AIMultiple ResearchAIMultiple Research

5 RPA Programming Options You Need to Know in 2024

RPA is increasingly becoming an enterprise-level opportunity, and ~64% of respondents on the RPA journey, it is a strategic or enterprise-wide initiative. However, there are numerous options for RPA tools in terms of programming requirements, cognitive capabilities, or human in the loop functions, and choosing the right tool for your business can be a difficult task.

In this article, we explore the different RPA programming options (code-based, low code, no code, recording, and self-training), and the pros and cons of each to help businesses choose based on their capabilities and needed outcome.

1. Code based

As expected, the most powerful interface for programming bots is coding the program with a programming language. Need to explicitly code the tool to perform the necessary actions to replicate a process or functionality. However, coding requires training and patience so this method is relevant for technically inclined personnel. Programming instructions essentially tell the bot which programs to use and how to interact with those programs.

Limitations

Though these tools can be programmed in extremely flexible ways and can automate up to 70-80% of rule-based activities in an enterprise, they have some limitations:

  • non-Windows environments: Most RPA vendors do not offer solutions for non-Windows operating systems such as Mac OS or Linux. This is not a major problem most of the time because a majority of human dependent company processes are conducted on Windows machines.
  • Reliance on programming effort: From a purely theoretical perspective, any process can be fully automated. However, as process complexity increases, programming time and cost make automation financially infeasible.
  • Reliance on programmers: While bots are relatively easy to program, they still need to be programmed by tech-savy personnel. Enterprises solve this problem with several measures:
    • Enterprises are founding centers of excellence (CoE) where they gather RPA talent who help departments with their automation efforts and guide them in their RPA journey
    • Enterprises outsource programming to RPA implementations specialists or other consultants
  • Edge cases: These are problematic for all automation solutions. When bots encounter cases that programmers had not anticipated, results can be unexpected. This requires auditing bots during first roll-out to ensure that such cases are encountered and fixed. However, not all edge cases can be identified during the first week of operations. For example changing market conditions can generate new cases months after bots are rolled out. This requires building a warning system and carrying out regular audits in light of changing market, regulatory or technology conditions.

2. Low code / No code RPA solutions: Graphical User Interfaces (GUI)

Most modern RPA vendors offer low code and no code solutions to program simple RPA bots with drag & drop interfaces. Technically proficient personnel (those that can code excel macros) should be capable of setting up simple bots. For more technical users, these tools also offer code based bot programming interfaces.

Limitations

Although no code RPA solutions reduce development time and cost, they may lack business-specific bot functionalities (e.g. invoice automation, resume validation). Customizing the RPA bot to automate business-specific processes will require a certain level of coding skills.

3. Recording

Just like macros in excel, bots can complete recorded actions. Recording a complex set actions and having them automatically translated into a bot program facilitates programming. Most vendors offer such macro recorders. Recorded actions can involve numerous enterprise software such as taking data from Salesforce, merging it with a report from mailchimp in excel to identify which customers to target during the company’s routine customer activation SMS campaign.

Limitations

Recorder function is an important advantage in an RPA tool because it enables rapid bot programming. However, recorders have some limitations as well:

  • Recording a complex set of functionality can be difficult and error prone
  • Maintaining recorded bots is difficult as their code is machine produced and may not be easy to read. Re-recording actions after each small change in the process can also be time consuming

4. Self-learning bots

Programmable RPA solutions require significant coding, increasing the time until RPA roll-out. Self-learning bots are able to program themselves by monitoring employee activity in order to learn automatable tasks. They also leverage OCR and NLP to understand unstructured data (e.g. images, PDFs). They are the easiest to deploy but they are currently a relatively emerging field of RPA.

There are various approaches to train RPA tools:

  • Using historical (when available) and current data; these tools can monitor hours of employee activity to understand the tasks completed and to start completing them after they have reached enough confidence to complete the process.
  • Real time monitoring; Employees can use the tool as they complete tasks in the manual manner as they used to. As tasks are completed, tools learn the necessary activities and start automating them. Employees provide feedback to the tool as it increases its automation levels.

Limitations

There are 3 major limitations to self learning:

1- Time requirement:

RPA bot training takes time, especially with large and/or unstructured training datasets. A self-learning RPA bot will need to:

  • Extract the data needed for the task from unstructured data using machine learning techniques (e.g. OCR or NLP)
  • Create process rules by understanding workflow relations (e.g. if there is an invoice attachment in the email, download and scan invoice, then input data to spreadsheet)
  • Embed rules to bot memory to create a template for similar processes.

To ensure result accuracy, training data needs to be extensive and detailed, thus increasing the time to roll out the self-learning RPA solution.

There’s just too many possible inputs in a process. The key ingredient in machine learning is data and you need to feed a bot months of data for it to be an effective learner. If such data is available, that’s great. However, most of the time enterprises have access to the outcomes of a process, the structured data but the unstructured inputs are not stored for more than a month.  If that’s the case, then data collection will need to start ASAP and can take a few months depending on the specific process to be automated.

2- Lack of maturity of the solution: This is more of an area of research now. There are a few vendors that claim this capability but we have not been able to verify their effectiveness from their customers yet.

3- Learning errors: Learning from screenshots is not always perfect since learning relies on identifying images in scraped screenshots. Especially during initial deployment, these bots could be making mistakes and their activity needs to be audited. Most of the time mistakes are avoided as bots understand when they see cases they don’t know how to complete. In such cases, they contact employees for guidance.

Recommendations to business leaders

  • Choose the RPA development platform and tool based on your teams’ coding skills:
    • Choose code-based and low-code RPA tools if your team includes employees with good coding skills.
    • Choose low-code and no-code RPA tools for teams with limited programming knowledge.

To explore different RPA development options, read our in-depth articles:

If you still have any questions about RPA, read our in-depth whitepaper on the topic:

Get RPA Whitepaper

And if you believe your business will benefit from an RPA solution, scroll through our data-driven list of RPA tools, and other automation solutions.

And we can guide you throughout finding the right tool:

Find the Right Vendors

This article was drafted by former AIMultiple industry analyst Alamira Jouman Hajjar.

Access Cem's 2 decades of B2B tech experience as a tech consultant, enterprise leader, startup entrepreneur & industry analyst. Leverage insights informing top Fortune 500 every month.
Cem Dilmegani
Principal Analyst
Follow on

Cem Dilmegani
Principal Analyst

Cem has been the principal analyst at AIMultiple since 2017. AIMultiple informs hundreds of thousands of businesses (as per similarWeb) including 60% of Fortune 500 every month.

Cem's work has been cited by leading global publications including Business Insider, Forbes, Washington Post, global firms like Deloitte, HPE, NGOs like World Economic Forum and supranational organizations like European Commission. You can see more reputable companies and media that referenced AIMultiple.

Throughout his career, Cem served as a tech consultant, tech buyer and tech entrepreneur. He advised businesses on their enterprise software, automation, cloud, AI / ML and other technology related decisions at McKinsey & Company and Altman Solon for more than a decade. He also published a McKinsey report on digitalization.

He led technology strategy and procurement of a telco while reporting to the CEO. He has also led commercial growth of deep tech company Hypatos that reached a 7 digit annual recurring revenue and a 9 digit valuation from 0 within 2 years. Cem's work in Hypatos was covered by leading technology publications like TechCrunch and Business Insider.

Cem regularly speaks at international technology conferences. He graduated from Bogazici University as a computer engineer and holds an MBA from Columbia Business School.

To stay up-to-date on B2B tech & accelerate your enterprise:

Follow on

Next to Read

Comments

Your email address will not be published. All fields are required.

0 Comments