RISING_SUN BIOS v3.14
Copyright (C) 2025 Rising Sun Industries
Initializing system...
Memory check: 64GB OK
Loading kernel modules...
[OK] display.driver
[OK] network.stack
[OK] ascii.renderer
[OK] terminal.emulator
Mounting filesystems...
/dev/projects mounted
/dev/updates mounted
/dev/portfolio mounted
Starting services...
creativity.daemon [RUNNING]
code.compiler [RUNNING]
caffeine.monitor [CRITICAL]
System ready.
Welcome to RISING_SUN
Press any key to skip...

Chorus

BETA

Multi-agent ensemble forecasting

#ai#forecasting#python#api#multi-agent#fastapi
────────────────────────────────────────────────────────────────────────────────────────────────────
────────────────────────────────────────────────────────────────────────────────────────────────────

Chorus replaces expensive human analyst teams with an ensemble of 17 AI specialists, each approaching your question from a different domain. The result is a calibrated probability estimate with transparent reasoning — not a black box, not a single model's opinion.

Key Features

  • >17-Specialist Ensemble: Reasoner, Intelligence Analyst, Market Analyst, Researcher, Data Scientist, and more — each with a domain-specific analytical playbook
  • >Parallel Fan-Out: All specialists run simultaneously via asyncio — full ensemble in the time it takes one specialist
  • >TOML-Driven Playbooks: Each specialist's analytical framework is loaded from a manifest at runtime — update the playbook, update the behavior
  • >Confidence-Weighted Aggregation: High-confidence specialists count more; low-confidence signals are down-weighted automatically
  • >Consensus Labels: `strongly_yes` / `lean_yes` / `uncertain` / `lean_no` / `strongly_no` — human-readable in addition to the probability
  • >Synthesis Pass: Optional final LLM call explains where specialists agreed, disagreed, and what the key residual uncertainty is
  • >Full Reasoning Chains: Every specialist returns its step-by-step analysis, not just a number
  • >REST API: Single `POST /v1/forecast` call returns the complete ensemble

How It Works

  1. 1.1. Submit a binary probabilistic question: *"Will the Fed cut rates before September 2026?"*Submit a binary probabilistic question: *"Will the Fed cut rates before September 2026?"*
  2. 2.2. Chorus fans it out to your chosen specialists in parallelChorus fans it out to your chosen specialists in parallel
  3. 3.3. Each specialist applies its domain playbook and returns `{probability, confidence, reasoning, assumptions, uncertainties}`Each specialist applies its domain playbook and returns `{probability, confidence, reasoning, assumptions, uncertainties}`
  4. 4.4. Responses aggregate: confidence-weighted average probability, consensus label, optional synthesisResponses aggregate: confidence-weighted average probability, consensus label, optional synthesis
  5. 5.5. You receive the ensemble result + every specialist's individual takeYou receive the ensemble result + every specialist's individual take

Use Cases

  • >Hedge Funds: Calibrated probability estimates for macro and market events
  • >Strategy Consultants: Quantify scenario likelihood for client presentations
  • >Political Risk Analysts: Multi-perspective assessment of geopolitical outcomes
  • >Researchers: Structured elicitation of AI domain knowledge across specialties

Revival Story

Cindicator raised $15M to combine crowd-sourced human predictions with ML for institutional-grade forecasts. The crowd was too noisy and expensive. Chorus replaces the human crowd with 17 AI specialists — faster, cheaper, and more consistent.

Tech Stack

FastAPI, Python, Anthropic Claude API, asyncio parallel execution, TOML manifest system. Part of the AIOS specialist mesh.