How the research is made
How Synthos works
An AI research engine with a named human signing every conclusion. Every claim we publish carries its source, speaker, and date — you can always see who said what, when.
What the machine does
Synthos listens to thousands of hours of the world's most credible investors and scientists — alongside filings, fundamentals, and market data — and distills it into a living knowledge base of 53,000+ dated, attributed claims spanning 44 research fields across macro, AI, crypto, medicine, investing, and frontier technology. From that base it maintains live conviction scores, screens for what is real but not yet priced, writes research, and constructs portfolios — then logs every prediction and grades it in public when it matures. A week's new evidence moves a live score within days. The weighting machinery is proprietary; the conclusions, sources, and grades are public. Disagreement among credible voices is shown, never averaged away. Buy and Avoid, both published.
The honesty commitment
We never fabricate or inflate performance. Hypothetical figures are labeled hindsight wherever they appear, at the same size as their benchmark. When a call lags, we publish it anyway. Our own failed experiments are published reading. Every performance claim is approved by a named human.
— Ari Sagherian, founder. This standard is permanent.The real-money Flagship
The Flagship is funded with real money and tracked in public from its public launch (date to be announced) — every position and every return against the S&P 500, published up or down. Until it has history, no track-record claims are made anywhere on this site.
The accessibility commitment
Synthos was founded by a legally blind investor. This site targets WCAG AAA: dark, high-contrast surfaces (7:1 or better), resizable type with no pixel-locked sizes, full keyboard navigation, screen-reader semantics on every table and chart, and no information encoded in color alone — every gain and loss carries an arrow and a sign. Details and a feedback channel are on the accessibility page.
What this is not
Not investment advice, not personalized, not a promise of returns. The quantitative layer is descriptive, not predictive. The AI layer drafts; a named human reviews and signs. How our public posting works — including the disclosed use of AI agents — is documented on How Synthos posts.