Are you an engineering leader with a 2025 mandate of transforming productivity across your org? Are you an engineering leader handling complex, enterprise-scale projects, with a lot of work happening on top of legacy applications, and with change requests aplenty? Are you handling, say, large scale SAP or Salesforce or ServiceNow implementations for your global customers?
If yes, your situation, as you are well aware, is very different from an engineering leader who is building brand new products from scratch. The complexities you must deal with are orders of magnitude different; you are always operating within a deeply established tech, business, and process context.
In such scenarios, how can you move forward to incorporate AI into your SDLC in low-risk and non-disruptive ways? Here is a tip sheet for you to consider as you close out your 2024.
Before you go any further with tech modernization or SDLC efficiency, first get a handle on everything – every application, every service, every module – that you are currently sitting on.
Are you sitting on hundreds of thousands (millions?) of lines of code, across systems, built over years (decades?), interconnected in ways that will leave anyone looking at it closely with a headache? The single biggest – and quickest – impact you can have on your SDLC is to simply extract all the knowledge that is folded within your codebase – and your various documents – into one single, comprehensive, easy-to-navigate portal.
Every SDLC persona will benefit from having the entire application knowledge available for easy access, described in simple English. LLMs with programming language-specific adapters can do this – by learning everything there is to learn from raw codebase and all prior documentation, and then turning this into a comprehensive knowledge portal – scroll through your application, query a chatbot to understand more about some aspect of the application, view code quality checks, etc.
This in itself would be a high-impact project for you in 2025. It might take months to do if you assign a person or two to it. There are models out there that can get this done – for immensely large and complex codebases – in a few hours. We know, because at Prodapt we have built one that is doing exactly this, for large global enterprises. (PS: See tip 5 below).
Most AI applications for SDLC are built for individual devs (or testers, etc.) If I need to write a block of code and I can describe it well
enough on a chat prompt, the code gets generated for me. Which is great.
But what about a few steps upwards in the chain – the leads, the managers? Particularly in the complex enterprise-grade brownfield context we spoke about? The user stories, the test scripts, the design and architecture documents – all these are still being created by your leads and managers, with not much help, AI or otherwise.
Enabling this SDLC layer will have a cascading effect downwards. 2 user stories in the time of 1, reduction in the time it takes to create test scripts, a first cut of a project plan pre-created – these don’t seem like earth shattering changes, but for large customer engineering teams with high velocity new-feature and change requests, this will be significant. You can do up to 3X more work with the same team, you can ship faster, and you can check off multiple boxers in your calendar 2025 list.
Start with one of the personas (your BA lead, say) and build for that. Focused problem statement. There are tools available (again, see tip 4 :-). See the results for yourself, and then move into other personas, stitching them all together for significant productivity gains.
There are so many AI tools for SDLC enhancement, seemingly new ones every day. You could leave your teams free to play around in this landscape, but your job as an engineering leader will just get harder unless you have a way of bringing order to all of this.
You require a platform that brings just enough of all the major tools into one single location, that integrates with other tools as required to provide a simple interface. That will help you to drive systematic process efficiency across SDLC, instead of patches of high productivity with multiple bottlenecks.
What is better – having 100% of the capabilities of some of the AI tools, but have your teams work on them in disorganized and disconnected ways? Or have all your teams on the same unifying platform, even if each SDLC persona gets only 75% of the capabilities they would have obtained if they had gone solo.
Our Synapt AI labs team has built a platform that does all the above. We are excited about what we are building, and we would love to show it to you.
We are certain that engineering leaders like yourself will appreciate getting a behind-the-scenes look at what we are doing for Gen AI-driven SDLC productivity.
Nil
Nil
Deliver measurable outcomes for your business with #PracticalAI. Let’s talk!
