Generative AI Powered SDLC

Where’s Customer’s Voice?

Before diving into the details, we’d like to quickly revisit the topic of a previous article AI and Product Management (click here). Our concerns today directly stem from the growing number of Product Management jobs advertised today that explicitly reference AI in the position/title, e.g. AI Product Manager.’ Is the fact that organisations are including a reference to the solution - AI, fuelling the issues described in the latest MIT article: The GenAI Divide: The State of AI in Business 2025 (click here)?

Product Managers are responsible for brokering a 3-way collaboration between: investors, customers (users), and the implementation teams. Their principle role is to be the voice of the customer (user). Technology is a solution to a commercial problem or opportunity, (i.e. financial position), not the solution itself. Is this lack of focus leading to 95% of AI pilots failing to produce measurable financial returns, despite substantial corporate investment?

Note. Technology, such as Generative AI, cloud computing, ERP or business automation are all possible solutions to organisational challenges or opportunity, e.g. increase customer NPS, increase market share, reduce staff attrition or optimise IT project delivery.

Generative AI enabled IT & Software Development Life Cycle (SDLC) aims to:

1. Optimise Idea to Investment Timeline

2. Reduce Requirements Authoring Effort

3. Deliver Value Earlier and More Frequent

4. Reduce Reworking and Operational Risk

5. Optimise Engineering Effort

Optimise Idea to Investment Timeline

Generative AI provides a great opportunity to rapidly create a Lean business case. The cost saving (time), comes from its ability to ingest and process large quantities of material (e.g. competitive, market and associated environmental information (trends, technologies and growth)). The output is summary text that highlights such factors are core value drivers & aims, commercial landscape and trends. The speed at which Generative AI content analysis and creation can operate is staggering. Saving Product Management teams hours weekly as the environment within which they’re developing a strategy is under constant change from many factors that are is commonly referred to as PESTLE (political, economical, social, technical, legal legislative and environmental).

However, product Managers should not take everything the Generative AI produces at face-value.  True opportunities are discovered, this is of particular importance for commercially focused solutions, where teams are required to identify things that aren’t there and exploit them (i.e. an unserved need). Generative AI is exceptional at discovering and presenting patterns contained within large amounts of material (data) and these sometimes impresses product management teams at the expense of fully evaluating its true (low) value. Nevertheless, it’s currently less able to identify the absence of something (see Blue Ocean Strategy). Therefore, the successful development of a competitive product strategy can only truly emerge as a result of a human-led cross-discipline approach.

Reduce Requirements Authoring Effort

There’s a lot of excitement surrounding the promise of exploiting Generative AI to produce user stories. Certainly, its output can be consistent, adhering to the common user story template, (i.e. As a, I want to, So that); complete with a series of recommended set of acceptance criteria and conditions of satisfaction.

While several Agile Application Lifecycle Management tools are releasing plug-ins, which is freeing up a lot of Product Owners time that was previously necessary to produce these user stories, don’t be fooled by the quality of this output. Quality will always trump volume and speed.

The opportunity and ambition of leveraging certain types of Generative AI capabilities has been considered for many years by project management. Specifically, exploiting Machine Learning to identify critical success factors that influence the content, priority and allocation of individual activities to create a Lean Project Schedule is one prime example. (see MONETICAL own machine learning design, as way back as 2009 for a great example). Whilst exploiting Generative AI to create user stories with acceptable criteria in rapid time, thereby freeing-up Product Management’s time to focus on the added-value tasks, they need to fully appreciated and account for the constantly evolving plans and resource schedule.

Deliver Value Earlier and More Frequent

In a similar manner to those opportunities discovered in the optimisation of ideas section discussed earlier, Generative AI is an ideal tool to divert time and energy away from repetitive tasks, e.g. data gathering, analysis and presentation. Enabling every member of the team to focus 100% of their time on delivering value early and more frequently, as the time spent on ‘non-value add / administrative tasks’ is dramatically reduced, if not eradicated all together. The opportunity to leverage Generative AI to ingest a wealth of project related content, e.g. schedules, status requirements, issues and risks, possibly topics raise during a retrospective; enabled then to quickly produce highly informative status reports, complete with diagrams, high-level summaries and also perform its distribution to a predefined list of recipients. Dramatically reducing the administrative burden that typically falls to the appointed Facilitator (i.e. Scrum Master or Project Manager). This automated approach will certainly lead to the discovery of issues earlier (i.e. left shift), and where designed, the suggestion of ‘best-course of corrective measures’. If designed and implemented correctly, it may also lead to no longer needing extensive facilitation support, as other team members take-on responsibility for triggering the Generative AI action.

Nevertheless, those leading an initiative must not forget Generative AI only knows what it’s been told (i.e. the content its been requested to ingest). And as we know, there are many cultural factors that influence performance, which regularly go undocumented, such as, team moral, interaction, motivation, leadership style, the clarity and appetite to risk, and most importantly, tacit knowledge, which is rarely documented.

It’s for these reasons, teams need to exploit Generative AI primarily as a tool for facilitating better human interaction and management, not a substitute across certain aspects of the SDLC.

The opportunity to leverage certain types of Generative AI capability has been exploited for many years. In recent years, its growing use across tasks such as, enhanced sentiment and scenario analysis, performance (scaling) security testing (cyber) have all led to the successful adoption of a left shift strategy. A left shift - is a strategy that focusses on moving critical practices e.g. testing, security, and quality assurance (QA) to earlier stages of the SDLC. By resolving issues, defects, and vulnerabilities earlier, leads to cost savings, faster development cycles, improved product quality, and higher customer satisfaction.

Reduce Reworking and Operational Risk

For two decades the industry has been fully aware of the cost of change. With reports claiming the cost of change through the SDLC increases by 3fold from the requirements definition to design phase, 7fold by coding and 15fold by testing. For these reasons, the SDLC community has been seeking ways to reduce the cost of change, and this has led to the growing adoption of Agile ways of working.  The question has now become, how can the rapidly developing capabilities of Generative AI further advance this cause (i.e. idea)?

To answer this question, and before we consider exploiting Generative AI, we need to fully understand the reasons for change and its impact. It’s safe to state that many factors are driving the need for change, including, the teams constantly evolving understanding of customers’ needs and priorities, effort required to address defects uncovered during unit and integration testing, discovering certain technologies are unsuitable for specific requirements, and the constant updating of regulatory policies (industry and government bodies).

For SDLC teams to be able to manage change more effectively they would need to develop a greater understanding of the probability of each of these ‘need for change’ scenarios and the rapid access and knowledge on any suitable corrective measure. With these two parameters in-hand, they are well placed to prioritise investment towards reducing rework and operational risk.

One of the most staggering facts about Generative AI is the speed at which it can analysis vasts amount of data to determine probabilities, (i.e. the likelihood of events happening) and any present inferential factors, (i.e. statistics interprets patterns). Then, using the appropriate training models, that explicitly apply to those capabilities identify critical performance attributes. In other words, Generative AI has the capacity to identify the probability of change being required (in the future) by analysing historical data (similar projects and attributes), use regression (machine learning) techniques that surface (identify) potential risk and then present from a vast collection of historical scenarios the right course of action, before the events (need for change) is actually upon them.

Optimise Effort whilst Increasing the Value Delivered Ratio

Once agains SDLC teams have the opportunity to optimise the effort they invest in a new initiative whilst increasing the value of their work-product (i.e. output). Stakeholders and senior managers would much prefer greater accuracy of teams say/do ratio, than them infrequently delivering early (which rarely ever happens, simply because effort is always absorbed in software projects). They demand reliable estimates to ensure they allocate funds (i.e. opportunity cost), resources on those initiatives that are most likely to meet their strategic business goals. Therefore, accurate estimating is a fundamental component of an effective SDLC. The Triple Constraint Model, consists of the scope of a project, its schedule (time), and the estimated cost (effort). All three are interconnected and interdependent constraints of any project and every project owner is constantly having to weigh-up the importance of each.

The role of Generative AI and why

As we have discussed multiple times in this and previous articles, accurately identifying the most important requirements (scope), is a critical success factor. We have also discussed how Generative AI can be used to produce requirement specifications (i.e. user stories), but with some cautionary advice regarding the identification of those that aim to create a differentiation with its competitors, i.e. unique selling USP (unique selling proposition). Albeit, when appropriate, Generative AI offers an unrivalled opportunity to identify the required means (competencies) to implement, previously associated risks and effort of requirements of a similar type.

Unfortunately, for those initiatives that aim to be truly groundbreaking, committed to creating a new market or exploiting existing or merging technology in a new, previously unconsidered application, Generative AI’s contribution is limited right now.

The adoption of software development frameworks (pre-built collection of tools, libraries, and structures that provide a foundation for building software applications, allowing developers to focus on unique features rather than repetitive code) and the generic functional capability of most software solutions (e.g. user management, records creating and editing, workflows, reporting and operating environments), creates a perfected opportunity for Generative AI to become the principle code creator.

Having created an extensive, structured and standard set of requirements (i.e. user stories) and exploited it to create software code, initiative teams have created the opportunity for Generative AI to be used to automatically generate, execute and maintain test scripts. Thereby, dramatically reducing the time and effort associated with creating software of the highest quality.

Here we have described how three highly interdependent factors of an initiative, the scope (requirements), the schedule (time), and the effort (cost), can be seamlessly integrated with Generative AI to rapidly increase the effort to value delivered ratio within both types of SDLC methods, linear SDLC events (i.e. waterfall) and Agile software development methodology.