How does Quotewise use AI responsibly?
Quotewise uses AI for retrieval and curation, never generation. Every quote in our database comes from a documented source: books, speeches, social media posts, interviews. We find quotes. We never fabricate them.
AI assists human judgment but never replaces it. Semantic search uses vector embeddings to find quotes by meaning rather than keywords. Verification agents cross-reference sources and calculate attribution confidence scores. Quoteworthiness scoring helps surface signal from noise in our hundreds of thousands of quote corpus. And human curators still make final editorial decisions on what belongs in collections.
We're aware that quote databases carry the biases of history. Dead white men dominate because they wrote more books, held more speeches, got more press coverage. We address this through active curation of contemporary voices, including tweets, social media posts, and modern thinkers before the book deal, plus community contribution where anyone can submit wisdom regardless of credentials.
Quotewise will never include AI-generated quotes in the database. If AI hallucinated it, it doesn't belong here.
What is the Zero-Training Pledge?
Quotewise will never use your interactions to train AI models.
This covers:
- Search queries (semantic or keyword)
- Conversations with QuoteGPT or AI agents
- API requests and responses
- Collection creation and curation
- Quote submissions and edits
- MCP tool calls from Claude or other AI assistants
Your searches stay private. Your conversations stay private. Your data is for your use.
If this policy ever changes, we will provide public notice with a deadline before any change takes effect. You will have the opportunity to export your data and delete your account before any training use begins.
This is a firm commitment. We believe trust requires constraints you can rely on.
How does Quotewise protect user data?
All connections are encrypted with TLS via Cloudflare SSL in Full Strict mode. Our database runs on Aurora PostgreSQL with encryption at rest, hosted in AWS us-west-2. Login uses OAuth 2.0 through GitHub and Google, so we never see or store your passwords. We collect as little data as possible and measure what people do, not who they are.
For analytics, we use PostHog to track search patterns, feature usage, and errors, but we don't follow individual users across sessions. We know that 20% of searchers save what they find. We don't know which ones. We don't want to.
There are no cookies, no consent banners, no behavioral profiling. We don't sell data or share it with advertisers.
GDPR compliance is built in. EU-hosted servers where GDPR applies by default. We comply with privacy regulations because we never collected the data they would restrict.
What are Quotewise's content policies?
We prioritize verification quality over database size. A smaller corpus with reliable attribution beats a massive collection where half the quotes are misattributed.
When we discover a quote is misattributed, we correct it, whether or not anyone reported it. QuoteSightings document the evidence trail so you can verify our reasoning. Disputed attributions are flagged, not hidden.
Found a misattribution? Submit a correction with evidence. We review all reports and update attribution when the evidence warrants it.
Our attribution methodology is public. Attribution confidence scoring (launching 2026) will make our confidence level visible on every quote. Critics can find the holes before users stumble into them.
We don't say "verified" and expect you to trust us. We show you the evidence chain through QuoteSightings and let you decide. Verification is a process we're building, not a state we've achieved.
Quotewise includes quotes on controversial topics and from controversial figures. We preserve historical context, not sanitized comfort. Offensive language, hateful ideology, and difficult history are documented because they were said. Context and source attribution are how readers evaluate them.
How does Quotewise work with AI crawlers?
We use a few structured signals to tell AI systems and crawlers what we're about and how to interact with us.
AIPREF meta tags declare our preferences for how AI systems should use our content (index, nosnippet, etc.). Compliant crawlers respect these signals. Our llms.txt file is machine-readable documentation of what Quotewise does, our API endpoints, and usage policies. The traditional robots.txt handles search engines and well-behaved bots. We allow indexing but ask for rate-limiting so the servers stay happy.
Every quote page also includes schema.org JSON-LD markup, giving AI systems structured access to quote text, attribution, source documentation, and the relationships between quotes, originators, and collections.
Why bother with all this? AI systems that understand our data structure can verify quotes more accurately, attribute them correctly, and surface them in context. These signals are how we avoid becoming yet another source of AI hallucinations.
How can I contact Quotewise?
Drop us a line:
- General inquiries: [email protected]
- Misattribution reports: [email protected] (include evidence: links, citations, page numbers)
- API/developer questions: [email protected]
- Media/press: [email protected]
We're also on GitHub. Join discussions on Issues for our public repositories. We welcome contributions, corrections, and challenges to our methodology.