Canary Data launches Stanley, an AI agent for cyclical-stock analysts

Canary Data, the AI software startup founded by former short seller Joe O'Donnell, has released a new agent called Stanley that scans markets for mispriced cyclical stocks and produces chart-filled investment memos, according to The Wall Street Journal.

The agent is named after Stan Druckenmiller. O'Donnell told the Journal that Canary's software is designed to perform in hours the kind of work that used to take him days or weeks as a professional analyst.

The pitch is workflow compression at the idea-generation stage: screen the universe, flag candidates, draft the memo. Cyclicals are a defensible starting wedge because the analytical pattern is more repeatable than, say, a special-situations or distressed name. Earnings revisions, inventory cycles, commodity inputs, and historical multiples are the standard inputs an analyst would run anyway.

O'Donnell's positioning, that he is automating his own past job, is a familiar founder narrative in this category. It is also the most credible version of it: a former short seller knows exactly which steps in the workflow are mechanical (pulling comps, building the chart pack, drafting the bull-bear summary) and which are judgment (deciding whether the setup is real, sizing the position, defending it to a PM).

Canary enters a crowded field. The Journal names Anthropic and OpenAI as horizontal model providers chasing financial workflows, AlphaSense as the incumbent financial-software player, and a set of fellow fintech startups all targeting the same analyst seat.

What to watch next: pricing disclosure, the size of the design-partner book, and whether Canary publishes any backtest or accuracy framing for Stanley's mispricing calls. None of those were disclosed in the Wall Street Journal piece.