There has been a lot of attention given to the benefits of using generative AI to accelerate code development – Microsoft CoPilot, Amazon Q, and Gemini Code Assist, to name a few – but the real potential lies in holistically looking at the entire SDLC and using generative AI across all phases.
Naturally, the opportunity for greater developer productivity centers on the work to actually write code. And the data points are now backing this up. Across a range of studies, developers are realizing productivity gains of 27% to 55%, depending on skill level and code complexity.
When you widen/broaden your scope to the entire software development process – requirements, design, build, test, deploy, run – and consider the multitude of people and actions involved, what if they were all generative AI-enabled and generating similar or even greater productivity?
Why Requirements and Design Matter Most
The Standish Group’s CHAOS Report has been a benchmark in understanding software project success and failure rates. Key findings from the 2020 CHAOS Report include:
- Successful Projects: Only 31% of projects were delivered on time, on budget, and with the required features.
- Challenged Projects: 50% were late, over budget, or lacked necessary features.
- Failed Projects: 19% were canceled before completion.
A significant contribution to challenged and failed projects is poor requirements gathering and management. Getting requirements and design done right, reduces cost and time in development later. If we can automate this process with generative AI, combined with what can be done in code acceleration, there’s a significant opportunity to accelerate and change how software is built.
Reducing Friction Across the SDLC
In practice, no one in the software development process works in an autonomous silo; in addition to each phase being more productive, the handoffs between phases could also benefit from greater automation. If the requirements and design are not only more accurate, but more easily produced, shared and consumed by other teams (e.g. development, testing, support) the entire project moves faster. If requirements and design is done more accurately and quickly, the code would be done more accurately and even rapidly without generative AI. And even more so with generative AI.
Reducing the friction between discovery, design, and coding is a challenge seen across the industry, particularly in digital transformation. This has led to the introduction of Product Operations or ProdOps – and a similar opportunity to accelerate the benefits of ProdOps with generative AI.
The Rise of Product Operations (ProdOps)
To unlock all this new velocity, you have to look at the three pillars of any software development effort – the people, the process, and the technology. Changing any one of these independently of the other,s as you adopt generative AI tools across the SDLC, will only create new friction and bottlenecks. The process has to be validated with the new technology by people trained in the new tools. In an ideal world, retraining a team on new tools wouldn’t be necessary – and that is where there is an opportunity for real innovation. This is emerging in places, with plug-ins in IDEs and integration interfaces in DevOps and ProdOps platforms – allowing teams to leverage the investment in the tools they know, while powering them with new generative AI capabilities.
New Capabilities Are Accelerating Everything
And the pace of innovation continues at breakneck speed – consider the recent release of the Claude 3.5 Sonnet ‘computer use’ enabling Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking buttons, and typing text. The possibilities for this type of capability to transform software testing alone will drive significant acceleration – especially when combined with a framework to coordinate and track handoffs from development to production.
The Future: End‑to‑End Product Acceleration
All of these new technologies are powerful in and of themselves, and when combined with a standard orchestration fabric, they reduce friction and integration costs. Now, it’s possible to move beyond just code development acceleration to end-to-end product acceleration. By using a platform that orchestrates the entire process, one can improve one’s organization’s prod ops productivity by at least 50% while significantly reducing costs per project. The future is bright in ProdOps, and it’s just getting more colorful with the use of gen AI, revolutionizing processes to help us build better products faster.