Reducing Investor Meeting Prioritization Turnaround to Under One Hour with AI

Industry Overview

Investor Relations teams attending capital markets conferences need to decide which institutions offer the strongest meeting opportunities. Attendee information, however, can arrive in different formats and naming conventions, making manual comparison against investor data slow and difficult to standardise.

The Problem

Before each conference, Investor Relations Officers spent one to two days manually reviewing attendee lists and cross-referencing institutions against available data.

Institution names appeared in multiple variations across PDFs, Excel files, and email bodies, making consistent matching difficult.

Prioritization also relied heavily on judgement rather than quantitative factor analysis. In some cases, the manual process was slow enough that reports were not ready before the conference began.

Our Role

Ganit automated the investor meeting prioritization process so attendee lists could be converted into ranked, data-backed investor recommendations within an hour, while retaining a lightweight human review step for accuracy.

Methodology

The system automatically polls the inbox every 20 minutes and triggers the prioritization workflow when a new attendee list arrives.

Claude Sonnet through Amazon Bedrock, combined with AWS Textract, extracts institution information across different document formats and normalizes institution names.

A weighted matching algorithm then compares institutions against the Fitzcores universe while accounting for aliases, abbreviations, and naming variations.

Each matched institution is evaluated using a 19-factor scoring engine, consisting of 13 growth factors and 6 value/leverage factors.

A final 30-minute human-in-the-loop review validates low-confidence matches before the prioritized investor report is generated.

How It Works

Email Parsing → Attachment Storage → Institution Extraction → Matching & Normalization → Data Enrichment & Scoring → Report Generation

The end-to-end pipeline connects incoming email and OCR/Claude-based extraction through institution matching, data enrichment, and final report generation.

Our Impact

The solution achieved approximately 99% extraction accuracy and reduced the complete investor prioritization workflow to under one hour.

IROs can now receive a ranked, data-backed report shortly after an attendee list arrives. The previous one-to-two-day analyst process is replaced by automated extraction, matching, enrichment, and scoring, with human review retained where confidence is low.