Accelerated Project Delivery with genAI

With the emergence of genAI and rapid adoption of genAI in multitudes of solutions across the world , I would like to present few use cases which can be developed to leverage genAI and can enable rapid deployment of IT implementation projects.

The present project implementation schedules, which can take months or even years, are rather lengthy. This has always been the case and hasn’t really changed over time. However, we can now fully realize some significant advantages to automate numerous project lifecycle phases thanks to genAI.

The concept revolves around leveraging genAI to automate key aspects of each stage in the software delivery process. Let’s delve into the specifics of each stage below.

Solution Design

A requirement gathering session and a BRD (Business Requirements Document) are standard components of the solution design phase. The team responsible for delivering the solution will discuss and assess potential high-level solution designs based on the requirements.

A business analyst or a functional consultant will usually detail the design and prepare what is known as an FDD, or functional design document, based on the high-level solution design that was selected.

I suppose you’ve realized by now that genAI offers the possibility to automate this. A trained & well-grounded LLM (Large Language Model) which has studied thousands of business requirement documents would be able to suggest a solution based on the selected underlying technology with ease.

Furthermore, it can create functional specifications with ease using any template that is needed. Magic. Again, though, I believe it’s far easier said than done. I have no doubt that human interaction will still be required to review, fix, or even completely rewrite the solution. But even at 40–50% automation, we still significantly reduce the amount of time needed for this phase.

Build & Programming Phase

Most of us have already come across a variety of genAI-based code generating tools. Thus, I don’t think many people will find this area new. But the objective is to continue from where we left off in the last phase. You can use the functional spec that was created during the Solution Design phase as an input to create code during this phase. Organizations offer a variety of code generating technologies that can generate entire stack code—from frontend to backend services—all at once.

Additionally, SAP has introduced build code, which appears to be a potent tool.We may train the LLMs with current code bases, semantics, and dictionary elements of a particular organization to make the tools more accurate. This will improve the real life usage of code generated.

Once more, while this won’t totally replace human intervention, it can tremendously aid the developer and shorten the development process lead time.

Data Transformation & Migration

Data migration is one of the most important and time-consuming phases of an ERP adoption project.

There are many processes involved in data migration, including extraction, transformation, load, post-load, and pre-load validations. Certain of these processes can be automated by using a trained LLM who is familiar with the structures and semantics of the source and target databases and files. For instance, transformation is easily automated.

Furthermore, we can utilize genAI to perform pre- and post-load validations, which not only automates the process but also minimizes conversion errors that would otherwise be detected in later stages of data migration.

There are numerous benefits, but the main one is that fewer mock data cycles will be required, and there won’t be a constant need to create new transformation tools. This can significantly shorten project durations.

Test Automation

As the name implies, we are now automating testing, which is a crucial stage of project delivery. Using genAI, test scripts may be automatically generated based on the selected technology and solution. Millions of automatic test scripts can be used to train the LLMs, thus creating and integrating the resulting test scripts with industry-leading test tools should not be too difficult.

It seems perfect and touchless. Isn’t it? However, given the maturity of many test automation technologies already in place, I believe the percentage of automation here can be much higher. It is easy to shorten the testing period from several weeks or months to a few days. This can significantly shorten the project’s total duration.

Production Support or Application Management Services

It will not be fair if we stop our genAI journey post project delivery and go live. Tools, Technology, Automations are of no use if they don’t work in production. genAI journey won’t be complete if we are not able to support production operations.

The documentation created during project lifecycle such as FDDs, User manuals, KDD (Knowledge Discovery in Databases) can be leveraged to create a Digital Assistant where users can ask queries or raise incidents in case they run across problems in production. This idea is nothing new and is available forever.

Goal is to minimize need of L1 support teams and completely automate tasks such as user query resolution, incident creation, incident assignment to right queue etc. Some of the advanced Digital Assistants can also aim to automate L2 operations using backend API integration.

Once again, a plethora of documentation pertaining to systems and organizational processes can be used to train LLMs in this situation.

Conclusion

As was initially mentioned, genAI has opened previously unimaginable possibilities in real life. We have a wonderful potential to quickly deploy IT projects when we can shorten the project lifetime. This can greatly assist organizations in embracing change and transforming more quickly, enabling them to stay up with the rapid advancements in technology occurring worldwide.