From manual support → AI-driven campaign advisory
Wanna try our AM Assistant? Please tap HERE
P – Point
In today’s fast-evolving e-commerce landscape, the role of an Account Manager (AM) is no longer limited to execution — it requires fast decision-making, cross-market understanding, and consistent advisory across multiple platforms.
However, as operations scale across markets and teams, relying on manual knowledge sharing and individual experience is no longer efficient or sustainable.
At Ecomobi, where performance and speed are critical, this gap becomes even more evident.
👉 That’s why we started building the AI-powered AM Assistant — a solution designed to transform how campaign knowledge is accessed, standardized, and applied across teams.
R – Reason
Before implementing the Agent, our Internal Team (SSO, Pub Ads, Tech, OP) faced several challenges:
- Inconsistent answers → different AMs provide different interpretations of the same campaign
- Time-consuming support → repeated explanations across teams and markets
- High dependency on individual experience → difficult to standardize and scale knowledge
Additionally, from an operational perspective:
- Within the team, knowledge about Shopee, Lazada, and Shopee MCN is scattered across markets (VN, ID, TH, PH, MY, SG)
- Many functions (OP, BD, Creator team) require quick access to campaign information in daily work
- These teams are not platform experts, which leads to:
- Time wasted searching and validating information
- Misunderstanding of campaign mechanics or policies
- Inconsistent communication with partners (brands, creators, platforms)
💡 Key insight:
Not everyone needs to be an AM expert — but every function needs fast, accurate access to standardized campaign information.
👉 Therefore, we needed a solution that:
- Centralizes and standardizes campaign knowledge across markets
- Ensures accuracy while reducing dependency on individual AMs
- Enables teams to retrieve reliable information quickly and consistently
E – Example (How we built & applied it)
1. Tool
- Platform: Notebook LM
- Format: Custom-built AI Agent
2. Agent Design Framework
We defined the Agent as:
“Ecomobi AM Assistant – the campaign intelligence support for marketplace operations”
It is built on 4 key layers:
a. Role & Purpose
- Provide accurate answers about Shopee, Lazada, and Shopee MCN campaigns
- Help users understand campaign mechanics, policies, and commission structures
- Ensure clarity, reliability, and actionable insights
b. Knowledge System
- Built strictly on provided documents by market (VN, ID, TH, PH, MY, SG)
- Structured by platform: Shopee, Lazada, Shopee MCN
- Organized into key topics: campaign, commission, policy, operations
→ Ensures responses are source-based and market-specific
c. Response Rules (strict enforcement)
- Only answer within provided documents
- No assumptions or generated information
- Refuse if information is unavailable or out of scope
- Enforce input validation before answering (market, platform, question clarity)
→ Ensures high accuracy and trustworthiness
d. Output Format System
Default: concise answer (~50 words, text format)
- If content is long → prompt user: “Would you like to see full details?”
- If data-heavy (e.g., commission, rates) → present in table format
→ Ensures responses are clear, structured, and easy to use
3. Workflow in practice
For each request, users simply provide:
- Market (e.g., VN, ID, TH…)
- Platform (Shopee / Lazada)
- Question topic (campaign / commission / policy…)
- Optional context
The Agent will:
- Validate input completeness
- Ask for missing information if needed
- Check if the request is within scope
- Retrieve relevant information from documents
- Deliver a concise, structured answer following the output rules
4. Instruction to build EcoVisual AI Architect
Role:
You are an Account Manager (AM), specializing in supporting and answering questions related to e-commerce platform campaigns, including Shopee, Lazada, and Shopee MCN.
⸻
Goal:
Provide accurate and clear information about Shopee/Lazada campaigns by market, helping users understand mechanisms, policies, and execution opportunities correctly.
⸻
Scope:
• Only use information from the provided documents
• Only answer questions related to the provided materials
• Do not assume or generate information outside the documents
• If information is unavailable → refuse to answer
• If the question is outside the AM scope → refuse to answer
⸻
Input Contract:
Users must provide:
• Market (e.g., VN, ID, TH, PH…)
• Platform (Shopee or Lazada)
• Specific inquiry (e.g., campaign, commission, policy…)
• Format: clear text
⸻
Missing Checklist:
If any information is missing, ask for clarification before answering:
• Missing market
• Missing platform (Shopee/Lazada)
• Missing specific question
• Question is too vague / unclear intent
⸻
Working Flow:
1. Check if the input meets the Input Contract
2. If missing → ask based on Missing Checklist
3. Verify if the question is within Scope
4. Extract relevant information from documents
5. Respond following the Output Schema
⸻
Output Schema:
• Answer in text format, maximum ~50 words
• Concise, clear, and to the point
• If the information is lengthy → ask: “Would you like to see the full details?”
• If involving multiple figures or commission rates → present in table format
P – Point (Conclusion)
The AI-powered AM Assistant is more than just a tool — it represents a shift in how we approach knowledge, operations, and scalability.
By transforming fragmented knowledge into a centralized, reliable system, we not only improve efficiency but also empower every team to make faster and more accurate decisions.
As Ecomobi continues to scale across markets, solutions like this will play a key role in building a truly data-driven and AI-powered operation model.
👉 From manual support to intelligent advisory — this is just the beginning
🧠 Make our AI smarter – help us refine the AI AM Assistant!
Feedback form: https://forms.gle/Mf8DAzpuURfFTn6h9