Unlocking Document Intelligence In Capital Markets with Box Skills and Alkymi
The following is a guest blog post by Alkymi. Alkymi applies artificial intelligence to help financial services companies automate manual tasks on documents that require human comprehension. Follow them on Twitter and LinkedIn.
Despite years of digitization, documents still underpin most financial services workflows, with employees using ‘copy-paste’ to capture valuable financial and client data in business processes. The result has been that financial services companies — banks, asset managers and insurance firms — spend billions on manual document processing that creates stubborn barriers to automation. This means that front-office decision makers are missing insights that generate an edge, when transactions, risk, and relationships are locked within documents.
The combination of Box Skills and Alkymi targets this challenge by turning documents into data, helping enterprises with data extraction and document automation.
Background: Using Alkymi and Box Skills to Understand Equity Research
Alkymi enables document automation for financial services companies across a variety of workflows and unstructured data such as email, PDF, excel, and scanned images. For this blog post, we’re demonstrating the benefit of integrating Box Skills with Alkymi for insights on equity research, a familiar data set if you’re dabbling in stock-picking. Sell-side firms publish millions of equity research reports annually, which means buy-side recipients only open and read less than 5% of these reports due to information-overload and overflowing email inboxes. This gap represents lost opportunity for insights that drive ideas and trading strategies for analysts and portfolio managers.
Alkymi and Box Skills solves this problem with machine learning algorithms that instantly transform equity, credit, commodity, and economic research reports into structured data that can be used for analytics (Figure 1).
Diving deeper: Developing with Box Skills
When equity research documents are uploaded to a Box folder, the Alkymi platform generates tags in real-time, using Box Skills to populate topic cards with factual information from the document (Figure 2).
When receiving a new document, Alkymi’s Django backend starts an asynchronous Celery task to analyze the document and generate the relevant topics. Once complete, this data is pushed back to Box via an Express.js app that uses the Box Skills Kit API.
Here’s a snippet of our Express.js app with an endpoint for receiving the topics and the context of the uploaded document. It then pushes those as topic cards to Box and returns a success message back to the Alkymi platform.
const express = require('express')
const { SkillsWriter } = require('./skills-kit-2.0');
const app = express();app.post('/push-card', (req, res) => {
console.log('Got event from Alkymi and forwarding to Box')
const skillsWriter = new SkillsWriter(req.body.fileContext); const cards = [];
cards.push(skillsWriter.createTopicsCard(req.body.topics));
skillsWriter.saveDataCards(cards);
res.status(200).send('All Good!');
});
Check out this demo to learn more about Alkymi, or get in touch for a live demo on your data.