Kickstart your automation journey with faster ROIs within a specified payback period and reduced TCO
What is RPA: Robotic process automation (RPA) is a software technology that makes it easy to build, deploy, and manage software robots that emulate human actions interacting with digital systems and software.
In simple parlance, RPA is automated code for repetitive tasks to enable digital transformation and improve efficiency while reducing costs.
“Automate repetitive tasks without human intervention- How? Create intelligent chatbots for automatic workload management and for human interaction where needed. Use a batch process for automating workflows and tasks”
The Foxolabs Mantra to be successful: Integrate RPA with AI and IoT to turn data efficiency into dollars.
Can RPA replace legacy systems:
Yes, Automation, processes, invoices, payroll, reports are few things which RPA has accomplished at multiple organizations. Virtually any high-volume, business-rules-driven, repeatable process is a great candidate for automation—and increasingly so are cognitive processes that require higher-order AI skills.
These are some of the common use cases using RPA:
All repetitive and recurring processes can be replaced by RPA. RPA automation can help in various processes like customer orders and billing(purchase, invoice, payment etc.). Customer onboarding, extracting data from PDFs, generating mass emails, invoices, L1 tech support, bank statements, daily P&L. Procure to pay, extracting data from different formats, CRM systems, automated testing-arena, bank statements and PnL preparation are some common RPA daily life workflows.
Leaders of functional areas from finance to customer service to marketing to human resources and beyond find that RPA improves many processes, yielding higher capacity, faster throughput, and fewer errors for key processes.
“Real life example: While RPA can be used to digitize or automate repetitive processes, it can also be used to sift through large amounts of data in order to provide relevant answers using ML”
How Foxolabs implementation proved to be a game changer:
Website scraping-It is automation. You can pull the content from various websites automatically with all the links to subsequent pages on the main page.
But now you want to search through these thousands of pages. That is where AI, BOTS and RPA come together to be a game changer.
How many of you have heard of Pulse app for android?
It lets you store all of your favorite magazines, blogs and newspapers in one place. Once you install it, you have to choose topics of your interest and you can swipe through tons of headlines and articles and share them via Facebook or Twitter. Pulse loads the stories you read so you can read them later offline.
AI and CHATBOTS
= How about creating an AI based personalized self-learning assistant or chatbot called Akinator that can search, refine, analyze and provide the precise answer to your question. This has changed the way we solve problems.
BERT can search within documents
= It can search within emails and attachment
Is there any help available for search?
Yes, BERT from Google. Bert(is basically an NLP system and ML model which uses LSTM (Long Short term memory) was trained for billions of words from Wikipedia and other sources to make it smart and can outperform humans in certain cases. You do not have to read a 500 pager mortgage document or 500 pager claims document or 1000 pager document to find who did what. Bert will search based on the question I have asked and provide an answer in 3-5 seconds. Google just announced last week of October that they started using BERT in google search-one of their biggest updates in the last 5 years.
If BERT answers are not good, you can manually correct it and it learns since it is a self learning BOT.
“No hard feelings Microsoft and Amazon but there is a reason Google is a verb in Webster dictionary-so you can understand the impact. Google creates something and when they get bored with it, they release a white paper and the technology to the world“
Our AI based self learning BOT Assistant AKINATOR
(combination of web crawler and deep learning search models along with various questions and answer datasets it is trained on) searches through various sources to get the articles and then runs very powerful deep learning models like BERT from Google to find precise answers for you.
So what did we do
=BERT is a general model and we customized it for question and answering tasks.
How did you train BERT
= We used various data sets like Corpus and World Bank open data and Genome Dataset and Stanford question and answer dataset. Then we saw similar questions and built our own deep learning network.
So we feed back the learnings from machine learning algos or manual answers to our Akinator BOT, it becomes smarter each time. RPA combined with ML and supply chain helped Macys improve their supply orders. ML can correlate the sales process RPA to supply chain and prescribe what actions are needed to improve sales. ML can predict the sales for winter clothes based on weather forecast and past data to help Macy’s with their inventory and order which can be automated as a recurring process.
RPA is not AI; AI is not RPA.
But RPA technology now makes it possible to insert advanced AI skills in the form of machine learning models, natural language processing (NLP), character and image recognition, and more into RPA robots.
Understanding documents including semi-structured or unstructured data.
Comprehending speech and carrying on conversations and chats
Next time, we shall give another live use case of how Foxolabs built Voyager – a speech and text algorithm for conversations and chats which is used by banking professionals, back offices and certain embassies for visa and citizenship interviews.
Enterprises in industries ranging from financial services to healthcare to manufacturing to the public sector to retail and far beyond have implemented RPA in areas as diverse as finance, compliance, legal, customer service, operations, and IT. And that’s just for starters.