Building a live-social platform requires more than delivering individual features. Video, chat, presence, reactions, notifications, and social interactions must work together reliably while users are actively engaging with one another. The project required a delivery and quality approach that could support: • Low-latency streaming through OBS and browser-based broadcasting. • Real-time chat, audience presence, and interactive engagement. • Live voting with token-based and non-token participation. • Recording, video processing, and video-on-demand experiences. • Social features including following, blocking, profiles, emails, and recommendations. • Creator tools for stream configuration, moderation, and analytics. • Member and Supporter subscription access and gated content. • Secure authentication and role-based access control. • Responsive experiences across desktop and mobile devices. The challenge was not simply to test whether individual features worked, but to validate how the entire platform behaved when real users interacted with it in real time.
Business Challenges
Solution
Gurzu approached the project through a combination of scalable web development, real-time infrastructure, structured quality engineering, and AI-assisted issue investigation.
Development & Real-Time Integration
The platform was developed using Next.js, TypeScript, PostgreSQL, and a real-time architecture designed around live streaming and interactive audience experiences.
LiveKit provides RTMP and WHIP ingress for creators broadcasting through OBS and other streaming software. WebSockets power real-time chat and audience interactions, while Server-Sent Events deliver live voting results.
The platform also integrates video processing and hosting, authentication, subscriptions, payments, file management, and notification services to support the complete creator and viewer experience.
Streaming & Broadcasting
Creators can configure and manage live broadcasts through dedicated creator tools, with support for:
- RTMP and WHIP streaming
- OBS and streaming software integration
- Live viewer counts and presence detection
- Stream ingress management
- Live status indicators
- Stream configuration and moderation controls
- Theater-style viewing experiences
Real-Time Chat & Moderation
The WebSocket-based chat system enables audiences to interact while streams are live.
Features include:
- Real-time messaging
- Viewer presence
- Reactions and audience interaction
- Slow chat mode and rate limiting
- Followers-only chat
- Chat enable/disable controls
- Participant moderation and removal
Interactive Voting & Audience Participation
Interactive voting is a core part of the OurWorld experience. Creators can engage audiences through live polls and episode voting while streams are active.
The platform supports:
- Real-time voting results through SSE.
- Token-based voting.
- Free/non-token voting.
- Per-world custom currencies.
- Viewer token balances and participation tracking.
This allows creators to turn audiences from passive viewers into active participants.
Social & Content Discovery
The platform includes a social layer that helps audiences discover creators and communities and continue interacting beyond individual streams.
Key capabilities include:
- Following and follower relationships
- Blocking
- User profiles and public browsing
- Community-focused tabs
- Following-based recommendations
- Stream discovery and filtering
- Search functionality
- Live-first content discovery
Creator Dashboard
Creators have access to dedicated tools for managing their presence and content on the platform.
The dashboard supports:
- Stream configuration and management
- OBS/streaming ingress setup
- Thumbnail uploads
- Stream moderation settings
- Performance and analytics visibility
- Creator-focused content management
Subscriptions, Tokens & Monetization
OurWorld incorporates multiple monetization mechanisms designed around creator communities.
The platform supports:
- Member and Supporter subscription tiers.
- Stripe-powered subscription billing.
- Premium and gated content.
- Per-world currencies.
- Creator token economies.
- Viewer token balances.
- Token-based audience participation.
This gives creators multiple ways to build communities while monetizing participation and premium experiences.
Technology Stack
Framework & Core
- Next.js 16 with App Router and React 19
- TypeScript with strict type checking
- Node.js 24.x
Database & ORM
- PostgreSQL with Neon for production
- TypeORM with migration-based schema evolution
- PostgreSQL full-text search
- Cascading relationships for data integrity
Real-Time & Streaming
- LiveKit for RTMP/WHIP video ingestion
- WebSockets (
ws) for real-time chat - Server-Sent Events (SSE) for live voting results
- Firebase/Firebase Admin for supporting real-time capabilities
- Mux for video processing and hosting
Authentication & Authorization
- Clerk for authentication and session management
- Webhook-based user synchronization
- Role-based access for Users, Creators, and Admins
- Protected routes and middleware-based access control
File Management & Media
- UploadThing for server-side file uploads
- AWS S3 SDK with presigned URLs for secure file access
- Sharp for image optimization and processing
UI & Design
- Tailwind CSS 4
- shadcn/ui component system
- Radix UI accessibility primitives
- Lucide React icons
- Motion for animations
- Sonner for toast notifications
State Management
- Zustand for client-side state
- React Hooks for component state
- React Server Components for server-side rendering and application workflows
Payments, Email & Webhooks
- Stripe for subscriptions and payments
- SendGrid for email notifications
- Svix for webhook management
Testing & Quality Engineering
Because much of the platform’s risk exists in human interaction and real-time behavior, Gurzu adopted an 60/40 QA model: approximately 60% manual testing and 40% automated testing.
Manual Testing - 60%
Manual testing provided broad coverage across user-facing workflows, including:
- Authentication and access control.
- World and episode creation.
- Following workflows.
- Notification workflows
- Creator workflows
- Other high-regression-risk application flows.
Exploratory testing was also performed during delivery cycles to identify timing, interaction, and usability issues that are difficult to capture through scripted automation.
Automated Testing - 40%
Automation was focused on stable, high-risk workflows where regressions would be expensive to discover late in the development cycle.
Automated coverage included areas such as:
- Going live and configuring streams through OBS.
- Connecting streaming workflows to YouTube and Twitch.
- Real-time chat and moderation.
- Audience voting and interactive experiences.
- Notifications and email workflows.
- Subscription and billing flows.
- Creator workflows.
- Cross-browser and responsive behavior.
- Exploratory testing of timing, usability, and interaction issues.
The approach intentionally avoided automating everything. Automation was expanded where a workflow proved stable and repeatable, while human testing remained responsible for complex, interactive product behavior.
Technology Stack
Framework & Core
- Next.js 16 with App Router (latest React 19)
- TypeScript (98.5% of codebase) with strict type checking
- Node.js 24.x runtime
Database & ORM
- PostgreSQL with Neon for production
- TypeORM for data management with migration-based schema evolution
- Support for PostgreSQL-specific features (fulltext search)
- Cascade deletes for data integrity
Real-time & Streaming
- LiveKit - RTMP/WHIP streaming protocols for video ingestion
- WebSockets via ws library for real-time chat
- Server-Sent Events (SSE) for live voting results
- Firebase/Firebase Admin for additional real-time capabilities
- Mux - Video processing and hosting
Authentication & Authorization
- Clerk - Modern authentication and session management
- Webhook-based user synchronization to database
- Role-based access (User, Creator, Admin)
File Management & Media
- UploadThing - Server-side file upload management
- AWS S3 SDK with presigned URLs for secure access
- Sharp - Image optimization and processing
UI & Design
- Tailwind CSS 4.0 with custom PostCSS configuration
- shadcn/ui components for consistent design system
- Radix UI primitives for accessibility
- Lucide React for icons
- Motion for animations
- Sonner for toast notifications
State Management
- Zustand for client-side state
- React Hooks for component state
- Server Components for server-side state
Additional Services
- Stripe - Subscription billing and payments
- SendGrid - Email notifications
- Svix - Webhook management
AI-Assisted Issue Investigation
Gurzu also incorporated AI into the engineering workflow by connecting Codex with the Linear workspace through the Linear MCP server.
When a tester reports a defect, the issue can contain reproduction steps, environment details, severity, and supporting context. Codex can retrieve the issue, investigate the relevant code, trace potential causes, and add its findings back to the same issue.
This creates a connected workflow from:
Issue → Investigation → Diagnosis → Implementation → Verification
Rather than separating bug reports from technical investigation, the relevant context remains connected throughout the development lifecycle.
The Results
Reliable Real-Time Experiences
The combination of real-time streaming, WebSocket communication, SSE-based voting, and dedicated QA coverage helped validate the interconnected experiences that define the platform.
Interactive Audience Engagement
Live voting, chat, reactions, social connections, and token-based participation transformed the viewing experience into an interactive community environment.
Flexible Creator Monetization
Member and Supporter subscriptions, premium content, custom currencies, and token-based participation provide creators with multiple mechanisms for engaging and monetizing their audiences.
Scalable Streaming Infrastructure
Multi-protocol streaming through RTMP and WHIP, combined with LiveKit and Mux, established an infrastructure capable of supporting live broadcasting and video-on-demand workflows.
Comprehensive Quality Coverage
The 60/40 QA model combined broad manual coverage of complex real-time behavior with automated regression testing for stable, high-risk workflows.
Faster Issue Investigation
Connecting Codex with Linear through MCP reduced the hand-off between defect reporting and technical investigation while keeping the original issue context attached to the investigation.
