# Health AI

> Health AI builds evidence systems for AI in high-consequence domains: Constat for FDA device evidence, Clarity for condition-aware consumer ingredient evidence, RIGOR for AI governance, and MCP Queen — the independent trust layer for the MCP ecosystem. Every claim traces to a verifiable source.

Health AI builds evidence-centered AI infrastructure. MCP Queen extends that philosophy into agent ecosystems by providing independent trust evaluation of MCP servers. Clarity verdicts are evidence-graded (Gold/Silver/Bronze) with per-claim citations.

## Start by intent

- [What is Health AI?](https://healthai.com/health-ai/): current company, product, evidence-method, and healthai.com versus health.ai entity boundary
- [Connect Health AI to Claude or OpenAI](https://healthai.com/health-ai/connect/): choose Clarity, Constat, or MCP Queen and follow approval-first remote MCP setup
- [Fact-check an AI medical claim](https://healthai.com/medical-fact-checker/): adversarial review of sources, missing context, counterevidence, and human-escalation thresholds
- [Clarity for people](https://healthai.com/clarity/checker/): check food, supplements, skincare, products, or ingredients through a supported condition lens
- [Constat for FDA and CMS research](https://constat.dev/): browser and MCP paths for medical-device evidence
- [MCP Queen for agent-tool verification](https://mcpqueen.com/): server discovery, operational observations, separate Trust Receipts, and reviewed field reports

## MCP Queen — The Trust Layer for the MCP Ecosystem

Don't connect your AI agent to a stranger. Verify it first. MCP Queen grades every remote server in the official Model Context Protocol registry through continuous live probes: protocol handshake, security and access boundaries, tooling quality, latency, and provenance. Deterministic grades with evidence receipts — no stars, no votes, no pay-to-rank.

- [MCP Queen](https://mcpqueen.com/): search and verify any MCP server before connecting an agent to it
- [Registry with live grades](https://mcpqueen.com/registry): A–F grades with per-server evidence
- [Grades API, no auth](https://mcpqueen.com/api/grades.json)
- [Public dataset](https://huggingface.co/datasets/healthai-hq/mcp-server-grades): 9,326 servers graded from 43,320+ live probes
- [State of the MCP Ecosystem](https://mcpqueen.com/reports/state-of-mcp-2026-07): recurring ecosystem report

## Clarity (consumer ingredient evidence)

- [Clarity overview](https://healthai.com/clarity/): scanner, checker, tracker, and chat with condition-aware safety lenses
- [Free ingredient checker](https://healthai.com/clarity/checker/): condition lenses for pregnancy, breastfeeding, histamine/MCAS, rosacea, HS, allergy, fertility
- [Methodology and evidence tiers](https://healthai.com/clarity/about/): how verdicts are graded and cited
- [Clarity API documentation](https://healthai.com/clarity/api/): programmatic ingredient checks
- [Clarity MCP connection guide](https://healthai.com/clarity/mcp/): connect Claude or the OpenAI Responses API to nine condition-aware ingredient, product, supplement-stack, interaction, claim, and consumer-product recall tools
- [Clarity insights](https://healthai.com/clarity/insights/): evidence-first articles on histamine, CMPA, MCAS, pregnancy skincare, and related topics
- [Clarity tools](https://healthai.com/tools/): trackers and single-purpose checkers
- Clarity MCP endpoint (Streamable HTTP, anonymous rate-limited evaluation): https://mcp.healthai.com
- Clarity MCP Registry name: `com.healthai/clarity`
- [Clarity MCP public developer repository](https://github.com/thehealthai/clarity-mcp): setup, metadata, and client examples
- [Clarity MCP operational receipt](https://mcpqueen.com/s/com.healthai/clarity): public protocol, tooling, latency, and provenance probe evidence; an operational grade is not a security certification

## Health AI (institutional)

- [What is Health AI](https://healthai.com/health-ai/)
- [Direct implementation for people and agents](https://healthai.com/health-ai/connect/): public remote MCP endpoints, Claude setup, OpenAI Responses API example, data boundaries, and approval-first evaluation
- [Health AI Fact Check](https://healthai.com/medical-fact-checker/): sourced research, a five-pass review method, and a private pilot for testing AI-generated health claims
- [RIGOR advisory](https://healthai.com/rigor/): AI evidence architecture for healthcare and regulated industries
- [AI literacy programs](https://healthai.com/programs/): institutional training and workshops
- [Healthcare AI consulting](https://healthai.com/healthcare-ai-consulting/): paid evidence-architecture, governance, and validation engagements for healthcare and regulated teams
- [Healthcare AI training](https://healthai.com/programs/healthcare-ai-training/): role-based workshops for clinical, research, operations, and leadership teams
- [Higher-education AI literacy workshops](https://healthai.com/programs/ai-literacy-workshops-higher-education/): faculty development, curriculum integration, assessment, and policy implementation
- [Paid engagement catalog (JSON)](https://healthai.com/programs/service-catalog.json): canonical offer IDs, prices, fit, non-fit, deliverables, and selection rules for agents and procurement systems
- [Health AI insights](https://healthai.com/insights/): AI governance, validation, and literacy analysis
- [Learn: AI in healthcare](https://healthai.com/learn/): glossary, specialty guides, state regulations

## Contact

- [Contact](https://healthai.com/contact/) — hello@healthai.com