Generative AI
How much faster can coding assistants really make software delivery?
As we head into the second half of the decade, business leaders are beginning to appreciate the incredible promise generative AI has for unleashing innovation in their enterprise. Tools like ChatGPT and a stream of others that followed in its wake are simplifying interactions with AI to an extent previously unimaginable. But while 2024 might have been a year for experimentation and kicking the tires, this year it’s time to get serious.
One area of particular promise is the software development lifecycle. In this article, I’ll be sharing examples of the things we’ve discovered that really work.
Almost as soon as ChatGPT burst on the scene, questions arose about its suitability within the enterprise. What were the dangers of hallucinations? Can enterprises really use the code it creates in their demanding production systems? Over the past year, we systematically addressed many of those challenges. As these tools evolve, so do the strategies for leveraging them effectively. Through rigorous testing and innovation, we’re developing methods to ensure responsible and impactful use. This technological advancement highlights exciting opportunities, and we are committed to helping clients confidently utilize its full potential.
Enabling transformation across the software lifecycle
As we’ve seen firsthand, GenAI integrates neatly with everyday workflows, business applications, and general productivity goals. It allows you to do more with less, and speeds up time-consuming tasks. Fundamentally, it’s a tool that augments the capabilities of the person using it. And that augmentation can take many different forms. When it comes to software development, there are three main areas in play. The tooling is maturing across the entire lifecycle and going beyond editing code. This is providing a clear path for tremendous delivery acceleration and compressed timelines.
1. Faster, more cost-effective modernization
Many companies rely on systems based on creaking legacy code. At best this makes their operation slow and frustrating for users. At worst it can lead to a critical outage of service that damages both revenues and reputation. But bringing code up to date is expensive and time consuming. GenAI excels when it comes to reading, analysing and converting such legacy code.
Tools like Thoughtworks’ CodeConcise Legacy Assistant — a code discovery and intelligence tool designed to accelerate work involving large legacy systems — help businesses better understand their systems and tackle capability gaps, reducing the time and effort required to reap the rewards of enterprise modernization. Historically, a reverse engineering team would do this work, but it's a long, slow and complex job. GenAI is used to make sense of legacy codebases, for instance by spotting user roles, structures and relationships within the code, significantly speeding up the analysis process.
2. Accelerating software development
GenAI offers extended cognition to generalists, acting as a high-powered research tool that brings specialist knowledge to their finger tips. This means more time spent in the sought-after, high productivity ‘flow state’, and less time spent down Google rabbit holes, searching for domain-specific knowledge and insights.
AI is also an excellent collaborator, taking on other roles within a development team (without disturbing the actual team member from their own flow state) for brainstorming, ideation, and story creation.
With minimal prompting, generative AI can provide developers with context, content and expertise, in a fraction of the time it would take otherwise.
AI can also assist with client buy-in, helping move projects along by demonstrating at earlier stages exactly what’s possible. Rather than talking a customer through a presentation or showing them static slides to get them on board with your vision, GenAI makes it possible to work up a user interface at record speed. It may not be polished, but being able to show even basic functionality so early can have a huge impact on a project’s direction.
We’ve found that AI can help with so much more than just writing code. Which is why the accelerator we’ve developed, Haiven, is designed for use across the entire software development lifecycle by all team members — from those who build the software to those who create the business goals. Haiven is an AI-enabled knowledge amplifier that accelerates tasks across the software delivery lifecycle.
3. Shaping better products
Using GenAI within product lifecycle management is another ready-made win. Modernization, acceleration, and code-writing assistance all ladder up to improved product development. Returning to Google Cloud’s survey, 47% of organizations plan to use GenAI to develop new products, services, and business models in the short term, and in the long term that rises to 55%.
And those that have already started are seeing the benefits, with 84% of organizations moving a GenAI use case from idea to production in less than six months. When integrated fully into the product process, from inception all the way through to delivery, GenAI can have a significant impact on time-to-market and time-to-value.
Proven benefits for business
Time, efficiency, productivity. All gain a massive boost when GenAI is used as an enabler. Things that were once possible, but simply not economical, now become both.
Maintaining focus may sound like a simple win, but the cumulative impact is huge. One or two interruptions a day can wipe out hours of productive work. A recent survey by Atlassian explored the potential of GenAI when workers use it as a collaborator rather than just a simple tool. The survey found that using AI this way saved workers on average 105 minutes daily, equal to an extra workday each week.
And the evidence continues to grow. A Google Cloud survey on GenAI investment returns found 45% of organizations reporting improved employee productivity saw it at least double with GenAI use. The same survey also found that gains came quickly, with 74% of enterprises using GenAI reporting ROI within the first year.
From revenue growth to better security to improved user experience, GenAI is having a major effect across the board. But that’s not to say the road ahead is totally smooth.
Tackling barriers to adoption
GenAI has its flaws — the technology isn’t perfect. But why treat it differently to any other application you think can benefit your business? Start by establishing a way to understand what ROI looks like, what success looks like, and adhere to those metrics and understand that to its core before you start looking for a problem with this new tool.
Over the past year, we’ve taken huge strides in finding ways to use it for value. At Thoughtworks, we continue testing and advising our clients about the most appropriate application of the existing and next generation of GenAI tools. Whether that’s getting under the hood on emerging new tools, such as DeepSeek, or through identifying straightforward steps to GenAI success, our teams continue to find innovative ways to exploit the technology safely.
Of course, challenges remain. Issues like accuracy, scalability, and workforce adoption require thoughtful navigation. These are not roadblocks but stepping stones to progress. Our approach involves managing expectations, fostering psychological safety for experimentation, and guiding adoption through comprehensive change management frameworks. These steps ensure organizations reap the benefits while addressing both technical and human concerns.
Moving out of the sandbox into production
The time to act is now. GenAI is not an endpoint; it’s an enabler — a tool to support talented teams in achieving extraordinary outcomes. By integrating it widely, creating safe environments for innovation, and treating its adoption as a significant organizational shift, businesses can unlock new efficiencies and possibilities.
CodeConcise and Haiven exemplify the types of initiatives we think can accelerate companies AI programs, they’re just a small part of what’s possible. We’re excited to continue testing, learning, and advising on the most impactful ways to implement GenAI. The journey is just beginning, and we’re here to help our clients take bold steps forward.
