10 ways GenAI improves software development

If you’re helping to lead your company’s software development, generative AI (GenAI) can boost your productivity and speed by 20-50% — right now. GenAI can also improve product quality and increase end user satisfaction. It can even give rise to whole new classes of products and services. You can achieve these benefits whether your focus is product development, supporting the business or both. They’re available whether your developers are in-house or at service providers.

We know these productivity, speed and quality gains are real because we’re already achieving them at PwC with our in-house software teams. GenAI, working side-by-side with our people, is taking ideas and turning them into requirements, then turning requirements into user stories, user stories into test cases, test cases into code and code into documentation. It is speeding up and enhancing each of these steps.

This first-hand success has convinced us too that what we’re seeing today is just the beginning. Soon, GenAI will automate or augment every stage of software development. It might even make Agile as we currently know it obsolete.

What follows are 10 of the most achievable, impactful ways for GenAI to add value to software development today — and some guidelines for how to get ready for what GenAI will potentially do next.

Right now: How GenAI is changing the SDLC

At PwC, our firm-wide GenAI deployment has included software development — with especially gratifying results already achieved in ten areas. If you’re developing your own software, consider these use cases too. If you’re procuring software, consider providers that can generate this value and pass it on to you.

  1. Generate product features — automatically. With the right inputs, GenAI can generate relevant, detailed features in standardized, easy-to-use templates.
  2. Fast, consistent solutioning. GenAI can standardize and automate both solution architecture and systems design, saving time and money — and helping teams reuse components for further cost and time savings.
  3. Fast, accurate user stories. GenAI can break down complex requirements and features into user stories and generate acceptance criteria — saving teams time.
  4. Generate wireframes — automatically. GenAI can input both desired features and context, then generate wireframes to accelerate development.
  5. More test cases, faster. GenAI can analyze acceptance criteria, then generate test cases for a broader range of scenarios than human teams typically cover. That saves time — and can reduce the need for costly fixes later.
  6. Synthesize data to close gaps. GenAI can synthesize domain and use case specific data for regression testing, prototyping and more — saving time, increasing quality and reducing the need to use sensitive “real” data.
  7. Generate test scripts — automatically. Based on acceptance criteria and known uncertainties, GenAI can generate user acceptance testing (UAT) scripts — increasing the speed and productivity of quality assurance.
  8. Rapid, granular troubleshooting. GenAI can compare intended and actual outputs to facilitate root cause analysis. It can also automatically upgrade software library versions and replatform software.
  9. Complete and review code. Although GenAI can’t (yet) generate complex code from scratch, it can augment human coders through intelligent code completion, code refactoring suggestions and automated review.
  10. Automatic documentation. GenAI can document software development, boosting troubleshooting and enabling smoother handoffs between teams. GenAI can also automatically generate release notes and user guides.

Coming soon: New kinds of software — and business models

What kind of software will you develop when GenAI can help you do it in half the time and at half the cost? What new markets will you reach, and what new business models will emerge? These are questions worth considering because this future is approaching. Based on our everyday use of GenAI, the work in our innovation labs and our alliances with all the major developers in the GenAI ecosystem, we’re confident as to what GenAI will soon do for software development.

Over the next year, we anticipate the ten use cases above will improve rapidly. Skilled users will be able to instruct GenAI to generate high-quality artifacts for user stories, acceptance criteria, test cases, documentation and so on. Documentation, for example, will be dynamic and real-time. GenAI will generate APIs automatically too.

Then, GenAI will augment our work at every stage of the Agile life cycle. It will, for example, generate not just code “snippets,” but high-quality code. It will automatically conduct highly sophisticated simulations as well as performance and security testing. With GenAI’s help, a sprint that today takes two weeks will take two days.

The final stage: GenAI agents will be the next Agile

As GenAI matures further, it could redefine Agile, as it automates most Agile stages and continually shifts among them. AI “agents” will, likely for example, autonomously understand requirements, break down problems and generate code. Different AI agents will likely communicate and collaborate, much as people do today. These agents will improve themselves, automatically upgrading algorithms and strategies. With so much “experience” (i.e., data) on generating, testing, reviewing and improving code, these GenAI agents could predict user needs, maintenance requirements and potential system failures.

Development would no longer start with studies and plans. You could go straight to prototyping, telling GenAI to give you options. The drop in costs and time, and the rise in quality, could make even more new business models possible. Yet the need for skilled engineers will remain. Humans will need to creatively develop algorithms, architectures and user experiences to carefully instruct AI agents and to rigorously oversee them every step of the way.

Time to rethink ways of working

You’ll only get these benefits if your developers can dynamically prompt AI, with continual validation and iteration. They begin by dividing the task into small pieces since GenAI models excel with finely segmented projects. Developers next prompt the model to generate preliminary outputs — which they evaluate and use to give GenAI an even better prompt. As the cycle repeats, outputs can quickly approach optimal solutions.

If your people can identify “patterns” in GenAI use cases — similar tasks in different software projects and development stages — they can help scale up GenAI (and its value) quickly. And as GenAI automates routine tasks and enables developers to “try out” complex solutions quickly, developers can be more innovative and imaginative than ever. In our experience at PwC, developers thrive when you let them “play” with GenAI, either freeform (with appropriate guardrails) or in hackathons. Playful exploration can quickly become innovation and a serious competitive advantage.

What to do next

Whether you develop software in house, procure it or both, asking these questions can help you chart your near- and long-term course.

  • Are our providers passing on value in the right places? If you’re using third parties to deliver software, they should be using GenAI to give you structural savings and performance gains.
  • Where do we have Agile maturity? For in-house development, the fastest route to value with GenAI is often by using it to enhance Agile processes. But a mature Agile framework, such as the Scaled Agile Framework (SAFe), must be up and running and delivering benefits. If you aren’t fully Agile yet, assess which processes are mature enough for GenAI.
  • Do we have the technical foundation? For your people to use GenAI effectively, they’ll likely need a secure enterprise chat application, scalable plug-in architecture and applications toolkits and extensions.
  • Do we have the skills? Software engineers and their overseers may need new GenAI skills — such as prompt engineering, AI oversight and model management — and be coached to better collaborate with colleagues across development stages.
  • Can we manage risks and measure value? Besides a Responsible AI framework to manage enterprise-wide GenAI risks, you may also need to address software-specific GenAI risks, such as the risk of using GenAI where it isn’t appropriate. New frameworks, both for internal processes and providers, can help you track value delivered and ROI.
  • Is our AI operating model up to speed? The operating model we use at PwC — an AI factory — drives deployment through cross-functional teams, structured use case selection, rapid iteration, shared tools and knowledge, and robust oversight. It’s designed to help identify patterns of GenAI training and deployment so you can— if you choose — scale successful pilots quickly.
  • What new business strategies are possible? Assess how more cost-effective and timely software can open up opportunities for the business. One of our six 2024 AI business predictions is that GenAI will give rise to new classes of products and services. With GenAI transforming software development, your company may be able to deliver more customized products and services, reach new markets and create new business models.

Generative AI

Lead with trust to drive outcomes and transform the future of your business.

Learn more

Drive business reinvention through cloud engineering

Shape “the new” and drive growth with the transformative power of cloud engineering. 

Learn more

Bret Greenstein

Data and Analytics Partner, PwC US


Scott Petry

Principal, Cloud & Digital, PwC US


Andrew Carlson

Principal, Products & Technology, PwC US


What can generative AI do for you?

Generative AI is already transforming business. Contact us to learn more about this rapidly evolving technology — and how you can begin putting it to work in a responsible way.

Get in touch

Next and previous component will go here

Follow us