Let’s imagine a scenario where repetitive and tedious tasks in a business are carried out flawlessly, error-free, and at an impressive speed. This is the true potential of Generative AI, and unfortunately, not everyone sees it that way. The growing popularity in recent years of Robotic Process Automation (RPAs) has led to the false belief that they are the ultimate solution. However, as we will discuss, Generative AI goes far beyond, breaking the inherent limitations of RPAs.
Step 1. Forget everything you’ve learned about RPAs
Automation technologies – traditional, RPA, DPA, and BPA – have been around for a long time. Companies like UiPath or Microsoft have made great efforts to reduce the barrier to deploying these automations making them seemingly accessible even to non-programmers (low-code/no-code). However, the arrival of Generative AI completely changes the game:
1. We must rethink our approach. Everything we take for granted in this field requires reconsideration; for example, unstructured data – handwritten papers – that was previously not useful is now valuable, advanced learning and creativity are no longer exclusive to humans, most business decisions can be delegated to computers…
2. Automation robots are obsolete. The traditional concept of “preconfigured software instance for the autonomous execution of processes” has become inadequate to understand the capacity of Generative AI. As we mentioned in the previous installment, we recommend to stop thinking about automation robots and start thinking about autonomous workers.
3. Uncertain horizon for current automation partners. Many companies rely on partnerships with third parties to carry out their automations – such as UiPath, BluePrism, or Automation Anywhere. However, it is uncertain whether these companies will be able to offer Generative AI solutions beyond superficial integrations.
“Let’s imagine a scenario where repetitive and tedious tasks in a business are carried out flawlessly, error-free, and at an impressive speed. This is the true potential of Generative AI”
Step 2. Select tasks to automate
The effective implementation of Generative AI begins with the precise identification of tasks that will benefit the most from this technology. Here, we propose a strategic framework for selecting the right tasks:
1. High repetition: performed frequently and have a standardized process. Examples: transaction processing, data entry, routine tasks, pricing, claims acceptance, request acceptance, order processing, bank reconciliation, etc.
2. Bottleneck: slow down the overall workflow and their automation helps the efficiency of various processes simultaneously. We do not provide any examples here because it will depend on each company.
3. Multiple systems, variables, or data: require considering multiple systems – web apps, device apps, mobile apps -, variables or data to be able to be carried out. Example: to approve a request, I must consider solvency, history, age, medical records, etc.
4. Real-time: they need to be executed in a constantly changing and rapid environment. Example: recommendation systems, real-time price adjustments, payment processing, customer service.
5. Business impact: their accuracy and effectiveness directly impact business results. Example: data analysis for strategic decision-making.
Step 3. Develop and evaluate your automations
After identifying several potential tasks for automation, it’s time to develop and evaluate their effectiveness regularly. It is advisable to have the help of a Generative AI expert to assist with the following tasks:
1. State of the art of AI: it is essential to stay at the forefront of AI capabilities evolution to be skilled in determining the viability of each automation.
2. Necessary inputs to perform the task: locating all the instructions, data, variables, and systems needed to carry out the task is required. If we fail to locate them, we will not be able to trigger the automation at the right time, and it cannot be performed optimally.
3. Results of this task: it is necessary to identify the records left by performing the task. Determine if the result is recorded in software, a database, an application, if it is the input to trigger a subsequent task, if it becomes a report or deliverable, etc. If the desired result is not well defined success of the automation is at high risk.
4. Tracking automations: specific metrics must be implemented to measure the performance and effectiveness of automations in relation to business objectives, using real-time tracking tools and continuous feedback.
5. Security audit: regular security audits must be carried out to identify and mitigate risks, ensuring compliance with data protection regulations and maintaining high security standards in the automation infrastructure.
In a market where innovation determines leadership, it is essential that entrepreneurs do not settle for automation systems like RPAs. In exchange for offering robust and fast automations, these systems have limited their scope, precisely the category on which the entire potential of Generative AI rests.
Generative AI is not just an improvement on existing systems; it is an evolutionary leap that allows companies not only to compete but to lead in a constantly changing business environment. It is important in a Technological Revolution era to build defensive moats around our businesses.
