Automation began with simple rule-based operations around structured data. Built on databases, enterprise solutions operate with structured data. Most of the standard workflows are already preprogrammed or could be scripted within the enterprise applications, such as enterprise resource planning (ERP) or customer relationship management (CRM) systems, during the setup process.
Robotic Process Automation (RPA) solutions allow adding a level of automation on top of existing solutions, transferring data across them, and tying them together within specific processes. But again, RPA rules have to be strictly defined upfront, and they only allow for dealing with structured data.
The next level of automation, however, requires additional capabilities:
- Dynamic process tracking and automation discovery. While processes are constantly expanding and changing, automation technology needs to monitor those changes dynamically and discover new processes and automation opportunities by tracking and analyzing user actions, identifying repetitive patterns, and then suggesting automation.
- Understanding unstructured data. Despite extensive use of databases for storing business data, most researchers agree 80% of data in the enterprise is still unstructured. From documents to audio and video recordings, modern companies operate and store unstructured content, which contains critical information used for making business decisions.
This content is typically processed by human workers, which often creates a bottleneck due to slow, error-prone, and often more expensive procedures. Leveraging various artificial intelligence (AI) technologies, such as computer vision, optical character recognition (OCR), voice recognition, and natural language processing (NLP), enables automatic access and transformation of unstructured data into a structured format, which could then be processed by automation solutions.
- Automatic data analysis and decision-making. Traditional automation follows simple prescribed rules and cannot adapt to changing business requirements. However, recent developments in machine learning (ML) technology and big data analysis enabled the creation of dynamic self-learning models, which can process incoming data and make intelligent decisions based on prior training. Such models are used for risk assessments, fraud detection, financial modeling, and other areas where an abundance of existing data allows sufficient training of ML models.
Intelligent automation leverages AI and ML to expand the capabilities of traditional RPA and support the use cases described above. As a result, an intelligent automation system is built on five key elements:
A rock-solid RPA platform
The RPA platform forms the foundation of the automation system. It facilitates the creation of bots, Digital Workers, and other automations through the intuitive user interface that makes it easy for a business user to build bots.
The platform should also provide a collaborative environment, ideally with an integrated development environment (IDE), that enables RPA developers to build upon business users’ bots by adding programming code, as well as by integrating AI and ML technologies.
The web-based interface in Automation Anywhere Enterprise makes it easy for business users, aka citizen developers, to create basic bots and then collaborate with RPA specialist developers who can add sophistication, such as AI modules, API integrations, security, and governance controls (see Figure 1).
2. Process discovery
An intelligent technology that can identify what processes can and should be automated, process discovery leads to process automation. It records, analyzes repetitive user actions, and then creates RPA bots to perform those actions. User actions can be followed across multiple applications, such as ERP, CRM, and other enterprise solutions, as well as Excel, Outlook, and other productivity applications.
An intelligent process discovery solution can record users’ actions while they follow the process end to end. In this way, it can identify common patterns and repetitive tasks. This data is used by the intelligent automation platform to create RPA bots automatically.
3. AI components for processing unstructured data
Artificial Intelligence is the key technology that transforms RPA into intelligent automation. And, just as the RPA platform must be robust, the AI components must be as well. More importantly, they must be designed to work together easily and seamlessly.
The AI system should be flexible and yet complex enough to address all areas of AI, including computer vision, OCR, voice recognition, and NLP.
Let’s take NLP as an example. NLP is designed to read, decipher, understand, and make sense of human language in a manner that’s useful in machine-to-human interaction. It can detect sentiment from text communications.
If a customer writes an email that says, “I need you to address this ASAP!” for instance, NLP can detect that the customer is not happy and escalate the issue to a live customer service agent — while at the same time letting the customer know he or she should hear from the agent within X number of minutes.
NLP components should be extensible — that is, easily added to the RPA system. Enterprise A2019 enables NLP modules to be incorporated into the platform (see Figure 2), where the NLP technology can be integrated into various bots and processes.
One of the most critical use cases where business data is converted from unstructured into structured form is intelligent document processing. Business documents — such as invoices, purchase orders, contracts, and shipping documents — all contain business-critical data hidden in unstructured content.
Automation Anywhere IQ Bot is an intelligent document processing solution that automatically reads documents in various formats, sorts them, extracts data from them, and converts the data into a structured format. Hence, this data becomes available for automatic processing by business applications and RPA bots, enabling end-to-end process automation.
4. Advanced data analysis and decision-making
"What gets measured gets done" is a famous quote by HP Co-founder Bill Hewlett. In the realm of intelligent automation, this is truer than ever. An advanced intelligent automation system offers analytics to monitor both automations and the business. Operations managers and other practitioners can continuously track the efficiency and effectiveness of their Digital Workforce and effortlessly quantify processes' operational performance on demand.
Implementing ML models that allow swift decision-making in typical situations enables automating even complex processes, which used to require human intervention. That brings automation to the next level, significantly growing its applications.
5. A secure, efficient yet flexible infrastructure
For maximum effectiveness, the intelligent automation infrastructure must be secure, cost-effective, and, above all, available. It must offer many different deployment options, such as on-premises (bare metal or private cloud), cloud, or Software as a Service (SaaS). It must have high availability — 99% or higher — low latency, and lightweight design.
Automation Anywhere Enterprise A2019 was architected from the ground up as a true cloud-native RPA platform. As a result, it’s more efficient, secure, and cost-effective than any RPA platform on the market. Its design enables it to meet or exceed all major security and compliance (regulatory) standards.