AI-Powered IPOs: How Wall Street Is Automating Prospectus Creation 

A quiet transformation is underway in investment banking: the use of large language models (LLMs) to generate IPO prospectuses.

Goldman Sachs and JPMorgan are among the firms that now automate up to 95% of their prospectus documentation, a task that once involved weeks of work from junior analysts.

These LLMs are trained on hundreds of past IPOs and regulatory frameworks, allowing them to instantly generate sections like company background, risk factors, and financial overviews.

Human reviewers remain essential—particularly for legal compliance and sensitive disclosures—but the bulk of the drafting is handled by machines.

The result is dramatic:

• The average timeline from mandate to filing has dropped by 30–50%
• Internal cost savings are notable

As generative AI tools mature, they are also being tested for:

• Secondary offerings
• Investor presentations
• Roadshow materials

Regulators are beginning to respond.
The SEC is rumored to be considering new guidelines that would require issuers to disclose the use of AI in preparing public documents. In Europe, ESMA is reportedly exploring traceability standards for AI-assisted financial disclosures.

While automation improves efficiency, it also raises new concerns:

• Model hallucinations
• Data integrity
• Legal liability in the case of omissions

For now, however, AI-generated prospectuses are becoming the new normal in capital markets.