Artificial intelligence is changing the game in many industries and now in automation tools.
Companies can no longer ignore the digital transformation of their competitors and customers. This digitalisation implies being master of one’s own processes and their state of optimisation. Moreover, artificial intelligence is the new revolution that is changing the operational model of companies and the way they deliver their products and services to their customers.
The combination of different automation technologies at the service of the business, combined with AI technologies, has become a vital necessity. Vendors of IPA solutions have perfectly understood this issue since they are integrating more and more AI in their solutions. To understand the value of automation through AI, it is important to know the contribution of each technology in the automation process pipeline.
- Intelligent Process Automation in general (IPA)
- Task & Process Mining (TM/PM)
- Robotic Process Automation (RPA)
- Business Process Management (BPM)
- Intelligent Automation (IA)
- Intelligent Virtual Assistant (IVA)
- Intelligent Document Processing (IDP)
- An example!
- Finally…
Quadrant Knowledge Solutions defines “Intelligent Process Automation (IPA) as solutions and services that combine robotic process automation (RPA) with technologies such as process mining, artificial intelligence (AI), intelligent character recognition (ICR), optical character recognition (OCR) and advanced analytics.”
IPA solutions and services are also known as hyper-automation, intelligent automation and digital process automation (DPA).
They leverage RPA, BPM, AI, OCR, ICR and task and process mining to improve business operations through end-to-end automated business processes.
IPA’s solutions and services deliver value to customers through their ability to learn autonomously and act at the right time using contextual information gathered by the analytical tools.
IPA also offers customised components for data mining, content processing and decision making models to deliver smarter business processes.
Process Mining is an approach that enables the discovery, monitoring and optimisation of business processes by analysing the event logs generated by information systems (ERP, CRM, etc.)
Task Mining allows the same operation as Process Mining to be carried out but by recording the execution of tasks directly on the users’ workstations, thanks in particular to artificial intelligence (AI).
These technologies will enable the automatic discovery of the company’s processes and operating rules.
The advantages: automatic formalisation of paths, process monitoring, assistance in implementing digital twins.
These task mining techniques can be good introductions to RPA by enabling the discovery, formalisation and monitoring of user operations. AI is used widely at this level for the recognition of elements manipulated by the users. AI-driven analysis techniques are used also to predict the time and cost impact of automatically identified opportunities to improve the workflows.
These solutions collect data (event logs or screenshots) by referring to constant identifiers such as a user name, a folder ID, etc. This data is then analysed to generate a report on the user’s behaviour and help to generate process maps.
Use cases for exploration:
- process discovery,
- compliance checking,
- process improvement,
- change simulation,
- supervision.
Robotic Process Automation is a technology for automating repetitive activities of users.
These technologies are non-invasive and typically interact through the user interface of software applications.
RPA scenarios range from creating a single automatic response to an email, to deploying thousands of bots, each programmed to automate work on one or more systems.
RPA-eligible processes include those that involve a lot of repetitive human handling, such as sorting emails, downloading attachments, entering values from a file into a business application, etc.
Eligibility :
- existing processes,
- volumetry & duration,
- manual entries,
- standard vs. exception,
- time slots.
RPA solutions are generally equipped with a graphical studio to record the work of a user on his workstation, an editor of the recorded processes and a platform for launching and supervising the robots.
RPA solutions come with additional functionalities: rules engine, document understanding (invoice, order…), IDP and Intelligent Automation tools like image recognition, email classification, language detection, etc.
Mature RPA solutions have evolved into complete IPA platforms.
Business Process Management (BPM) offers an overview of business processes: their organisation, their interactions in order to optimise and automate them as much as possible.
BPM is generally associated with the BPMN formalism and the monitoring of processes through BAM.
BPMN (Business Process Model and Notation) is a business process modelling method for describing the value chains and business activities of an organisation in the form of a graphical representation.
BPMN provides a notation that is understandable to all automation stakeholders: business analysts, designers and developers who will be responsible for implementing the automated processes.
Business Activity Monitoring (BAM) involves the real-time acquisition, aggregation, analysis and presentation of data associated with business processes.
The BAM dashboard displays key performance indicators that summarise the health of key business activities.
A BPM platform includes :
- a BPMN and/or rules modelling capability,
- a process repository to manage the modelling metadata,
- a process execution engine,
- a state and/or rule management engine.
Example of a BPMN diagram for a purchase request business process passing between the purchasing and accounting departments
The BPM allows the transmission of information to be automated and facilitates the approval of the request by purchasing.
Like RPA vendors, AI-driven tools also are part of the BPM solutions now.
Intelligent automation (IA), sometimes also referred to as cognitive automation, is the use of machine learning (ML) techniques and natural language processing (NLP) technologies to streamline and adapt decision making in organisations.
Machine learning and complex algorithms can be used to analyse structured and unstructured data.
This allows companies to develop knowledge bases and make predictions from existing data.
The notions of algorithm explainability and sustainability (green computing) are to be taken into account in AI-based approaches, especially when AI is involved in a critical system.
Examples of use cases :
- prediction of failures,
- visual analysis of welds,
- recommendation of the next best action,
- maintenance assistance based on documentation,
- automatic responses to emails,
- resume summary,
- applicant correspondence,
- demand forecasting,
- diagnostic assistance,
- fraud detection,
- route optimisation.
Intelligent Virtual Assistants (IVAs) are computer programs that interact directly with a user. In this category are conversational assistants, which are computer systems designed to interact with users in real time via a natural conversational interface, such as speech or text. These solutions are often referred to as Chatbots. IVAs use NLP techniques powered by ML and NLP technologies. The most talked about conversational assistant at the moment is ChatGPT from OpenAI.
Conversational assistants can be used to perform a variety of tasks, such as answering questions, searching the Internet, scheduling appointments, controlling connected devices, etc.
Conversational assistants are generally based on different technologies, such as speech recognition, speech generation, natural language understanding and neural networks and deep learning.
In particular, they can be deployed on different channels:
- mobile applications,
- social networks,
- websites,
- connected boxes,
- reception terminals,
- metaverse.
Conversational assistants are increasingly used to improve the user experience, automate repetitive support tasks and provide self-service solutions to users.
To date, there are several hundred solutions that allow users to configure their own assistant.
On the Internet, the user interface has become relatively standard, making it easier for users to access.
The use cases for Conversational Assistants go beyond customer service and now include areas such as HR support, IT helpdesk, sales, marketing and supply chain/purchasing…
Intelligent Document Processing solutions are any product or software solution that captures data from documents, categorises it and extracts the relevant data for further processing.
Examples of documents include:
- email,
- business document,
- contract,
- invoice,
- order,
- resume…
Processing is based on computer vision, optical character recognition, and natural language processing. The processing steps consist of pre-processing the captures (image correction), classifying the document, extracting the data, and validating the extracted data. At the end of this processing, the product or software solution returns structured data that can then be easily processed by a computer program, RPA script, automated process, etc.
The processing may require the intervention of a human to validate or correct the extracted information. This processing is called “human-in-the-loop”.
Chapati receives several thousand emails a day. Several people in the company are responsible for sorting them, processing the orders and then answering questions about the status of certain previous orders or about the products. These employees then take each email containing an order, open the completed and scanned form and, according to the list of designated products, carry out checks, make requests in the company’s various tools and then respond to the requesters. Some requests require validations and the employees call the decision-makers to obtain the various confirmations.
This company wanted to relieve its employees of this tedious data entry work so that they could be entrusted with more rewarding activities such as helping customers. The Chapati company therefore embarked on a transformation project. As he did not control all the stages of the various processes, the transformation manager used automatic task exploration by placing agents on the workstations of the people in charge of sorting and answering. The resulting map, using task mining (TM), was then transformed into a graphical representation containing activities and associated sequences (BPM). As there is no API on the internal applications of the company, RPA scripts are implemented to simulate the input work of the employees. Depending on the characteristics of the ordered products, the system automatically asks for validation from the decision makers (“human-in-the-loop”).
In the business workflow, an e-mail connector is used to retrieve requests and an AI-based subroutine is used to classify these requests (request classification). Based on the result of the sorting, orders are retrieved and the information contained in the scanned documents is extracted using intelligent document processing (IDP). Finally, an intelligent virtual assistant (IVA) automatically answers any remaining questions from customers.
The former sorting staff now perform customer advisory operations, thus enhancing their knowledge of the company’s internal operations and product knowledge.
It is no longer possible to ignore the digital transformation of consumers and, by extension, businesses. For companies, this transformation means having control over processes and their optimisation. Artificial Intelligence is the current revolution that is changing the operational model in depth and the way products and services are delivered to customers. The combination of different technologies, at the service of the company, has become a vital necessity.
However, it is important to first define the company’s ambitions: revenues, costs and risks. This also requires a thorough understanding of the company’s processes, their structure, their level of optimisation and the corrections that need to be made before any optimisation.
The contribution of IPA is to provide an organisation and a technical framework for assembling all the automation technologies to serve the business processes.
When thinking about automation, it’s important to first think about understanding the business processes, break those processes down into activities, and then determine which IPA components are best suited.
The landscape of automation solution providers is changing as they become AI-enriched suites. They are increasingly incorporating models based on machine learning approaches. It is therefore no longer possible to simply think in terms of RPA, BPM, etc., but also to understand the contributions of AI and the biases carried by these technologies.
