Proxy Contests being evaluated by AI models

When I first started the site I wanted to have good data that would help me, as an individual investor, vote my proxies and I did in fact vote all my proxies during 2025. As the vast majority of these are regular annual meetings, the data focus on the site was on executive compensation and benchmarking and board nominees. This year I've put new focus on proxy contest analytics and used GenAI tools to come up with an analysis for each proxy contest. The first proxy contest added was Lululemon $LULU and subsequently the site was updated to automatically add new proxy contests and as a result Jack in the Box $JACK and Warner Brothers $WBD were added last week.

The AI analysis requests an analysis of each proxy fight from the four major LLMs. The prompt is the same to each and asks the model to review both the activist and management materials, come back with an analysis, and a confidence score. Each of these requests is done individually, with no analysis sharing between models. Then Claude does a consensus score and review, identifying points of alignment and disagreement.

The reason all this is being done is under the same mission as my original focus - helping investors with their proxy voting decisions. While there are Institutional Proxy Advisor firms that provide analysis (paid) to Institutional Investors, no such service is available to individual investors. Although it's early days for AI analysis of proxy contests, with just three covered contests on the site, I like what I'm seeing and the quality of the analysis looks high.