Solutions Architecture
Build version: 1.4.2
This document outlines the technical architecture for the MCP Agent Solutions platform, detailing the flow of data and technology stack integration.
System Overview
The MCP Agent Solutions platform operates as a lightweight, secure gateway (Client Portal) that
aggregated various Agentic Applications hosted on the dwaas.ash-de.au infrastructure.
The architecture utilizes a static frontend wrapper pattern to provide a unified user experience
while keeping the core agent logic decoupled.
Tech Stack
- Frontend Host: Static Web Server (HTML5, CSS3, Vanilla JS)
- Authentication: Client-side Access Code (LocalStorage Session)
- Integration Pattern: IFrame Encapsulation
- Backend/Agent Host:
dwaas.ash-de.au(External Service)
Data & Process Flow
Sequence of Operations
Enterprise Integration Model (Internal Host)
This model illustrates the architecture when the solution is hosted within a client organisation (e.g., Insurance or Superannuation Fund) using enterprise platforms like Microsoft 365 (Copilot) or Google Workspace (Gemini). In this scenario, Give More consultants recruit and hire staff who then utilise these internal tools.
Workflow Description
- Start: A prompt or instruction is initiated by the Recruitment Consultant.
- Orchestration: The Enterprise AI (Copilot/Gemini) processes the request.
- Context Retrieval: The AI securely accesses internal "Referential Documentation" (HR Policies, Job Specs) to ground its responses.
- Artefact Generation: The AI produces necessary documents (Screening Reports, Candidate Summaries).
- Storage & Indexing: Generated artefacts are stored in the organisation's repository (SharePoint/Drive) and indexed for future retrieval.
(User)") Prompt["Prompt / Instruction
(e.g. Email / Teams Chat)"] end subgraph ClientOrg ["Client Organisation Environment (e.g., Insurance/Super Fund)"] subgraph AIPlatform ["AI Platform (e.g. MS Copilot / Google Gemini)"] Orchestrator["AI Orchestrator
(e.g. Copilot Studio)"] LLM["Language Model
(e.g. GPT-4 / Gemini)"] end subgraph KnowledgeBase ["Enterprise Knowledge Base"] RefDocs[("Referential Docs
(e.g. SharePoint / Intranet)")] CandidateDB[("Candidate Database
(e.g. Salesforce / Workday)")] end subgraph ArtefactMgmt ["Artefact Management"] GenDocs["Generated Artefacts
(e.g. Word / PDF reports)"] Index["Search Index
(e.g. MS Graph)"] Storage[("Enterprise Storage
(e.g. OneDrive / Azure)")] end end %% Flow Consultant -->|1. Define Requirements| Prompt Prompt -->|2. Submit| Orchestrator Orchestrator <-->|3. Retrieve Context| RefDocs Orchestrator <-->|3. Check History| CandidateDB Orchestrator -->|4. Generate| LLM LLM -->|5. Produce| GenDocs GenDocs -->|6. Store & Index| Storage Storage -.->|Updated Index| Index Consultant -.->|7. Retrieve/Search| Index Index -.->|8. Access| Storage %% Styling style Consultant fill:#f9f,stroke:#333 style ClientOrg fill:#fff3e0,stroke:#e65100,stroke-dasharray: 5 5 style AIPlatform fill:#e3f2fd,stroke:#1565c0 style ArtefactMgmt fill:#e8f5e9,stroke:#2e7d32