AI developers target workplace data and Wall Street alt-data to train models

What's Happening

• Artificial intelligence companies are acquiring internal corporate data, including Slack messages and memos, to train models on workplace tasks, as BI did a deep dive on the trend.
• A fast-moving market has emerged for this proprietary information, featuring exploding offers for corporate datasets.
• Robotics developers are paying human workers to record everyday tasks like cooking and cleaning to generate fresh training material.

Why It Matters

AI models require deep context to automate complex tasks. The shift from scraping public web data to purchasing proprietary corporate archives creates a new revenue stream for businesses with extensive records. It also provides a lifeline for alternative data vendors who previously struggled to expand beyond hedge fund clients.

The Context

• Wall Street hedge funds popularized alternative data, such as satellite imagery and credit card transactions, in the mid-2010s to gain trading advantages.
• Data providers hit a ceiling in the financial sector because selling the same datasets to multiple funds means buyers lose your potential edge.
• Law firms possess detailed archives and are already leveraging their databases to build specialized AI tools.

What They're Saying

• If AI tools are going to automate work, they need a deep understanding of what the work is, according to the report.
• Former Point72 executive Kirk McKeown launched a data startup last year to capitalize on this specific market opportunity.

What to Watch

• Monitor alternative data providers as they pivot from selling real-time market signals to hedge funds toward offering a wider, deeper catalog to AI developers.

The Bottom Line

The AI industry's appetite for training data is transforming internal corporate communications and Wall Street alternative datasets into valuable commodities.