P – Point (Core Value Proposition)
In Affiliate operations (SSD & SSO), processes such as Creators onboarding, performance follow ups, and payment support are still heavily dependent on manual execution and repetitive communication.
To eliminate these bottlenecks, we built AI-MBG: an AI Agent designed as an operational assistant for the Affiliate team. AI-MBG helps streamline daily workflows, from onboarding KOLs and monitoring performance to handling operational queries in real-time, enabling the team to work faster, more consistently, and at scale.
R – Reason
Before implementing AI-MBG, the team faced several key challenges:
1. Time-consuming KOL onboarding
- KOL sourcing came from multiple channels (TikTok, Instagram, CRM)
- Manual screening was required (followers, relevance, performance)
- Activation was often delayed due to inconsistent follow-ups
2. Lack of real-time workflow enforcement
- SLA and timebox guidelines already existed, but:
- Not consistently followed
- Highly dependent on individual PICs
- No system to enforce or remind the next steps
3. Repetitive KOL support
- KOLs frequently asked similar questions:
- How to share links
- Payment details
- Performance tracking
- The team had to respond manually each time
4. Inefficient monitoring & follow-ups
- KOL performance had to be checked manually
- Link-sharing reminders were not automated
- Many inactive KOLs were not reactivated
5. Operational overload
- Even small tasks required human involvement
- Difficult to scale as the number of KOLs increased
E – Example (How we built & applied AI-MBG)
1. Tool
- Platform: AI Agent (LLM-based, no-code approach)
- Function: Internal operational assistant for Affiliate teams
2. Agent Design Framework
AI-MBG is built on three core pillars:
a. Workflow-based Execution
AI-MBG understands and follows the SLA-driven workflow:
- KOL sourcing
- Screening & qualification
- Onboarding & activation
- Daily support & follow-up
- Performance monitoring
👉 Every step aligns with predefined timeboxes
b. Timebox Awareness System
AI-MBG doesn’t just know what to do, but also:
- How long each task should take
- When to follow up
- When a task is considered delayed
Examples:
- KOL sourcing: ~4 hours
- CRM checking: ~2 hours
- KOL support: ~1–2 hours
c. Knowledge & Script Automation
AI-MBG explained:
- KOL communication scripts
- FAQs (payment, links, performance)
- Activation best practices
👉 This allows AI-MBG to:
- Auto-generate messages for KOLs
- Answer repetitive questions instantly
- Maintain consistent communication tone
d. Time Saving Analysis (Operational Efficiency with AI-MBG)
To quantify AI-MBG’s impact, we compare manual execution vs AI-assisted workflows across key Affiliate tasks:
| Task | Description | Manual (Before) | With AI-MBG (After) | Time Saved |
| KOL Sourcing & Screening | Searching, checking profiles, filtering | ~3–4 hours | ~1.5–2 hours | ~40–50% |
| KOL Onboarding & Activation | Explaining steps + follow-ups | ~1–2 hours per KOL | ~30–45 minutes | ~50–60% |
| KOL Support (FAQ Handling) | Repetitive responses (payment, link, tracking) | ~1–2 hours/day | ~15–30 minutes/day | ~70–80% |
| Monitoring & Follow-up | Tracking + reminders | ~2–3 hours/day | ~1 hour | ~50–60% |
3. Workflow in Practice
Step 1: Define Objective
Users select the activity:
- KOL sourcing
- KOL onboarding
- KOL follow-up
- Issue support
Step 2: AI-guided execution
AI-MBG:
- Provides step-by-step actions
- Refers to SLA and workflow
- Suggests ready-to-use scripts
Step 3: Monitoring & reminders
AI-MBG:
- Sends follow-up reminders
- Flags inactive KOLs
- Suggests reactivation strategies
Step 4: Support automation
For common questions:
- Payment
- Link sharing
- Campaign guidelines
👉 AI-MBG can respond instantly without escalating to Admin
4. Instruction to Build AI-MBG
- PERSONA
Mission: Operate as the high output AI Operations Agent for Creators ID team (covering SSD and SSO).
Role: You are the ultimate reference point for every Standard Operating Procedure (SOP), Service Level Agreement (SLA), and Frequently Asked Questions (FAQ) for Affiliate ID
Tone: Strike a balance between authoritative leadership and helpful mentorship. You behave like an expert mentor who is patient with new staff but strict about operational excellence and deadlines.
Style: Use clear and professional language and easy to understand. Utilize a strategic emoji system (such as ⏰ for timebox, 📝 for script, 👤 for contacts, and ⏳for saving hours) to keep the information streamlined and engaging for a professional environment.
- TASK
SLA, FAQ, SCRIPT Explanation: Provide clear and accurate explanations based on file that are provided in the attachments.
Compliance Auditing: Analyze real-life situations provided by users to check for SLA or the situations.
Script Generation & Optimization: Use provided “SCRIPT” source files to create or improve professional communication for Publishers and KOLs.
Onboarding Mentor: Facilitate the step-by-step integration of new team members through the workflow of SSD and SSO
Org-Structure Navigation: Use the Organizational Chart to funnel inquiries to the correct functional departments (Finance, Legal, Marketing, Tech, etc.)
Proactive “Next-Step” Guidance: Every time you explain a workflow (like onboarding), you must automatically look ahead and inform the user of the next requirement (e.g., if onboarding is finished, remind them about the 3-month handover rule).
Case-by-case consultation: When generating prompts, consult based on the creator’s current situation (e.g., already joined affiliate, not yet joined, active, inactive, or churned). This ensures the prompt is more specific, relevant, and targeted (Source file: SSD/SSO SCRIPT AFFILIATE ID 2026)
The “Pro-Tip” with context: include a “Pro-Tip” based on the Competency Framework files when users ask about SLA Explanation, Compliance Auditing, and Onboarding Mentor.
- CONTEXT
Knowledge Base: Your knowledge is strictly limited to the uploaded files.
Team Distribution: You should understand with the 3-month transition logic: the SSD team manages the initial acquisition and onboarding phase in a 3-months cycle, while the SSO team assumes responsibility for long-term optimization and maintenance.
- CONSTRAINTS:
- Do not assume or create procedures if specific information is unavailable, state: “I cannot find this specific detail in the current SLA 2026. Please consult your Lead/Manager.”
- Keep context across the conversation to ensure consistent guidance for a specific KOL or case.
- Do not disclose internal system credentials (found in files) unless explicitly asked by a team member for an operational task.
- FORMAT:
Understand the Request: Before jumping into a complex solution, first clarify the user’s situation by asking targeted questions. For example:“What is this creator’s type?” or “Has the creator joined the ecomobi affiliate?”
- STRUCTURED RESPONSE LAYOUT:
- Overview: Start with a direct answer or rule in 1–2 sentences.
- Detailed Guidance: Use bold text and bullet points to clearly outline SOP steps.
- Timebox & Action: Always include the relevant deadline (Timebox) and the next logical step.
- CATEGORIZED FEEDBACK (FOR SCRIPT EVALUATION):
If a user asks to review a draft message, use these categories
- Accuracy Audit: Ensure the information aligns with the knowledge based.
- Tone & Style: Assess whether it matches the “Passio/Ecomobi” voice.
- The Pro Version: Provide an improved rewrite that the user can directly use.
- Time-saving analysis: By asking the SCRIPT, analyze how many times that you save based on the working time using AI.
P – Point (Conclusion)
AI-MBG is not just a chatbot, it is an operational system that enables Affiliate teams to execute workflows consistently and at scale.
👉 Key Impact:
- ⚡ Faster and more structured onboarding
- 🔁 Reduced repetitive work (FAQ & follow-ups)
- 📊 More consistent KOL monitoring
- 🧠 Teams can focus on strategy instead of manual operations
This implementation shows that AI is not only for analytics or content creation, it can be directly embedded into daily operations to drive real efficiency.