IT Bridge Solution - Original product development: MCC Agent & GAIA/ORION/ZEUS Backend

Original product development: MCC Agent & GAIA/ORION/ZEUS Backend.

Explanation: Microservice Communication Core (MCC)

1. Introduction: Why This Architecture Was Necessary

Today's IT systems are rapidly evolving, combining microservices with various cloud services. However, this also presents challenges like complex inter-service dependencies and protocol inconsistencies.

Challenge 1: Tight Coupling

When Service A calls Service B's API directly, changes to Service B's specifications can have a major impact on Service A. This is called "tight coupling."

Challenge 2: Protocol Inconsistency

Different services and cloud providers adopt various protocols like REST API, gRPC, and Pub/Sub. This complicates communication logic and becomes a bottleneck for development and maintenance.

To solve these challenges, we developed our own communication layer: the MCC Agent.

2. The Role and Innovativeness of the Unique "MCC Agent"

The MCC Agent is an intelligent proxy positioned between microservices like GAIA, ORION, and ZEUS, and GCP products such as Vertex AI and BigQuery.

2.1. Innovation 1: Enforcing Loose Coupling

The MCC Agent's most important role is to completely sever dependencies between services.

From the Application's Perspective GAIA doesn't know the specifics of ORION's API. It simply asks the MCC Agent to "send a PredictWaitTime request to a service called ORION."
From the MCC Agent's Perspective When the MCC Agent receives this request, it locates the ORION service and communicates with it via gRPC.

This mechanism allows services to focus on their own business logic, significantly reducing the risk of a change in one service affecting another.

2.2. Innovation 2: The Protocol Abstraction Mechanism

The MCC Agent has a unique mechanism for achieving protocol unification and abstraction.

A "Common Language" for Services All microservices communicate with the MCC Agent using a single protocol: gRPC. This frees developers from worrying about the complexities of different protocols like HTTP or Pub/Sub.
Leveraging google.protobuf.Any Request payloads are wrapped in a generic type called google.protobuf.Any. The MCC Agent doesn't need to know the specific data format inside the Any type to forward the request. This keeps the MCC Agent's core logic simple, eliminating the need to modify its code even when the proto definitions for GAIA or ORION change.
The "Smart Translator" The MCC Agent dynamically translates incoming gRPC requests into the native protocols of external services, such as the BigQuery REST API or the Vertex AI gRPC API.

This protocol abstraction allows us to treat GCP services like Cloud Spanner or BigQuery as if they were our own self-developed microservices.

Explanation: Sequence Diagrams of Use Cases Using MCC

Legend

App: Core application logic for each service (GAIA App, ORION App, ZEUS App)
MCC Client: The client part of the MCC Agent within each service (receives requests from the App via UDS and sends them to other services via gRPC)
MCC Server: The server part of the MCC Agent within each service (receives requests from other services via gRPC and notifies the App via UDS)
UDS: Unix Domain Socket communication
gRPC: gRPC communication (between MCC Client and MCC Server)
Service Mesh/LB: Service Mesh or Load Balancer (Envoy/Istio/GCP Load Balancer)
SR: Service Registry

1. GAIA: AI Concierge Training / Personalized Dialogue

Use Case: Customer dialogue start notification, preference prediction, suggestions, order requests, and wait time prediction.

sequenceDiagram participant Customer as Customer participant Mobile_UI as Mobile UI (Vue.js) participant GAIA_App as GAIA App Logic participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant ORION_MCC_S as ORION MCC Server participant ORION_App as ORION App Logic participant VertexAI as Vertex AI participant CloudSpanner as Cloud Spanner participant SR as Service Registry
Customer->>Mobile_UI: Browses menu and starts a dialogue with the AI Mobile_UI->>GAIA_App: Dialogue start request (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: Vertex AI, method: PredictTaste) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] Vertex AI Endpoint GAIA_MCC_C->>VertexAI: [gRPC] PredictTasteRequest (Customer ID, historical data) VertexAI-->>GAIA_MCC_C: [gRPC] PredictTasteResponse ("Likes spicy food") GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse ("Likes spicy food") deactivate GAIA_MCC_C
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: Cloud Spanner, method: GetPreference) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] Cloud Spanner Endpoint (*Spanner is an internal API) GAIA_MCC_C->>CloudSpanner: [gRPC] GetPreferenceRequest CloudSpanner-->>GAIA_MCC_C: [gRPC] GetPreferenceResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App-->>Mobile_UI: Asks "You like spicy food, don't you?" Customer->>Mobile_UI: Replies "Yes, I do" Mobile_UI->>GAIA_App: Sends customer's reply (HTTP)
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: Cloud Spanner, method: UpdatePreference) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] UpdatePreferenceRequest (updates preference data) CloudSpanner-->>GAIA_MCC_C: [gRPC] UpdatePreferenceResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App-->>Mobile_UI: Suggests "How about our new super spicy ramen?" Customer->>Mobile_UI: Orders super spicy ramen Mobile_UI->>GAIA_App: Order request (HTTP)
GAIA_App->>GAIA_App: Starts internal processing after order confirmation GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: ORION_App, method: PredictWaitTime) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] ORION_App Endpoint GAIA_MCC_C->>ORION_MCC_S: [gRPC] PredictWaitTimeRequest (order information) activate ORION_MCC_S ORION_MCC_S->>ORION_App: [UDS] InboundRequestNotification (order information) ORION_App->>ORION_App: Wait time prediction logic ORION_App-->>ORION_MCC_S: [UDS] InboundResponse (Wait time: "5 minutes") deactivate ORION_MCC_S ORION_MCC_S-->>GAIA_MCC_C: [gRPC] PredictWaitTimeResponse ("5 minutes") GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse ("5 minutes") deactivate GAIA_MCC_C
GAIA_App-->>Mobile_UI: "Your order will be ready in about 5 minutes. What have you been into recently?"

2. GAIA: AI Friend Feature (Friend Registration)

Use Case: A customer sends an AI friend registration request and it's approved.

sequenceDiagram participant C1_UI as Customer 1 Mobile UI participant C2_UI as Customer 2 Mobile UI participant GAIA_App as GAIA App Logic participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant CloudSpanner as Cloud Spanner participant SR as Service Registry
C1_UI->>GAIA_App: AI friend registration request (to C2's AI) (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: Cloud Spanner, method: SaveFriendRequest) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] SaveFriendRequest (saves request history) CloudSpanner-->>GAIA_MCC_C: [gRPC] SaveFriendRequestResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App->>GAIA_App: Generates a notification for C2's AI GAIA_App-->>C2_UI: Notifies C2 "You have a new friend request" (HTTP/WebSocket)
C2_UI->>GAIA_App: Approves friend request (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: Cloud Spanner, method: RegisterFriendship) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] RegisterFriendship (registers friendship) CloudSpanner-->>GAIA_MCC_C: [gRPC] RegisterFriendshipResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C GAIA_App->>GAIA_App: Internal processing after friendship registration (e.g., notifies both AIs)

3. GAIA: AI Friend Feature (Getting Preference Info)

Use Case: Customer 1's AI gets food preference info from Customer 2's AI.

sequenceDiagram participant C1_UI as Customer 1 Mobile UI participant GAIA_App as GAIA App Logic participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant CloudSpanner as Cloud Spanner participant SR as Service Registry participant VertexAI as Vertex AI
C1_UI->>GAIA_App: Request to ask C2's AI for preferences (HTTP) GAIA_App->>GAIA_App: Checks for C2's AI and friendship status (internal processing, queries Spanner if needed) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: GetFriendPreferences) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] VertexAI Endpoint GAIA_MCC_C->>VertexAI: [gRPC] GetFriendPreferencesRequest (C2's AI ID, permission info) VertexAI-->>GAIA_MCC_C: [gRPC] GetFriendPreferencesResponse (C2's AI preferences: "Likes sweet things") GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse ("Likes sweet things") deactivate GAIA_MCC_C
GAIA_App-->>C1_UI: Displays "According to XX's AI, they seem to like sweet things."

4. GAIA: Prediction Quiz Feature / Customer Satisfaction Improvement / Enhanced Hospitality (Employee UI Integration)

Use Case: Customer order prediction, "Like" notification for employees, and favorite employee registration.

sequenceDiagram participant Customer as Customer participant Mobile_UI as Mobile UI (Vue.js) participant GAIA_App as GAIA App Logic participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant ORION_MCC_S as ORION MCC Server participant ORION_MCC_C as ORION MCC Client participant ORION_App as ORION App Logic participant Emp_UI as Employee UI (Vue.js) participant CloudSpanner as Cloud Spanner participant SR as Service Registry participant VertexAI as Vertex AI
Customer->>Mobile_UI: Confirms order (e.g., super spicy ramen) Mobile_UI->>GAIA_App: Sends order data (HTTP)
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: PredictOrder) activate GAIA_MCC_C GAIA_MCC_C->>VertexAI: [gRPC] PredictOrderRequest (customer data) VertexAI-->>GAIA_MCC_C: [gRPC] PredictOrderResponse (Prediction: "super spicy ramen", Probability: 90%) GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: ORION_App, method: NotifyOrderPrediction) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] ORION_App Endpoint GAIA_MCC_C->>ORION_MCC_S: [gRPC] NotifyOrderPrediction (Customer ID, predicted order, probability) activate ORION_MCC_S ORION_MCC_S->>ORION_App: [UDS] InboundRequestNotification (prediction quiz data) deactivate ORION_MCC_S ORION_App->>Emp_UI: [HTTP/WebSocket] Prediction quiz notification GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse (returns immediately for async notification) deactivate GAIA_MCC_C
Customer->>Mobile_UI: Views a simple employee profile Mobile_UI->>GAIA_App: Employee profile request (HTTP) GAIA_App->>GAIA_App: Gets employee profile (internal processing, could get from ORION) GAIA_App-->>Mobile_UI: Displays employee profile
Customer->>Mobile_UI: "Likes" an employee Mobile_UI->>GAIA_App: "Like" request (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: CloudSpanner, method: RecordCustomerLike) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] RecordCustomerLike (saves "like" history) CloudSpanner-->>GAIA_MCC_C: [gRPC] RecordCustomerLikeResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: ORION_App, method: NotifyEmployeeLike) activate GAIA_MCC_C GAIA_MCC_C->>ORION_MCC_S: [gRPC] NotifyEmployeeLike (Employee ID, Customer ID, service quality) activate ORION_MCC_S ORION_MCC_S->>ORION_App: [UDS] InboundRequestNotification (like notification data) deactivate ORION_MCC_S ORION_App->>Emp_UI: [HTTP/WebSocket] "Someone liked your service!" notification GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
Customer->>Mobile_UI: Sets a favorite employee (one per store, unchangeable) Mobile_UI->>GAIA_App: Favorite employee registration request (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: CloudSpanner, method: RegisterFavoriteEmployee) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] RegisterFavoriteEmployee CloudSpanner-->>GAIA_MCC_C: [gRPC] RegisterFavoriteEmployeeResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C

5. GAIA: Teamwork Improvement (Employee UI Integration)

Use Case: An employee sets a favorite employee and sends a "Like."

sequenceDiagram participant Emp_UI_Sender as Employee UI (Sender) participant Emp_UI_Receiver as Employee UI (Receiver) participant GAIA_App as GAIA App Logic participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant CloudSpanner as Cloud Spanner participant SR as Service Registry
Emp_UI_Sender->>GAIA_App: Favorite employee setting request (HTTP) (One for manager/employee each) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: CloudSpanner, method: RegisterInternalFavorite) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] RegisterInternalFavorite CloudSpanner-->>GAIA_MCC_C: [gRPC] RegisterInternalFavoriteResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
Emp_UI_Sender->>GAIA_App: "Like" request to an employee (HTTP) GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: CloudSpanner, method: RecordInternalLike) activate GAIA_MCC_C GAIA_MCC_C->>CloudSpanner: [gRPC] RecordInternalLike CloudSpanner-->>GAIA_MCC_C: [gRPC] RecordInternalLikeResponse GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C
GAIA_App->>GAIA_MCC_C: [UDS] OutboundRequest (target: GAIA_App, method: NotifyInternalLike) (to its own MCC Server) activate GAIA_MCC_C GAIA_MCC_C->>SR: [Lookup] GAIA_App Endpoint (routes to all GAIA MCC Servers in the store) GAIA_MCC_C->>GAIA_MCC_S: [gRPC] NotifyInternalLike (Receiver Employee ID, Sender ID) activate GAIA_MCC_S GAIA_MCC_S->>GAIA_App: [UDS] InboundRequestNotification (like notification data) deactivate GAIA_MCC_S GAIA_App->>Emp_UI_Receiver: [HTTP/WebSocket] "Someone liked you!" notification GAIA_MCC_C-->>GAIA_App: [UDS] OutboundResponse deactivate GAIA_MCC_C

6. ORION: Automated Shift Generation / Visualizing the "Why" behind AI Predictions / Creating Capacity / Notification Content

Use Case: Shift generation and notification to employees.

sequenceDiagram participant CloudScheduler as Cloud Scheduler participant ORION_App as ORION App Logic participant ORION_MCC_C as ORION MCC Client participant ORION_MCC_S as ORION MCC Server participant Emp_UI as Employee UI participant VertexAI as Vertex AI participant CloudSQL as Cloud SQL participant GAIA_MCC_C as GAIA MCC Client participant GAIA_MCC_S as GAIA MCC Server participant SR as Service Registry
CloudScheduler->>ORION_App: Shift generation trigger (HTTP/Pub/Sub) ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: PredictDemand) activate ORION_MCC_C ORION_MCC_C->>SR: [Lookup] VertexAI Endpoint ORION_MCC_C->>VertexAI: [gRPC] PredictDemandRequest (historical sales, space level data from GAIA, etc.) VertexAI-->>ORION_MCC_C: [gRPC] PredictDemandResponse (demand prediction, reason for shift generation) ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: CloudSQL, method: GenerateShift) activate ORION_MCC_C ORION_MCC_C->>CloudSQL: [gRPC] GenerateShift (generates optimal shifts, along with reasons) CloudSQL-->>ORION_MCC_C: [gRPC] GenerateShiftResponse (generated shifts, ares of shortage) ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>ORION_App: Logic to notify all employees of shift shortages ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: GAIA_App, method: NotifyShiftDeficiency) activate ORION_MCC_C GAIA_MCC_C->>SR: [Lookup] GAIA_App Endpoint (routes to all GAIA MCC Servers in the store) ORION_MCC_C->>GAIA_MCC_S: [gRPC] NotifyShiftDeficiency (detailed shift shortage, sales prediction, etc.) activate GAIA_MCC_S GAIA_MCC_S->>GAIA_App: [UDS] InboundRequestNotification (shift shortage notification data) deactivate GAIA_MCC_S GAIA_App->>Emp_UI: [HTTP/WebSocket] Shift shortage notification ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C

7. ORION: Resolution Notification / Contribution Visualization

Use Case: An employee fills a shift shortage, and their contribution is visualized.

sequenceDiagram participant Emp_UI_Worker as Employee UI (Worker) participant Emp_UI_Team as Employee UI (Other team members) participant ORION_App as ORION App Logic participant ORION_MCC_C as ORION MCC Client participant ORION_MCC_S as ORION MCC Server participant CloudSQL as Cloud SQL participant GAIA_MCC_S as GAIA MCC Server participant GAIA_App as GAIA App Logic participant SR as Service Registry
Emp_UI_Worker->>ORION_App: Request to fill a shift shortage (HTTP) ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: CloudSQL, method: UpdateShift) activate ORION_MCC_C ORION_MCC_C->>CloudSQL: [gRPC] UpdateShift (updates shift info, records contribution) CloudSQL-->>ORION_MCC_C: [gRPC] UpdateShiftResponse ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>ORION_App: Logic for notification after shift shortage resolution ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: GAIA_App, method: NotifyShiftContribution) activate ORION_MCC_C ORION_MCC_C->>SR: [Lookup] GAIA_App Endpoint (routes to all GAIA MCC Servers in the store) ORION_MCC_C->>GAIA_MCC_S: [gRPC] NotifyShiftContribution (Contributor ID, contribution details, badge info) activate GAIA_MCC_S GAIA_MCC_S->>GAIA_App: [UDS] InboundRequestNotification (contribution notification data) deactivate GAIA_MCC_S GAIA_App->>Emp_UI_Team: [HTTP/WebSocket] "XX filled a shift shortage!" notification + badge award ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C

8. ORION: Space-Level Visualization (ZEUS Integration)

Use Case: Aggregating customer space levels and teamwork levels and feeding them to ZEUS.

sequenceDiagram participant ORION_App as ORION App Logic participant ORION_MCC_C as ORION MCC Client participant ORION_MCC_S as ORION MCC Server participant ZEUS_MCC_S as ZEUS MCC Server participant ZEUS_App as ZEUS App Logic participant CloudSQL as Cloud SQL participant BigQuery as BigQuery participant CloudScheduler as Cloud Scheduler participant SR as Service Registry
CloudScheduler->>ORION_App: Periodic space level aggregation trigger (HTTP/Pub/Sub) ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: CloudSQL, method: AggregateSpaceLevels) activate ORION_MCC_C ORION_MCC_C->>CloudSQL: [gRPC] AggregateSpaceLevels (aggregates from customer "likes" and employee "likes") CloudSQL-->>ORION_MCC_C: [gRPC] AggregateSpaceLevelsResponse (customer space levels and teamwork levels by time period) ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: ZEUS_App, method: FeedSpaceLevels) activate ORION_MCC_C ORION_MCC_C->>SR: [Lookup] ZEUS_App Endpoint ORION_MCC_C->>ZEUS_MCC_S: [gRPC] FeedSpaceLevels (aggregated level data, anonymized) activate ZEUS_MCC_S ZEUS_MCC_S->>ZEUS_App: [UDS] InboundRequestNotification (space level data) deactivate ZEUS_MCC_S ZEUS_App->>ZEUS_App: Triggers data processing and writing to BigQuery ZEUS_App->>BigQuery: [API] InsertSpaceLevelData ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C

9. ORION: Autonomous Ordering

Use Case: The AI automatically creates an order list, and an employee performs a final check.

sequenceDiagram participant CloudScheduler as Cloud Scheduler participant ORION_App as ORION App Logic participant ORION_MCC_C as ORION MCC Client participant ORION_MCC_S as ORION MCC Server participant Emp_UI as Employee UI participant VertexAI as Vertex AI participant CloudSQL as Cloud SQL participant SR as Service Registry
CloudScheduler->>ORION_App: Autonomous ordering trigger (HTTP/Pub/Sub) ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: PredictConsumption) activate ORION_MCC_C ORION_MCC_C->>SR: [Lookup] VertexAI Endpoint ORION_MCC_C->>VertexAI: [gRPC] PredictConsumptionRequest (consumption pattern data) VertexAI-->>ORION_MCC_C: [gRPC] PredictConsumptionResponse (predicted consumption, recommended order list) ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: CloudSQL, method: SaveOrderList) activate ORION_MCC_C ORION_MCC_C->>CloudSQL: [gRPC] SaveOrderList (auto-generated order list) CloudSQL-->>ORION_MCC_C: [gRPC] SaveOrderListResponse ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C
ORION_App->>Emp_UI: [HTTP/WebSocket] Order list check notification Emp_UI->>ORION_App: Final check/approval of order list (HTTP) ORION_App->>ORION_MCC_C: [UDS] OutboundRequest (target: Vendor_API, method: PlaceOrder) (*Integration with external vendor API) activate ORION_MCC_C ORION_MCC_C->>SR: [Lookup] Vendor_API Endpoint (abstracted via MCC even for external service) ORION_MCC_C->>Vendor_API: [gRPC/HTTP] PlaceOrder (finalized order) Vendor_API-->>ORION_MCC_C: [gRPC/HTTP] PlaceOrderResponse ORION_MCC_C-->>ORION_App: [UDS] OutboundResponse deactivate ORION_MCC_C

10. ZEUS: Integrated Analytics Dashboard / Strict Data Governance / Co-creation with AI / Fostering Collaboration / Managerial Training Goals

Use Case: Displaying dashboard data, manager's future prediction input, and dialogue with AI.

sequenceDiagram participant CloudScheduler as Cloud Scheduler participant CloudFunctions as Cloud Functions (ETL) participant CloudSpanner as Cloud Spanner participant CloudSQL as Cloud SQL participant BigQuery as BigQuery participant ZEUS_App as ZEUS App Logic participant ZEUS_MCC_C as ZEUS MCC Client participant ZEUS_MCC_S as ZEUS MCC Server participant Mgr_UI as Manager UI (React) participant VertexAI as Vertex AI participant SR as Service Registry
CloudScheduler->>CloudFunctions: ETL trigger (Pub/Sub) CloudFunctions->>CloudSpanner: [API] Data acquisition CloudFunctions->>CloudSQL: [API] Data acquisition CloudFunctions->>CloudFunctions: Anonymization and pseudonymization processing CloudFunctions->>BigQuery: [API] Integrated data transfer
Mgr_UI->>ZEUS_App: Dashboard data display request (HTTP) ZEUS_App->>ZEUS_MCC_C: [UDS] OutboundRequest (target: BigQuery, method: QueryDashboardData) activate ZEUS_MCC_C ZEUS_MCC_C->>BigQuery: [gRPC/API] QueryDashboardData (integrated analytics data, customer space levels, teamwork levels, etc.) BigQuery-->>ZEUS_MCC_C: [gRPC/API] QueryDashboardDataResponse ZEUS_MCC_C-->>ZEUS_App: [UDS] OutboundResponse deactivate ZEUS_MCC_C ZEUS_App-->>Mgr_UI: Displays dashboard data
Mgr_UI->>ZEUS_App: Manager's future prediction input (HTTP) ZEUS_App->>ZEUS_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: ComparePrediction) activate ZEUS_MCC_C ZEUS_MCC_C->>SR: [Lookup] VertexAI Endpoint ZEUS_MCC_C->>VertexAI: [gRPC] ComparePredictionRequest (manager's prediction, AI prediction criteria) VertexAI-->>ZEUS_MCC_C: [gRPC] ComparePredictionResponse (basis of AI prediction, comparison result) ZEUS_MCC_C-->>ZEUS_App: [UDS] OutboundResponse deactivate ZEUS_MCC_C ZEUS_App-->>Mgr_UI: Displays prediction comparison and AI insights
Mgr_UI->>ZEUS_App: Asks the AI for the basis of a prediction (dialogue) (HTTP) ZEUS_App->>ZEUS_MCC_C: [UDS] OutboundRequest (target: VertexAI, method: ExplainPrediction) activate ZEUS_MCC_C ZEUS_MCC_C->>VertexAI: [gRPC] ExplainPredictionRequest (prediction ID, question content) VertexAI-->>ZEUS_MCC_C: [gRPC] ExplainPredictionResponse (AI's explanation) ZEUS_MCC_C-->>ZEUS_App: [UDS] OutboundResponse deactivate ZEUS_MCC_C ZEUS_App-->>Mgr_UI: Displays the AI's interactive explanation

Working Environment and Project Status: 2025.09.24

Working Hours

I'm working on the project from PM 6:00 to AM 3:00 (Asia/Tokyo) after my contract work is complete. The project is currently in the MVP (Minimum Viable Product) development phase.

Development Environment

As a pre-migration development environment for migrating to GCP, we've set up dummy servers for key GCP services to return responses.

Dummy Services

Vertex AI: A dummy service for machine learning models: This dummy service has methods like PredictTaste, PredictShiftDemand, ComparePredictions, and ExplainPrediction. It returns fixed dummy responses to requests from the MCC Agent.
Cloud Spanner: A dummy service for databases: This service mimics persistence requests from GAIA, such as RecordCustomerLike and SaveFriendRequest. It returns fixed dummy responses to requests from the MCC Agent.
Cloud SQL: A dummy service for a relational database: This service mimics shift management requests from ORION, such as GenerateShift and UpdateShift. It returns fixed dummy responses to requests from the MCC Agent.
BigQuery: A dummy service for a data warehouse: This dummy service has methods like GetDashboardData and FeedSpaceLevels. It returns fixed dummy responses to requests from the MCC Agent.

The project has already achieved significant milestones

GAIA -> Vertex AI (PredictTaste)

GAIA's test_client.go sends a UDS gRPC request to the MCC Agent.
The MCC Agent forwards a TCP gRPC request to the dummy Vertex AI.
The MCC Agent returns the response from the dummy Vertex AI to GAIA via UDS gRPC.
Result: Confirmed that GAIA can call Vertex AI's PredictTaste method via the MCC Agent.

GAIA -> Cloud Spanner (RecordCustomerLike, SaveFriendRequest, etc.)

Flow: Similar to the Vertex AI flow, a UDS gRPC request is sent from test_client.go, forwarded as TCP gRPC by the MCC Agent, and the response is returned to GAIA.
Result: Confirmed that GAIA can call Cloud Spanner's data manipulation methods via the MCC Agent.

ORION -> Vertex AI (PredictShiftDemand)

Flow: Similar to the GAIA-Vertex AI flow, with ORION's test_client.go initiating the request.
Result: Confirmed that ORION can call Vertex AI's PredictShiftDemand method via the MCC Agent.

ORION -> Cloud SQL (GenerateShift, UpdateShift, etc.)

Flow: A UDS gRPC request is sent from ORION's test_client.go, forwarded by the MCC Agent as TCP gRPC, and the response is returned to ORION.
Result: Confirmed that ORION can call Cloud SQL's shift management methods via the MCC Agent.

ZEUS -> BigQuery (GetDashboardData)

Flow: A UDS gRPC request is sent from ZEUS's test_client.go, forwarded by the MCC Agent as TCP gRPC, and the response is returned to ZEUS.
Result: Confirmed that ZEUS can call BigQuery's GetDashboardData method via the MCC Agent.

However, the project is still in its early stages, with plenty of room for growth

Full Implementation of ZEUS Use Cases

Remaining verification includes the flow for ZEUS to call Vertex AI's ComparePredictions and ExplainPrediction.
We also need to implement and verify the server logic for accepting FeedSpaceLevels requests from ORION to ZEUS.

Enhanced Error Handling

Currently, errors in the MCC Agent and dummy services are often handled with a simple fmt.Errorf. For production, we'll need more robust error handling, such as retries and circuit breakers.

Concrete Dummy Service Implementations

The dummy services currently return only fixed values. Adding more complex data processing and business logic to them would expand the scope of our Proof of Concept (PoC) verification.

Load and Performance Testing

We haven't yet verified performance aspects, such as the overhead of UDS gRPC vs. TCP gRPC or the throughput of the MCC Agent.

Page updated

Report abuse