BULX AI: Simulating 1024 Agents, Committees, and Markets

[netplux]

BULX.AI Is in an official pre-seed right now and they managed to make an AI prediction platform that forecasts outcomes by simulating people instead of extrapolating from history. A user asks a plain-language question and optionally adds context documents; BULX builds a knowledge graph of that world and populates it with up to 1,024 AI agents, each with its own personality, biases, incentives, and memory, with personas grounded in live social media posts on the topic. The agents post, argue, and change their minds on simulated social platforms, then an autonomous analyst agent interviews them and writes a report on where the crowd landed and why. The same behavioral engine powers four more systems: Market Prediction, which runs about 500 trader agents through an order book calibrated to an asset's 90-day history to produce a 30-day price path with confidence bands for stocks and crypto (beta); an Investment Committee of 12 AI specialists that interrogates a position or a Kalshi or Polymarket bet and delivers verdicts; Agent Foundry, where users design their own agents; and a Community Network where simulations, presets, and agents are shared and reused, so the platform compounds with every contribution. Usage is metered in BUX credits at a fixed $0.01 each, with a free monthly allowance and a full committee review costing about $0.30. BULX is built by ODDMock LLC.