Social Ace: Unlocking the World of Conversation through AI
Social Ace is an AI-driven social skills trainer that helps children with autism build communication confidence through interactive, gamified dialogue. Built on secure AWS architecture (Bedrock AgentCore and Nova Sonic2), it provides an affordable and engaging way to practice essential social interactions.
I am deeply honored that my vision for Social Ace has been recognized as a semi-finalist in the 10,000 Aideas competition. This project represents my commitment to using technology to solve real-world human challenges. This article describe my project in details.
Section 1: App Category: Social Impact
Social Ace is a dedicated Social Impact platform. I believe that digital innovation, especially AI, should serve the most underserved members of our society. Our application is designed to provide a "digital bridge" for children and teenagers facing communication barriers, specifically those within the Autism (ASD) community.
Section 2: Our Vision
Empowering the Silent Voice
For children with autism or language development delays, a simple conversation with a peer can feel like an impossible mountain to climb. They often struggle to initiate talk or maintain a topic, leading to social isolation.
Our vision is to transform this struggle into a journey of mastery. Social Ace is an engaging, AI-powered "Social Gymnasium" where youth can practice guided dialogue in a safe, judgment-free environment. By simulating daily life scenarios, we empower them to build the verbal confidence needed to navigate the real world.
Section 3: Why This Matters?
From Therapy to Play
Currently, social training for neurodivergent youth relies heavily on traditional methods like Applied Behavior Analysis (ABA) and speech therapy. While effective, the parents face three major hurdles:
- High Costs: Professional training fees often place a significant financial burden on families.
- Accessibility: Expert therapists are in short supply, leading to long waiting lists.
- Engagement: Traditional training can feel clinical and repetitive for a child.
Social Ace shifts the paradigm of social training. Our platform offers a 24/7 AI-driven alternative that bridges the gap between clinical sessions and daily life. By gamifying social practice and utilizing positive reinforcement in real-time, we ensure that mastering communication is both accessible and deeply rewarding for every child.
Core Functionality and Features:
- AI-Driven Dialogue: The system provides persona-based avatars across a variety of real-world social topics.
- Real-Time Feedback: The system delivers appropriate responses and provides gentle, constructive prompts for social course-correction.
- Positive Reinforcement: The system transform social practice into a rewarding "Mastery Journey." Users earn points and digital badges rewards upon completing scenarios.
- Smart Scaffolding (Hints & Cues): The system provides interactive "Cue Cards / Hints" and context-aware vocabulary hints to assist users when they face communication hurdles.
Section 4: How I Built It
a. Application Components Overview
Frontend Website
Social Ace provides an interactive web interface built with Next.js and ReactJs. Users can register through AWS Cognito authentication, browse different social topics (e.g. music, sports, travel), and engage in real-time voice conversations with AI avatars. The frontend features a responsive design with Tailwind CSS and real-time WebSocket communication for bidirectional audio streaming.
Backend AI Agent
The backend consists of AI agents developed using the Strands Agent framework and deployed on AWS Bedrock AgentCore. The system employs multi-agent orchestration with two specialized agents:
- Conversationalist Agent: Handles real-time voice conversations using Nova 2 Sonic model
- Evaluator Agent: Analyzes social behaviors using DeepSeek R1 model for real-time feedback
The agent architecture supports session persistence, dynamic prompt retrieval, and gamification features including points tracking and badge progression (Bronze, Silver, Gold).
b. Application Architecture
Social Ace leverages a serverless architecture on AWS for scalability and cost-effectiveness:
Architecture Diagram
Frontend Hosting
- AWS Amplify Hosting provides Git-based continuous deployment for the Next.js application
- Automatic builds triggered on every push to the main branch
- Built-in CloudFront CDN for global content delivery with automatic HTTPS
Backend Runtime
- AWS Bedrock AgentCore provides serverless runtime for the conversational agent
- Lightweight and cost-effective compared to ECS/Fargate
- Strong agentic capabilities with built-in WebSocket support
Data Layer
- DynamoDB stores prompts, user progress, and session history using a single-table design pattern
- Topic-specific conversation history isolation (each avatar maintains independent context)
- Automatic data expiration with 90-day TTL for privacy compliance
Serverless Functions
- AWS Lambda functions handle:
- Presigned URL generation for secure WebSocket connections (SigV4 authentication)
- Prompt retrieval from DynamoDB based on topic and age group
- User points and badge updates for gamification
- REST API proxy for hint and vocabulary features
API Gateway
- REST API with Cognito User Pool authorizer for JWT validation
- CORS-enabled endpoints for frontend integration
- Routes for WebSocket Presigned URL generation and chat assistant features
Authentication
- AWS Cognito provides OAuth2 authentication for the web frontend
- JWT tokens validated at API Gateway layer
- Presigned URLs with SigV4 signatures for AgentCore WebSocket connections
c. Agentic AI – Orchestration and Dialogue Evaluation
Multi-Agent Architecture :
Multi-agent achitecture
Social Ace employs a sophisticated multi-agent system with specialized roles:
1. Conversationalist Agent (Primary Agent)
- Model: Amazon Nova 2 Sonic
- Purpose: Real-time voice conversation and orchestration
- Features:
- Bidirectional audio streaming via WebSocket
- Dynamic prompt retrieval based on topic and age group
- Session management with conversation history tracking
- Tool calling for hint and vocabulary generation
2. Evaluator Agent (Sub-Agent)
- Model: DeepSeek R1 (us.deepseek.r1-v1:0)
- Purpose: Real-time social behavior analysis
- Features:
- Analyzes user responses against social behavior rubric
- Provides structured evaluation with confidence scores
- Triggers gamification updates via Lambda
3. Chat Assistant (Utility Agent - tools)
- Model: Amazon Nova Micro
- Purpose: Lightweight hint and vocabulary generation
- Features:
- Fast response time (<500ms)
- Cost-effective for frequent requests
- Generates 4-5 conversation hints or 10-15 topic-specific vocabulary words
Social Behavior Evaluation Logic :
The Evaluator Agent uses DeepSeek R1 to analyze user responses against a comprehensive rubric:
- Positive Behaviors (+1 point each, identify at most ONE):
- Joint_Attention: Acknowledging and building on the conversation partner's topic
- Social_Initiation: Starting a new relevant topic or asking engaging questions
- Conversational_Reciprocity: Balanced back-and-forth exchange with appropriate turn-taking
- Inquiry_Skills: Using Wh-questions (who, what, where, when, why, how) to show interest
- Topic_Maintenance: Staying on topic for multiple turns and developing the conversation
- Active_Listening: Using bridge phrases ("That's interesting!", "Tell me more") to show engagement
- Corrective Behaviors (-1 point each, identify at most ONE):
- Non_Contextual_Talk: Introducing topics that are clearly and completely unrelated with no connection to the current conversation
- Maladaptive_Language: Using rude, mean, or socially inappropriate language
- Repetitive_Speech: Repeating the exact same phrase or question 3+ times without variation
- Topic_Hijacking: Forcefully changing the subject without acknowledging the current topic
- Information_Monologuing: Long speeches without allowing turn-taking or checking for interest
- Interrogative_Overload: Asking multiple questions in rapid succession without waiting for answers
- Evaluation Flow:
- User completes a conversational turn
- Evaluator Agent analyzes the response with conversation context (last 5 turns)
- Identifies at most ONE behavior (positive, corrective, or none)
- Returns structured evaluation with confidence score
- Points Lambda updates user progress in DynamoDB
- Frontend receives real-time feedback via WebSocket
Agent Orchestration Framework:
The system uses the Strands Agent framework for agent orchestration:
Why Strands?
- Lightweight and simple architecture
- Excellent for rapid development and prototyping
- Built-in support for bidirectional voice streaming (i.e. BidiAgent)
- Easy integration with AWS Bedrock models
- Minimal boilerplate code
Model Selection Rationale:
| Model | Purpose | Key Characteristics |
|---|---|---|
| Nova 2 Sonic | Primary conversation (Speech to speech model) | Optimized for conversational AI with low latency and natural voice synthesis |
| DeepSeek R1 | Social behavior evaluation | Excellent reasoning capabilities for nuanced analysis (~1-2s response time) |
| Nova Micro | Hint/vocabulary generation | Ultra-fast and cost-effective for simple text generation (<500ms) |
This multi-model approach balances performance, cost, and quality across different use cases within the application.
d. Key Technical Decisions
1. Serverless Architecture
- Why: Cost-effective, auto-scaling, minimal operational overhead
2. Multi-Model Strategy
- Why: Optimize for specific use cases (conversation vs. evaluation vs. utility)
3. Single-Table DynamoDB Design
- Why: Efficient queries, reduced costs, simplified data model
4. Presigned URLs with SigV4
- Why: Secure WebSocket connections without exposing credentials to frontend
5. Topic-Specific History Isolation
- Why: Different avatars maintain independent conversation contexts
e. Development Process – Spec-Driven Development with Kiro
What is Kiro?
The entire Social Ace project was developed using Kiro, an AI-powered development tool that specializes in spec-driven development. Kiro assists developers through the complete software development lifecycle: from writing requirements, to designing solutions, to breaking down work into implementable tasks.
Why Spec-Driven Development?
Spec-driven development with Kiro help me to build better software than vibe-coding by:
- Formalizing requirements before writing code (requirements.md)
- Designing architecture with clear component boundaries (design.md)
- Breaking down work into testable, incremental tasks (tasks.md)
- Maintaining documentation that evolves with the codebase
My Development Approach:
I broke down the Social Ace project into five high-level requirements to make it more manageable and testable:
- Phase 1: Build a simple workable Voice Chat Agent with Amazon Nova Sonic 2 and Strands Agent as a base
- Phase 2: Enhance the backend agent with Evaluative feature through orchestration between Conversationalist Agent and Evaluator Agent
- Phase 3: Implement long-term persistence strategy. This enabled user recognition and conversation continuity
- Phase 4: Build Frontend Web App implementation with Next JS and React JS
- Phase 5: Enhance backend agent with hint/cue card/vocab features and security
Kiro helps me create a comprehensive set of requirements, design specifications, and technical details for each phase. The benefit of this approach is to ensure that every milestone is granular and verifiable, making it easier to monitor progress, conduct rigorous testing, and guarantee the technical excellence of the Social Ace platform.
Section 5: Demo
Project Video (demo start at 2:17):
Project Video
Screenshot:
Social Ace - Welcome Page
Social Topic Selection Page
Points awards and instance feedback provided by AI
Badges awarded - positive reinforcement
AI generated cue card / hint
Section 6: What I have learnt?
Developing Social Ace alongside AI agents provided deep insights into modern architecture and the shifting responsibilities of software engineers. Here are the key lessons from this journey.
1. Spec-Driven Development (SDD) vs. "Vibe Coding":
I have experimented with both "vibe coding" (ad-hoc development) and Spec-Driven Development (SDD). In my experience, SDD is the only professional way to build sustainable software. While vibe coding often results in disorganized and unmaintainable codebases, SDD provides the structural guidelines and clear documentation necessary to ensure a project's long-term success and scalability.
2. Multi-Phase Discussion and Collaboration with Kiro:
While I have utilized Kiro (my AI development partner) in the past, this project involved involve extensive discussion and collaboration with it. We partnered from the initial conceptualization through to complex architectural design. I have learned that the specification phase is the most crucial; once Kiro generates a spec, I engage in a rigorous process of questioning and clarification.
This iterative "handshake" between human and AI is what ensures the project’s foundation is sound.
3. Keep a Critical Eye: AI is Not always correct
A key lesson from this journey is that you must remain critical of AI suggestions. AI can provide technically incompatible designs. For example, we initially attempted to use Pre-signed URLs for both the WebSocket and REST endpoints of our AgentCore Agent.
After several rounds of "trial and error," we discovered that Bedrock AgentCore Agents do not support Pre-signed URLs for standard REST invocations in the same way they do for WebSockets. We explored alternatives like Custom JWT Authorizers, but found that certain configurations conflicted with the WebSocket handshake.
The Technical Solution: Ultimately, we implemented a hybrid security model:
- WebSockets: Secured via Pre-signed URLs for real-time streaming.
- REST Endpoints: Secured via AWS Signature Version 4 (SigV4).
This experience proved that you must carefully validate AI suggestions and conduct independent research before approving an architectural design.
4. Drive Early Testing
I learned to prompt Kiro to prioritize "testable" deliverables from day one. Instead of waiting for a full backend, I instructed Kiro to build UI-ready features early. This allowed me to visualize the Minimum Viable Product (MVP) immediately and catch design flaws before they became expensive problems.
5. Smart Architecture: DynamoDB vs. AgentCore Memory
One major pivot was our "Long-Term Memory" strategy. While AgentCore's built-in memory is easier to set up, Kiro and I determined that Amazon DynamoDB was a far more cost-effective and scalable solution for a multi-user game. Kiro helped me bridge the coding gap, resulting in a professional-grade architecture that keeps operating costs low.
6. The New Role of the Software Engineer
In this new era of AI, many argue that technology is replacing software engineer. In my opinion, it is not replacing but shifting our position. Our role has evolved from "Coder" to "Lead Architect." Your value now lies in your ability to maintain a high-level view, think critically, and rigorously judge the AI’s output. To build reliable software, the human must remain the ultimate authority on quality and logic.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article