RBC Capital Markets

Producing a post-earnings research note took analysts upwards of 45 minutes due to a highly manual workflow: sourcing announcements, extracting key figures, writing and formatting notes, then publishing. With dozens of companies to cover, this created a significant bottleneck. In a market where a 30-minute delay can mean missed opportunities, the model was unsustainable.

RBC Capital Markets built Aiden QuickTakes, a proprietary AI application on Databricks' unified cloud-native platform. When a company announcement goes live, the system automatically ingests the release, extracts financials, and produces a first-draft research note within minutes. Analysts then refine and add insight before publishing, maintaining human-in-the-loop control throughout. The platform uses Databricks Unity Catalog for governance and Mosaic AI for end-to-end AI capabilities.

QuickTakes reduced research turnaround times by 20–60% based on preliminary results, with the full workflow dropping from 45 minutes to approximately 15 minutes from announcement to publishable draft. The platform reclaimed thousands of analyst hours for higher-value activities and enabled Global Research Analysts to scale company coverage while maintaining quality standards.

• Human-in-the-loop design is critical in regulated capital markets: AI accelerates the process but analysts retain full editorial control and expert judgment.
• Enterprise-grade governance (Unity Catalog) is a prerequisite for deploying generative AI in regulated financial environments where trust and transparency are non-negotiable.
• Automating repetitive workflow steps (ingestion, extraction, drafting) unlocks analyst capacity for deeper commentary and client dialogue without sacrificing output quality.