AI Agentic

I. AI Agent Architecture and Tooling

The platform relies on a collaborative system of two primary Agents, with external tools providing critical context and orchestration.

Agent / Tool Platform Used Role & Function
Agent 1: The Research & Data Agent OpenAI + Tavily Search Input Processor & Fact Retriever. Reads user requests (sport, location, language). Uses Tavily Search for real-time, targeted web search to gather relevant, up-to-date facts, scores, and video links.
Agent 2: The Content Generation Agent OpenAI (GPT-4/o-series) Output Formatter & Translator. Takes structured data from Agent 1 and a predefined format (title, description, scores, etc.). It writes the news content and translates the final output into the user’s selected language.
Orchestration Layer Wakala AI Manager & Workflow Automation. Manages the flow between the two Agents, handles multilingual context, and integrates with the payment/token system APIs.

II. Core Feature Breakdown

1. 🌐 Dynamic Content Categorization

  • Primary Categories: Football (Soccer), NBA, Cricket, Tennis, Badminton, etc.
  • Secondary Entities (Filtering Layer): Coach, Sportsman/Athlete, Sports Event, Tournament/League.

2. 🌍 Hyper-Localization & Multilingual Support

  • Multilingual UI & Content: UI is instantly translated. Agent 2 delivers final news content directly in the user's chosen language.
  • Location Filtering: Users select Country/City, which is passed to Agent 1 for targeted news retrieval via Tavily Search.
  • "Near By" Functionality: Uses device geolocation to refine the search prompt, instructing the Research Agent to focus on local teams or events.

3. 💰 Monetization and Access Control

Subscription Model:

  • Tiered Access: Users subscribe for unlimited, automated news feeds within their chosen categories.
  • Category Selection: Subscriptions allow users to select a defined number of sports/teams for deep-dive news.

Token System (Pay-Per-Content):

  • Purchase Tokens: Users buy packages of digital tokens.
  • Gated Content: Tokens are spent to access specific, premium, or on-demand content outside their subscription, such as AI-generated deep analysis or single article access.

III. AI Agent Workflow Example

Scenario: User selects "NBA" and "Los Angeles" in Spanish.

  1. User Action: Sends request: {"sport": "NBA", "location": "Los Angeles, USA", "language": "es"}.
  2. Wakala AI (Orchestration): Routes the request to Agent 1.
  3. Agent 1 (Research & Data Agent):
    • Tool Call (Tavily Search): Queries for latest NBA news for Los Angeles teams (Lakers, Clippers) including scores.
    • Output: Returns structured JSON facts (e.g., score: "120-118", summary_en: "LeBron hits game-winner...").
  4. Agent 2 (Content Generation Agent):
    • Input: Receives JSON facts and the target language (es).
    • Output Generation: Writes the content:

      Title (es): ¡LeBron da la victoria! Lakers vencen a Celtics en un final de infarto.

      Description (es): Con un tiro ganador, la estrella de los Lakers aseguró el triunfo 120-118 contra su clásico rival.

  5. Final Delivery: Wakala AI displays the formatted, localized, and real-time news on spoorts.io.

Gallery

Contact Me

Chandigarh, India

Phone: +91 8699168417

Email: charnpreet1902@gmail.com

Lets get in touch. Send me a message: