What is cognitive automation?

How does cognitive automation work?

Cognitive automation integrates RPA technology with AI technologies like machine learning, natural language processing (NLP), optical character recognition (OCR) and computer vision to automate complex tasks that normally require human cognitive capabilities like thinking, analysis, judgment and creativity. Systems that combine intelligent technologies with RPA can consume and analyze vast amounts of complex data from multiple sources and take actions (or generate meaningful insights) that can improve human understanding and decision-making for complex scenarios.

Some cognitive services are customized, rather than designed from scratch, for certain organizations. This enables those companies to provision and use the specific capabilities they need to benefit their business. Such customizations usually require specialists with the expertise to ensure that the services match the company's needs, are resilient and secure, and incorporate specific requirements, such as those related to data privacy or regulatory compliance.

Larger organizations often set up a dedicated, in-house automation center of excellence to manage more sophisticated cognitive services. The CoE drives the firm's automation initiatives while working to ensure alignment between those initiatives and the firm's goals. They understand the best practices for scaling cognitive automation systems, improving user experiences and reducing the business risks of implementing those systems, such as unexpected costs or increased technical debt.

What are the benefits of cognitive automation?

Among the biggest benefits of cognitive automation is that it supports the automation of complex tasks that required human inputs or interventions, and could not be automated by RPA. Cognitive automation systems can consume, process and analyze vast amounts of unstructured data from multiple sources -- documents, emails, images, customer interactions, voice interactions, social media, etc. -- for a wide range of applications. Such automations eliminate common errors and inconsistencies due to human problems like lack of knowledge, fatigue, illness or poor judgment. Better results accrue for a specific application or business workflow. Cognitive systems also can learn from their experiences to continuously improve their performance and that of the underlying process or workflow.

Finally, IA can shrink labor costs. By automating certain complex or time-consuming tasks, organizations can reassign and optimize the human resources previously engaged in these tasks. Those employees can focus on higher-value activities to increase ROI.

What are examples of cognitive automation?

Numerous terms are in use today to describe cognitive automation, including the following:

• Intelligent process automation.
• Intelligent business process automation.
• Digital process automation.
• Hyperautomation.
• Cognitive services.

Here are a few examples of cognitive automation:

• Combining intelligent data capture with process automation using OCR, machine vision, speech recognition or natural language understanding.
• Automating process workflows and decisions using AI business rules engines to complement or replace traditional business rules management systems or business process management systems.
• Using process mining and AI tools to automate the process of identifying automation opportunities, then automatically provisioning them -- a concept called hyperautomation.
• Packaging a set of services that combine AI and automation capabilities provisioned via out-of-the-box or customizable APIs, software development kits and AI models. Azure AI Services is an example of such cognitive automation services.

To use intelligent automation capabilities for advanced tasks, cognitive automation systems and tools may incorporate any of these technologies:

• Machine learning.
• Deep learning.
• NLP.
• OCR.
• Predictive analytics.
• Intelligent document processing.
• Voice recognition.
• Emotion recognition.
• Sentiment analysis (opinion mining).
• Data mining.

What are the uses of cognitive automation?

Cognitive automation supports many processes that can benefit from extending RPA with AI. The RPA/AI integration enables cognitive automation systems to analyze data and automate more complex tasks, such as those that typically depend on human knowledge and reasoning.

Some good candidates for cognitive automation include the following:

• Virtual assistants. Chatbots enabled with AI technologies like NLP can understand and process human voices and text to process complex queries and provide personalized responses or recommendations.
• Product categorization. This involves automatically categorizing product data from various sources into one global set of structured data, which is important for improving product discoverability in e-commerce. Shoppers can then locate products quickly, which can increase product sales.
• Information technology service management (ITSM). Cognitive automation systems can streamline many ITSM tasks, such as incident response and management, service desk automation, security analysis, system monitoring and vulnerability management.
• Accounts payable. Information from differently formatted invoices can be copied into a standard format and then loaded into an accounting system, minimizing the need for manual entry and checks, and speeding up the accounts payable process.
• Customer service. Customer or support data can be retrieved automatically in response to an ongoing service call using speech recognition and natural language understanding. This helps call center agents to have more meaningful conversations, which can improve customer experiences and advance consideration of cross-sell or upsell offers.
• Employee onboarding. Many onboarding tasks that usually require HR efforts can be automated. Examples include creating login credentials and enrolling new participants in the onboarding program, enabling HR staff to focus on more important tasks.
• Customer onboarding and customer relationship management. AI-enabled systems automatically capture customer information that can help sales and other customer-facing teams improve customer engagement. The systems may also include communication features and capabilities like contact management and analytics that support the needs of sales, marketing, customer support and other processes.
• Recommendation engines. AI can generate personalized recommendations for customers based on information inferred about their intentions. This streamlines the customer experience. Combining RPA bots with conversational AI chatbots or virtual assistants can yield further improvements.
• Regulatory compliance. The focus here is tracking regulatory changes, identifying and flagging activities that may increase noncompliance risks, and generating compliance reports automatically. This saves time for compliance teams, helps them close compliance gaps and ensures that the organization adheres to applicable standards, laws and regulations.

What are the challenges of cognitive automation?

Perhaps the most significant challenge of cognitive automation is that it requires customization and integration work specific to each enterprise. This is less of an issue when the services are only used for straightforward tasks. More sophisticated automations require extensive planning, customization and ongoing iteration for optimal results. Smaller companies may not have the wherewithal, expertise or funds to tackle these tasks, which may hinder them from deploying cognitive automation and enjoying its benefits.

Other potential challenges include the following:

• Slow path to positive returns on the technical investment.
• Difficulty finding experts with automated business systems experience.
• The need to vet AI algorithms to minimize the potential for bias and AI hallucinations.
• Potential security issues arising from the system accessing a wider range of IT systems and workflows.
• Possible privacy or compliance breaches from feeding sensitive data, such as personally identifiable information, into cognitive automation workflows.

What are the differences between RPA and cognitive automation?

Simple RPA (RPA without cognitive automation) uses software robots (bots) on a computer to mimic and replicate human actions, such as data entry, website scraping, simple data analysis and automated help desk support, by following predefined rules. Unlike cognitive automation, RPA focuses on automating only repetitive or simple manual tasks, which is why it is not considered an intelligent technology.

Unlike RPA bots that need to be programmed with specific rules, cognitive automation systems do not need to be explicitly programmed. They go beyond simple rule-based operations to learn what to do and how to do it using the data they consume. These intelligent capabilities allow them to adapt and adjust to produce increasingly better outcomes for a specific application or use case. This ability to learn and adapt on the fly also allows these systems to retrieve meaning and context from ingested data, perform analyses to produce useful, actionable information, and even make decisions independently.

Here is a brief summary of the key differences between cognitive automation and RPA:

• RPA automates repetitive actions, while cognitive automation can automate more types of processes, even complex processes that typically require some human input or intervention.
• Traditional RPA only works with structured data, which limits its use cases. Cognitive automation can process both structured and unstructured data to support more real-world applications.
• RPA is simple to set up and manage, and often yields tactical wins and ROI quickly. Cognitive automation usually requires more time to set up the infrastructure and workflows and some additional management overhead, but this usually translates into a strong strategic long-term advantage.
• RPA bots are explicitly programmed to perform a certain task a specific way. Cognitive automation systems don't require rigid programming to learn the intent of a use case. They can learn and adapt over time to continuously improve their output and decision-making capabilities.

CIOs rely on cognitive automation and RPA to improve business processes more than ever. Explore the important distinctions between RPA and cognitive automation.