Do AI and ChatGPT signal the end of the SAP developer?

I’d like to start this blog with a provocative question: Will we have developers in 5-10 years? With low-code and no-code solutions becoming more available, and ChatGPT showcasing AI’s potential capabilities for generating code, the future does indeed appear uncertain. Does it mean that we will be able to do away with our developers? I suspect the answer is no – but that the developer’s tasks will be very, and I mean very, different.

My expectation is that developers will become the masters of driving automation, generating code and solutions, and leveraging the power of AI. I suspect they will be freed from having to be tied to specific coding languages, architectural design styles or even specific applications, platforms, or tools. In many ways, our developers will be much more able to develop, yet at the same time, become deskilled. By this, I mean moving away from having deep skills to code in specialized programming languages as they learn how to maximize their use of technologies like AI.

My initial use of ChatGPT has shown me how an automated solution can transfer specific requirements into technical delivery without needing to know the coding languages’ detailed syntax in all areas. The ChatGPT can be asked if it can deliver something in a specific language, and it can ensure the syntax (and quality) is correct. In the future, knowing what to ask for, how you want it delivered, and having the vision and diligence to drive a concept through to an end deliverable will probably be key requirements for developers.

My background is with SAP software, and the range of technical or functional solutions I have to deal with are extremely broad and complex. The range of skills and technical capabilities now required to fully realize effective solutions for business is truly amazing. I for one, would certainly see huge benefits from leveraging low code and/or AI solutions. Freeing myself up from constantly developing (and renewing) my specific technical delivery skills would certainly make life easier!

It's the end of SAP technical delivery as we know it

I also see huge amounts of basic tasks and delivery work that I do that could be automated. Building a document to explain your design and gain formal approval could easily be the starting point to drive ChatGPT to deliver a large portion of the solution, with only final refinement/review requiring more personal intervention. I admit that this would still probably require some coding skills today, but would that be the case in 5 years or even 3?

Developing in an agile manner can also be a full reality. Later iterations could retain the use of automation, low-code options or final in-depth in-person review. With automated development of a core solution being rapidly realized, AI can unlock a truly rapid development path increasing time for review and feedback, but still resulting in much more rapid sprint speeds. Maybe we should add the mantra of ‘AI enabled’ to common industry lingo like ‘Cloud First’ and ‘stick to standard.’

AI can unlock a truly rapid development path increasing time for review and feedback but still resulting in much more rapid sprint speeds.

We really are at an inflection point for SAP technical delivery. Below I’ve outlined the top ten areas I see AI revolutionizing SAP delivery:

Ten ways AI will change SAP delivery

1. Truly realize low-code and no-code developments: There are already numerous solutions in this space, including SAP’s application and platform solutions. For example, to build an app or a report, you only have to select an SAP Core Data Services (CDS), add in your preferred app or report template, and connect to a system; these three steps can be enough to then auto-generate an app or report. With side annotations, you can further refine to your needs without coding, which can also be re-used in different solutions. So, in the future, we should be able to just task an AI User to go away and do this – which is a significant change. An AI User does not have to be linked to only a developer; it could be an end user requesting this directly from an AI User, empowering end users and making business more agile.
2. Automated project and upgrade delivery acceleration: Many tasks in project delivery could be sped up with automation and/or AI. Being able to provide standard configuration in a more automated manner, or to take one example set of configurations and then expand for new organizational or business units. In fact, you have only to see how SAP delivers its own best practices and content to see how a step change has already occurred in this area. SAP implementors are following this lead with their own delivery. For example, HCLTech’s Base90 aims to provide that 90% solution base in a much more rapid and automated manner.
3. Automate the test automation: Many companies still struggle to automate their testing. Why not use AI to generate those self-same “automated test scripts”? Hence an AI User could pick up SAP’s Test cases (supplied with SAP’s best practices) and generate automated tests. Equally, your company’s own documented test cases could also be utilized to feed to an AI User to generate the actual test cases.
4. Intelligently manage data: Data rules mining is already with us, and options to automate some of the detailed developments to migrate data, check data and replicate data are all possible. Additionally, taking a more dynamic approach to data migration via selective data transition enables much faster replication, but could also be further enhanced with more automated transitions or support of organization or configurational change. Lastly, low downtime or zero downtime and support of in-flight transactions or long running processes (aligned with transactional event management) can all be possible. The data migration can link the transactional data being migrated to its own position in a sequence of business events, hence the management and control of long running business events can still be retained and driven as you transition to a new or upgraded system.
5. Ease integration complexity: The integration across APIs, protocols or systems can be eased, as can orchestrating and mediating these to provide richer functional solutions. HCLTech has already seen benefits in this area using bots to automate the migration of SAP’s PI/PO to SAP CPI, thus significantly reducing the developers’ workload.
6. Provide a step change in agile delivery: As noted in the introduction, more rapid developments and use of AI could truly transform how agile (and DevOps) delivery are perceived. A very rapid, automated delivery can bring core solutions to life and would enable much faster iterative development and expansion, which in itself could be realized using AI and automation.
7. Migrate solutions to new platforms: The ability to migrate to new platforms, architectures, or design frameworks would all be eased if these can be delivered via AI and automation, thus avoiding on-going technical lock in or company-specific technical debt or complexity. Being able to leave the automated AI workforce to move these across to new architectures would unlock developers from one of the classic long-term corporate challenges.
8. Orchestrate and compose at pace: Being able to really utilize architectural layers for differentiation and innovation via re-using underlying components or solutions would enable more complex orchestrated solutions to be developed and to compose available (or new) items quickly to meet business needs.
9. Automate training: Training of others - or even yourself - can be hugely aided in the brave new Chatbot-style world. You could ask any question or clarification, and probably have real world examples tailored to your needs generated and created as required.
10. Get some work life balance: I left this to last since this is almost a blog in its own right. Imagine being able to leave an AI User to complete end user analysis to confirm business needs? Understand that the AI User is now suddenly a key user of your system with its own needs and requirements. Imagine leaving your AI User to work overnight on all those tasks and just update the direction/guidance or check on progress in the morning! Maintenance of systems or solutions could be transformed if smaller changes can be automated and left to the AI User. Could we really and finally have a shot at the fabled “work life balance”?

The above may well seem a little way off, but conversely many component parts are already in place, I would be the first to say we can’t just do all of this today, but I think the opportunities are enormous, and if companies are willing to be some of the first to engage in these areas then the benefits can be significant.

HCLTech would be keen to talk to companies that have large SAP-focused landscapes to help them understand how they could transform their delivery architecture and capabilities.