Hackathon Semi Final 2 (S1) – Use Case 3 | RPA with help of AI | CA. Parth Chauhan

Summary

This presentation explores the power of Robotic Process Automation (RPA) and its integration with AI to automate tasks like GST data extraction and drafting responses to notices. The speaker demonstrates a practical application of RPA, using AI to generate code for automating tasks with minimal coding knowledge. Challenges include website changes, lack of specific APIs, and the need for context-specific AI training. While RPA is still in development, it offers significant potential for efficiency. The importance of security and business continuity in RPA deployment is also emphasized, highlighting the need for constant adaptation and oversight.

Takeaways

• 😀 RPA (Robotic Process Automation) is a software tool designed to mimic human tasks, not an actual robot. It automates repetitive tasks through a set of instructions.
• 😀 AI, especially with tools like ChatGPT, can be used to create RPA scripts, simplifying the process of automation, even for those with no coding background.
• 😀 A demonstration was provided showing how RPA can automate data extraction from the GST website, including entering GST numbers and handling captchas.
• 😀 RPA can be integrated with AI to automate various business tasks, such as drafting responses for GST notices. This process involves fetching data, text extraction, context identification, and AI-driven response creation.
• 😀 OCR (Optical Character Recognition) is used for extracting text from scanned documents, which can then be processed further for automation purposes.
• 😀 One challenge in implementing RPA is the lack of specific APIs from government sites, making RPA a necessary solution for automating tasks like extracting GST data.
• 😀 Context setting is a crucial part of AI. For instance, AI needs to be trained specifically on the relevant laws and regulations (e.g., GST laws) to generate accurate responses.
• 😀 The implementation of RPA involves both local AI models (easier to implement but slower) and cloud-based AI models (faster but may incur costs).
• 😀 A key limitation of current RPA implementations is the need for continuous human intervention to fine-tune prompts and ensure accurate results, especially with complex or changing websites.
• 😀 Security precautions are critical when deploying RPA systems to avoid data breaches. It's important to ensure that code runs securely on local systems and that data is protected from being exploited by third-party providers.
• 😀 A business continuity plan (BCP) is essential for handling failures in RPA systems. RPA processes should be supported by contingency plans to ensure business operations continue smoothly if automation fails.

Q & A

What is Robotic Process Automation (RPA) and how does it work?

RPA is a software technology that automates repetitive human tasks by mimicking them. It doesn't involve robots but rather uses software code to simulate human actions. For example, using RPA to retrieve data from the GST website involves opening a browser, entering the GST number, and extracting the required data automatically.

What are the key steps in automating processes using RPA for GST-related tasks?

The process involves several key steps: opening the web browser, entering the GST number, handling CAPTCHA verification, capturing the data from the GST website, and then presenting the extracted details on the screen. The automation mimics these steps through code.

How can AI assist in improving the automation process with RPA?

AI can enhance RPA by automating additional steps such as context identification and response generation. For example, once data is extracted from websites like GST or income tax portals, AI can be used to process and interpret the data and draft responses or identify relevant legal information.

What challenges are encountered when implementing RPA for automation in tax-related tasks?

Some challenges include the lack of official APIs from government websites, frequent changes in these sites, and missing notices or data. These obstacles may require continuous monitoring and adjustments in the RPA system to ensure proper automation and data extraction.

What is the role of context setting in AI for tax-related automation?

Context setting ensures that AI systems, such as GPT models, can focus on the correct domain (e.g., GST or income tax laws). It involves training the AI to understand specific regulations and provide relevant answers within that context, making the AI more effective for specialized tasks.

How does OCR (Optical Character Recognition) contribute to the automation process?

OCR technology helps in extracting text from scanned documents or PDFs, enabling automation tools to process notices and forms that are not directly available in digital text format. This step is crucial in extracting meaningful data from documents for further automation.

What is the difference between small AI models and large AI models in terms of implementation for RPA?

Small AI models can be implemented on local PCs and are easy to use but may have slower response times. Large AI models, like those in the cloud, require more computational resources but can provide faster and more reliable results, especially for complex tasks.

Why is there a need for continuous monitoring and modification of RPA systems?

Continuous monitoring is required because government websites and tax-related portals often change their layout or functionality, which can break existing RPA scripts. RPA systems need to be adjusted frequently to stay in sync with these changes and ensure they continue to work properly.

What precautions should be taken when deploying RPA in automation processes to ensure security?

It is crucial to deploy RPA systems on secure local PCs to prevent data leaks. Users should be aware that many software providers collect data during the automation process, so caution is needed to avoid exposing sensitive information. Developing custom RPA models can help mitigate these risks.

What should be considered when creating a business continuity plan for RPA systems?

A business continuity plan should address scenarios where RPA systems fail, such as when websites change or when the RPA scripts break. It's essential to have a backup plan, such as manual intervention, to ensure business processes can continue smoothly even when automation fails.