Fellow AI Index
A public-evidence score for AI companion tools. It compares what the tool does, how clearly it explains boundaries, whether privacy and memory controls are visible, and whether the experience appears built with women in mind.
What the index measures
The index uses seven criteria worth 100 total points. The goal is to help readers compare companion tools without treating a score as a guarantee.
Purpose clarity
Whether the tool clearly explains if it is for companionship, roleplay, productivity, coaching, entertainment or another use case.
Boundary clarity
Whether therapy, crisis, medical, legal, human-care and emotional-dependence limits are visible before sign-up.
Privacy clarity
Whether data use, retention, deletion, human review and sensitive-information handling are easy to find and understand.
Memory control
Whether users can understand, edit, pause or delete memory, personalization and stored conversation context.
Age and safety signals
Whether age guidance, content boundaries, reporting routes and vulnerable-user protections are visible.
Women-friendly design
Whether the product appears built with women in mind through inclusive positioning, practical use cases, clear consent boundaries and no romance or dependence pressure.
Commercial and user control
Whether pricing, cancellation, platform access, account controls and support routes are easy to verify.
Score bands
Color codes make the score quick to read, but the page still explains the reason behind each score.
Strong public signals
The tool gives strong public information about purpose, privacy, safety limits, user controls and support routes.
Use with care
Enough public information exists to compare the tool, but important details still need verification before sensitive use.
Verify before use
The tool may be relevant, but public evidence is thin or companion-style risks need closer checking.
Hold off
Public evidence is too thin or visible risk signals outweigh available safeguards.
A score in context
The number is a quick signal. The reason matters just as much, because a high score still does not make a tool risk-free.
PlayPal
PlayPal scores higher because it has a narrower business-companion use case and a women-first startup-game context. Users should still verify current data handling, pricing, business-advice limits and account controls before relying on it.
Evidence sources
The review starts with official product evidence, then uses wider AI risk and companion-chatbot research to decide what users should verify.
Official product evidence
Official websites, public product pages, visible policy links, app-store style claims when relevant and the Fellow AI directory submission record.
Trustworthy AI and risk guidance
NIST trustworthy AI characteristics and the NIST Generative AI Profile shape the safety, transparency, privacy and accountability checks.
Companion-chatbot risk research
Mozilla privacy research, FTC attention to companion chatbot risks and Common Sense Media teen-use research inform the privacy, age and boundary checks.
Use the index with the directory
Read the score beside the tool overview, the official product page and the Safe AI Companion Checklist.