100.Stream

• 

  High-level Architecture

Inspiration

In an era where misinformation spreads six times faster than truth on social media, we're witnessing democracy under siege, public health compromised by false medical claims, and innocent individuals destroyed by viral lies. The 2020 election misinformation, COVID-19 "miracle cures," and countless deepfake scandals have shown us that our current social platforms are failing to protect truth itself.

My inspiration for 100.Stream stems from recognizing this isn't just a technology problem—it's an existential threat to informed society. While traditional platforms rely on reactive fact-checking and opaque algorithms, I saw an opportunity to flip the script: What if we could build a community where truth-telling is financially rewarded, where users have skin in the game through reputation staking, and where AI agents work transparently to surface reliable information in real-time?

There's a growing hunger for authentic digital spaces where facts matter more than engagement metrics. 100.Stream is built to fill that gap—creating the first social platform where honesty pays, literally.

What it does

100.Stream runs on a multi-agent system, one of which evaluates the "truthness" of content especially for factual statements. For example, when a user posts a factual claim, our Truthfulness Agent searches the web for verification and assigns a confidence score. Users can toggle to see these scores, helping the community identify reliable information. Users are also gauged by a reputation scoring system where points are dynamically assigned and evaluated based on a reputation agent. The platform is auto-moderated by a moderation agent and users can choose to challenge / dispute take-downs through the 'Post Court', where they can stake their reputation points for a challenge and a jury will be formed to vote. High-reputation users earn a larger share of platform ad revenue, creating a direct financial incentive for posting authentic, valuable content rather than clickbait or misinformation.

Challenges we ran into

There were numerous challenges throughout the process of building 100.Stream, but the biggest was thinking and planning for the core agents that were to power the platform. It is incredibly important to understand how intricate content and platform moderation is, and how agents could power the 1st-layer and reduce the manual effort and load which moderation requires.

Accomplishments that we're proud of

I am incredibly proud of building out the MVP of the platform within an incredibly short period of time.

What's next for 100.Stream

I am looking towards expanding the moderation tier system and also the reputation points scoring, and actual ad-revenue distribution.

Built With

• fastapi
• gemini
• google
• nextjs
• python
• supabase
• tavily