skill-056

name	skill-056
description	Personalized content and product recommendations based on user behavior and interactions within enterprise systems.

Enterprise Recommendation Engine Skill (Smart)

This skill offers dynamic content and product recommendations tailored to user interactions across enterprise platforms, enhancing user engagement and satisfaction.

It is designed for environments where user behavior data is abundant, including interactions with documents, chats, and meetings.

When to Invoke This Skill

Invoke when ANY of the following is true:

1. The user needs personalized recommendations based on their previous interactions.
2. The recommendations should be based on real-time data, adapting to user preferences continuously.
3. The task involves suggesting relevant documents, tools, or resources to enhance productivity.

Why Use This Skill?

Without this skill: you rely on static recommendation lists that don’t account for user behavior, leading to irrelevant suggestions.

With this skill: a subagent:

• analyzes user interaction data to find patterns
• provides tailored recommendations that evolve over time
• enhances user experience by connecting them with the most relevant enterprise resources

Typical engagement increase: 15–30%.

Invocation

Use this format:

Task(subagent_type="enterprise-recommendation-engine", prompt="""
User ID: <user_id>
Context: <user_context>

Output requirements:
- Return a list of recommended documents and tools.
- Provide reasoning or evidence for each recommendation.

Constraints:
- Ensure recommendations are relevant to the user’s recent activity.
- Avoid recommending items that the user has already interacted with.
""")

Core Procedure (Must Follow)

Step 0 — Analyze user behavior

• Extract:
  • user ID (e.g., “12345”)
  • recent interactions (documents, conversations, tools used)

If user ID is missing, infer cautiously from existing session data.

Step 1 — Gather contextual data

• Retrieve user context (role, department, recent projects) to refine recommendations.

Step 2 — Generate recommendations

• Utilize collaborative filtering and content-based techniques to produce a ranked list of recommendations based on gathered data.

Step 3 — Validate recommendations

• Ensure that recommendations are within the enterprise’s product scope and are not outdated.

Step 4 — Return results

• Provide the final list with justification snippets for each recommendation, helping the user understand the relevance.