Media & Communication
300+ concurrent users, 75% faster decisions, 98% cost reduction vs traditional methods.
We built PSi, a real-time voice chat platform where groups hold anonymous discussions and vote to reach consensus faster than traditional meetings or surveys. Dynamic table splitting keeps conversations intimate at scale, while real-time analytics give organizers instant insight into group sentiment - achieving 75% faster decision-making and 98% cost reduction compared to conventional approaches.
Client
PSi
Industry
Media & Communication
Timeline
14 weeks
Team Size
9 engineers + 1 PM
Impact
Concurrent users
Decision speed
Cost reduction
Delivery timeline
Concurrent users: Participants in simultaneous real-time discussions with dynamic table management.
Decision speed: Faster consensus-building compared to traditional meeting and survey workflows.
Cost reduction: Reduction in decision-making costs versus in-person focus groups and multi-day workshops.
Delivery timeline: From kickoff to production launch with real-time audio, dynamic tables, voting, and analytics.
The Challenge
Traditional meetings and surveys are slow, expensive, and limited in scale - getting 300 people to discuss and vote on a topic required multi-day conferences or lengthy email chains that killed momentum and ballooned costs.
Small focus groups produce limited, potentially biased insights. Organizations needed a way to gather input from hundreds of participants simultaneously without sacrificing the depth that comes from real conversation.
Existing collaboration tools forced a trade-off between scale and intimacy - video calls break down past 10 participants, while survey tools strip away nuance. No tool combined real-time discussion with structured consensus-building.
What We Built
Built dynamic discussion tables that automatically split and rebalance participants in under one second as groups grow - keeping every conversation at an optimal 6-8 person size while scaling to 300+ concurrent users.
Integrated Agora for low-latency real-time audio and video streaming, for natural conversation flow even across distributed participants with varying network conditions.
Implemented SSO via AWS Cognito for enterprise authentication with anonymous discussion mode - participants are verified but identities are hidden during discussions to encourage honest input.
Created a data-driven insights dashboard showing real-time voting results, sentiment trends, participation rates, and consensus metrics so organizers can act on findings immediately.
Related Work
When a discussion group exceeds the optimal size (typically 6-8 people), the system automatically splits participants into smaller tables in under one second. The algorithm balances groups based on participation level and topic interest, then merges tables back when participants leave. This keeps conversations productive at any scale without manual moderation.
Next Step
One call with a founder. No sales team, no follow-up sequence. If we can't help, we'll say so.