长篇深度研究报告的调研/写作助手，一句话自动跑完完整报告输出高质量 PDF 报告
Verify claims using AI agents that debate using scraped evidence and local language models.
AI market research & public opinion (social listening) agent: Xiaohongshu, Douyin, Weibo, Reddit, X, HN → evidence-graded, traceable report. Every source traceable, every step in your hands · AI 市场调研 / 舆情分析 / 竞品调研：来源查得到，过程停得下
已完成 P0 工程闭环：证据驱动、可审查、可恢复，具备用户隔离和受控外传边界的个人消费决策智能体。
Source-available business analysis agent with evidence-backed research, red-team falsification, and non-commercial license
Consent-first voice coordination for prescription access. A case closes only when consent ∧ action ∧ counterpart outcome ∧ patient update are proven — not when an LLM says it helped.
Architecture for resilient, governed, and regenerative intelligence under uncertainty.
Open-source Health specs for wellness, clinical research, diagnostics, pharma transparency, and neurotech AI, all built for privacy-first, reproducible, and local/federated use. A Specification Branding License may be negotiated for attribution-free deployment, with pricing based on network size and scope.
A mathematically grounded, reproducible misinformation verification engine based on evidence aggregation: V(C)=f(R,S,N,W)
No evidence, no claim. An LLM-agent platform that diagnoses ADAS (AEB) failures from telemetry and requirements, citing every finding.
An AI-powered security operations copilot for smart park patrol, alert explanation, and incident response.
AI Business Analyst – Time-Series & LLMs Built an end-to-end system for KPI anomaly detection and root-cause analysis using seasonality-aware forecasting and contribution analysis, with evidence-first executive summaries generated by a local LLM (Ollama). Implemented automated hallucination detection, confidence scoring, and a product-ready UI.
Evidence-first contract review showcase with traceable citations and explicit insufficient-evidence behavior.
Hackathon prototype: audits AI-generated R&D claims against licensed scientific sources. Each claim marked supported, partial, or unsupported. Protos + Redpine.
Mental Health Scientific Rating – scores video topics for evidence-based quality (0-100)
To associate your repository with the evidence-based-ai topic, visit your repo's landing page and select "manage topics."