Wise.io

Wise.io provides predictive intelligence to help your customer support agents be more productive and focus where customers need them most.

“Mutually Human helped us kick start our application.”

– Joshua Bloom, CTO, Chairman & Founder

Background

Wise.io’s Support application uses machine learning to help companies rapidly handle customer service requests, lower response times and allow service agents to focus on solving customers’ issues.

Their algorithms handle tens of thousands of support requests a day; they must store and classify them in real-time, and retrain data models on the fly to respond to changes in the types of requests each client recieves.

Wise.io hired Mutually Human to evolve and grow the Wise Support web interface to keep pace with their growing business.

Approach

Because Wise.io came to us with an existing codebase, we began with some code archeology, delving into the code and its history to understand how it worked, and how it had changed over time. This work had to be done while the code was still under active development, adding features and fixing bugs to keep time with the business’ needs

Having understood the business needs, and the code that currently served them, we were able to start improving and fortifying the foundations, preparing it to handle increased load.

We also provided guidance and polish to Wise.io’s development process, allowing them to continue developing the app at a steady, sustainable pace.

Technology

“Code quality was consistently excellent, laying down a solid foundation for us to grow our application suite.”

Wise Support’s API was built in python, using Flask. Python was the language most familiar to the development team at Wise, so it made for a more comfortable hand-off of maintainence and continued development.

The frontend was built in AngularJS. Angular gave us the ability to make an extremely fast, dynamic interface, that had the flexibility to meet new customer needs as they surfaced.

All Wise’s infrastructure was hosted on Amazon Web Services (AWS); we hosted the frontend on S3, served with cloudfront, allowing it to load in milliseconds. The API was hosted on Amazon’s Elastic Beanstalk, allowing it to dynamically scale in response to changing load.

The Wise Support API interacted with multiple sources of support tickets, including Zendesk and Salesforce.

Challenges

Pretty Big Data

Wise Support took in a vast amount of data, and it needed to serve that data quickly and reliably. Rapidly sifting through millions of rows of data meant pouring over postgres query analyses, creating complex indexes, knowing when to normalize and when to denormalize the data, and building creative queries.

Respecting Others’ (API) Limits

Support tickets were created in a steady stream all day; and they were updated at a rate faster than the Zendesk and Salesforce API’s allowed us to fetch them. We built a safe, repeatable, and reliable batch processing system that ensured Wise’s data was always up to date, without violating the limits of any of the CMS APIs.

Staying Nimble

Wise Support was still a relatively new product; Wise.io’s understanding of their clients’ needs grew on a daily basis. Our process, and our software, helped us meet those new needs as quickly as possible.

Success

“Mutually Human helped us kick start our application building and SaaS integrations, bringing programming and organizational best-practices to the table.”

Wise.io was awarded the 2016 Product of the Year by CUSTOMER Magazine.

 

Technology
AngularJS
Backend
CoffeeScript
CSS
Frontend

 

Skills

Posted on

March 30, 2018