On June 14, 2017, Automation Anywhere’s chief marketing officer, Gary Conway, and principal product manager, Kelly Coupe, hosted an interactive webinar featuring David Jones and Edna Ezzell of our partner, Accenture. Entitled “Introducing IQ Bot,” our hour-long webinar focused on how artificial intelligence-based “bots” used in robotic process automation (RPA) were being enhanced with cognitive capabilities, and what that meant for today’s businesses.
Webinar participants were encouraged to send us their questions during the presentation. We are pleased to provide the answers.
Q: Do you need a complete systems development lifecycle (SDLC) team—for example, developers, project managers, and business analysts—to successfully deploy RPA?
A: Yes, a typical RPA implementation requires companies to take a structured and methodical approach executed through a SDLC-like framework. We recommend forming three types of teams:
- Bump Team: This team is comprised of business process experts or subject matter experts whose jobs are to identify processes that can be automated.
- Set Team: This team employs solution architects who architect the best way to automate a given process and design the solution (the bots). This team also has responsibility to review the accuracy and efficiency of the bots as they move through the lifecycle from development, to test, to production.
- Strike Team. This team is made up of automation experts who take the design from the Set Team and actually automate the selected processes.
Q: How is RPA different from workflow automation or business process management tools?
A: Workflow automation and business process management (BPM) tools are used for specific business scenarios and require specific business processes to be re-engineered. They’re driven by IT and are automated using backend application programming interfaces (APIs). They also typically require you to fulfil formal change-management requirements. All of this is both time-consuming and expensive.
RPA tools, on the other hand, are general-purpose tools that can be applied to a broad range of scenarios without needing to re-engineer any processes. RPA tools operate on the front end—through an intuitive application interface—and can deliver seamless automation without any need for back-end programming or integration. Because of this, you can use RPA tools to automate all kinds of business processes, including legacy applications and applications exposed over Citrix.
Q: Given all the RPA tools on the market today, how do you select the best one?
A: There are eight primary factors that must be considered when evaluating RPA solutions:
- Product Maturity: You should always check the number of years the product has been on the market. Ask for the number of enterprise deployments—make sure there are large as well as smaller deployments—and ask for the total number of bots in production overall.
- Breadth of Portfolio: Rather than a set of disparate products, look for a comprehensive platform that enables you to do an end-to-end RPA deployment. The platform should include three (3) components:
- Core RPA capabilities that allow you to implement rule-based automation
- Built-in cognitive automation that enables your bots to intelligently handle unstructured data as well as employ machine learning
- Built-in analytics engine to continuously improve the accuracy and efficiency of processes as well as of the bots
- Ease of use: The solution should be as easy to use as a consumer product—it should be simple for non-IT professionals to create, change, deploy, monitor, and maintain bots
- Scalability: The solution should allow you to reuse any bots you create and scale them on demand. It should also deliver built-in productivity tools that accelerate development of bots, such as recorder tools, collaboration tools, and an automation library.
- Reliability: Check that the product uses a client-server architecture, and that the vendor has performed large-scale deployments that you can contact as references
- Security and compliance: The solution should securely handle sensitive data both at rest and during transit. It should support best-in-class encryption of bots—such as AES-256—and be compliant with local, industry, and global standards and regulations such as the Federal Information Processing Standard(FIPS) and Veracode.
- Availability: The solution should not have a single point of failure, and must support high availability of both “control room” and “bot runner” machines
- Enterprise support and services: The solution provider must have a dedicated support and implementation team to deliver in-house support 24/7/365 using a service level agreement (SLA)-driven framework
Q: How is RPA implemented in a broken process?
A: RPA’s best characteristic is that it can be applied to any number of different business processes. RPA bots can work in both “attended” and “unattended” modes. For example, when the entire end-to-end process cannot be automated, RPA bots can work alongside humans to deliver attended automation. In such cases, the actions of RPA bots can be triggered by system-level events that can give and take data to and from human workers.
Q: What is the best strategy for implementing RPA? RPA as a service or build a Center of Excellence (COE)?
A: A center of excellence (COE) is the best route to go because it gives you more control, and enables you to deploy RPA in a way that is more secure and compliant. A COE also helps you build in-house automation capabilities, the benefits of which will grow in the future medium- to long-term. A COE also provides more flexibility and room for experimentation with different automation frameworks. If you want to automate a large number of processes within a shorter timeframe, you can often engage the services of a third-party expert partner rather than outsourcing the day-to-day operations as you would with RPA as a service.
Q: Is it possible to automate 100% of a process?
A: It depends on the process. Standard business processes that don’t require much decision-making can be 100% automated, and deliver a very high, very fast, return on investment (ROI). The percentage of a business process that can be automated depends on three things: whether the process has standardized rules and patterns; how many human touchpoints exist; and the overall complexity and variability of the data.
Q: How can businesses set expectations as they begin their RPA journeys? What are the criteria for success during the initiation phase?
A: Businesses need to plan for both the short term and the long term. We encourage enterprises to think big, start small, and scale fast. During the initial stages, they have to carefully select what to automate by studying end-to-end business processes. An RPA COE must be established. Then, begin to automate those processes that are highly standardized, and which have a large and engaged headcount.
Here are the steps, in order, of doing this:
- Identify processes that are candidates for automation
- Perform a feasibility assessment
- Perform a complexity estimate
- Calculate expected ROI
- Prioritize automation projects
- Manage development resources
- Review code
- Test code
- Deploy production bots
- Manage production operations
Key success criteria during initial phases should include the speed of automation, the number of automation fulltime equivalents that are created, and the ROI. Once you’ve achieved initial success, expand your COE and create a bigger pool of processes to automate.
Q: What is the disaster recovery process for RPA? What happens if a bot fails in the production system due to a small process change?
A: Disaster recovery (DR) of an RPA system occurs on two levels:
- DR of the centralized control room: The control room with the central bot repository must be available and able to automate the original schedules from another site in event of a disaster.
- DR of bot-runner machines: The bot runner machines on which the automations run should be able to execute from another site in event of disaster.
If a bot fails in a production system due to a small change in an automated application, it must be designed so it doesn’t need to completely recreate the entire automation sequence. Instead, the RPA solution should support app resilience so that minimal edits can be made to very small number of reusable bots. The rest of the bots should work as usual unaffected by the change.
Q: How should an organization present a business case for RPA?
A: An RPA Business case should include three dimensions:
- Cost Reduction: The business should calculate the number of fulltime employees replaced, the ROIs on various processes, and the impact on the organization’s bottom line. For example RPA mitigates the need to spend money on system integrations.
- TimeThe business should analyze the efficiencies gained, including reductions in cycle time, and faster time to market leading to sooner and higher realization of revenue
- Quality: Finally, the business should identify how automated processes eliminate the human errors that commonly occur in mundane, repetitive processes. It should calculate how higher-quality products and services can increase brand equity, reduce or eliminate compliance penalties, and minimize customer service complaints and costs
Q: What can existing human workers do to upskill and contribute value to RPA efforts rather than lose their jobs?
A: Only the most mundane, repetitive aspects of human jobs are automated with RPA. This frees up humans to perform tasks better suited to their intellect and creativity. They can also use the extra time to learn new business skills or acquire new competencies related to RPA. For example, by learning automation tools or automation implementation frameworks, they can uplevel their skills. They also help their organization by setting up or strengthening automation COEs, or identifying opportunities for further automation by becoming automation change agents.
Automation also creates jobs by making enterprises more productive, creating opportunities for more person power needed for medium and high-skill positions.