critically, it incorporates career interests (desired future roles) and skill interests (explicitly marked learning goals) as key influencing factors.
  Skill proficiency can be derived from manager ratings, self-ratings, or seniority as a fallback.
• Spotlight Courses: Administrators can tag specific courses to be ‘spotlighted’ for certain employees or groups, making them prominent in feeds.
• Course Eligibility: Courses can be limited to specific employees based on matching eligibility tags between the course and the employee's profile.
• Course metadata: The system ingests and enriches detailed course information, including the skills offered, difficulty level, and expected proficiency upon completion. Eightfold employs AI to infer additional skills from course titles and descriptions and to determine difficulty levels (For example, "Introductory," "Intermediate," "Advanced") when not explicitly provided.
• Organizational context: Recommendations consider attributes like business unit, job family, seniority, and job codes to ensure learning aligns with real-world audiences and organizational priorities.
• Relevance types: The platform supports different ‘relevance types’ that guide the recommendation algorithm, emphasizing what matters most for a given request:
  

• SKILL_GOALS: Prioritizes courses that teach skills the user has explicitly marked as learning goals.
• ROLE_GOALS: Uses the skills and context of roles the user aspires to, aligning courses to that target role.
• PROFILE_SKILLS: Concentrates on skills already present in the user's profile to deepen expertise.
• SKILL_GAPS: Focuses on skills where the user's current proficiency is below what their role or ambition requires.
• Intelligent matching: Eightfold follows a meticulous process to generate and refine recommendations.
  

• Catalog enrichment: The course catalog is continuously enriched using historical attendance data (to align courses with their real-world audience) and Large Language Model (LLM) inference from course descriptions to add rich, uniform metadata. This ensures recommendations stay relevant without manual curation.
• Matching algorithm: The platform transforms the ‘Recommendation Target’ into a search request, using skill terms with boost multipliers and context terms to nudge ranking. It filters out completed courses, aligns course difficulty with the learner's proficiency (for example, a course for a skill with 0.4 proficiency will have a difficulty of 0.4–0.8), and ensures each course teaches at least one highlighted skill. The algorithm also accounts for skill gaps, course relevance, and assigns weight to different skill types (Core Skills > Skill Interests > Other Skills).
• Filtering: After an initial search, a post-processing layer applies quality gates, including similarity scoring (skill and title similarity) and thresholds to discard weak matches. This acts as a ‘quality firewall’ to ensure recommendations are tailored and credible.
• Updates: Recommendations are continually adjusted as employees update their profiles, new courses become available, or organizational needs evolve. The system also adapts difficulty as a user's proficiency grows.

Key benefits

• Targeted upskilling: Employees focus on learning that directly closes personal skill gaps or advances career aspirations, ensuring learning investments are impactful.
• Business alignment: Recommendations reflect organizational priorities, such as specific business units or job families, ensuring that learning supports strategic talent needs.
• Time savings: Intelligent filtering removes irrelevant or duplicate courses, enabling employees to quickly find and engage with valuable content.
• Employee empowerment: Transparent explanations for recommendations and controls like the ability to dismiss courses give employees ownership over their growth journey, fostering trust and sustained engagement.
• Continuous improvement: Built-in feedback loops and the ability to adapt to new data (For example - attendance records, updated AI models) keep recommendations fresh and valuable over time.

Integration with learning platforms

The system supports various integration approaches with Learning Experience Platforms (LXPs) and Learning Management Systems (LMSs) to seamlessly deliver recommendations.

• File-based ingestion: This is the most common approach, where course catalogs and completion data are ingested via files. In this model, Eightfold typically serves as the primary discovery platform for learning, driving recommendations with its own algorithm and assessment data. It supports displaying upskilling pathways.
• App platform integration: In this less common approach, the LXP is the primary platform, and Eightfold requests relevant courses from the LXP on page load. Recommendations are driven by the learning vendor rather than Eightfold's algorithm. This method generally doesn't support displaying pathways.