It is no secret that AI has fundamentally changed the way people and companies work. What began as simple predictive text has evolved rapidly. Since the introduction of the Transformer architecture in Google’s Attention Is All You Need, large language models have become more sophisticated than ever. AI is no longer just a chatbot. It’s now a salient feature of the technology landscape, becoming deeply embedded into the way we work and interact with our tech and each other.
For technical teams, AI has changed the entire software development lifecycle and enabled a new model for team organization and leadership. AI has accelerated coding, allowing developers to dedicate more time to complex and creative tasks. Simultaneously, it's allowing teams to clear bottlenecks of repetitive tasks through automation, enabling leaders to create more agile teams and focus on higher-level strategic problems.
"It's not just about, do we write code faster? It's really about, can we ship software faster? And... the fact of the matter is [writing code] was never really the bottleneck for shipping software. The bottlenecks were all of the other things surrounding it. It was the operation of the software in production," Ryan J. Salva, Senior Director of Product at Google shared on episode 11 of the Leaders of Code Podcast. “And so what the AI is really doing is it's giving us a fighting chance at actually delivering our commitments each sprint or month.”
This article dives into how leaders can:
- Leverage AI automation to clear team bottlenecks
- Foster collaboration and enable agility within smaller teams
- Create a strong learning culture
- Maintain the quality of documentation and organizational knowledge
Clearing out bottlenecks with automation
Through coding assistants like Copilot integrated directly into workflows and IDEs, AI has reduced a developer's need to tab-switch, allowing them to spend less time manually coding and more time focusing on higher level priorities. The true benefits for technical teams come not just from generative coding assistants, but also from AI’s ability to automate the repetitive tasks that often lead to bottlenecks for programmers.
Developers now have an assistant in their pocket that helps with routine but important tasks, including writing documentation, staying organized, and managing administrative work. According to a research report by McKinsey, the use of AI among developers helps expedite manual tasks. AI can now document code functionality as a developer is coding, and auto-fill standard functions for them. Developers have increased speed when it comes to updating existing code, as AI assistants can facilitate changes both through automation and code generation.
AI also gives developers enhanced functionality and efficiency when it comes to bug detection, quality assurance, and testing, according to Carnegie Mellon’s School of Computer Science. “It does help us go faster and reduces cognitive load in the areas of overhead,” Peter O’Connor, Stack Overflow’s Senior Director of Platform Engineering shared about AI’s ability to automate repetitive work, “I don't need to go back to my Jira task and mark it complete. Can something just do that for me? That would be great.”
Reducing cognitive toil allows teams to focus on the bigger picture, instead of spending their days checking off repetitive tasks. For Google, this has translated to engineers at all levels thinking about strategy and architecture. Ryan J. Silva explained, “A lot of those architecture conversations that used to only happen among our most senior principal staff engineers and Uber tech leads, those conversations are now the everyday domain of our level two engineers, because we all need to be thinking at the architecture level."
With AI, developers no longer need to address rote tasks every time, freeing up their workloads and minds to focus on high-level thinking, allowing innovation and creativity to flow in the organization. For Salva and Google, “[AI automation] clears a lot of the work from my engineers to then go fix the issue rather than do the paperwork, the bureaucracy, the project management of bringing it together.”
Becoming more agile through the reorganization of teams
AI automation is helping organizations by improving agility and collaboration. With less cognitive load and routine tasks, teams are now able to focus on solving problems instead of just completing a to-do list. At Google, this has manifested itself as a reorganization of teams into smaller, more collaborative parts. “The thing that I'm starting to see really just this year is that teams that used to be organized around 30 to 60 people to deliver a single capability or a single service are starting to break down into smaller components,” Ryan J. Salva shared about how his teams are being reshaped.
This restructuring reduces the collaboration tax on teams, making smaller teams more agile and creative when addressing challenges. Communication becomes easier, leading to faster iteration and a quicker response to new information, while reducing back-and-forth that often occurs in larger teams. “The more people you pile onto a problem, you got to get everyone on the same page... trying to get 15, 20 people on the same page is very different than you and the person sitting next to you,” Salva explained.
But this has not led to a reduction of developers at Google. Instead, small teams of developers are able to focus more clearly on problems as they arise, working in a more organized and collaborative way to produce better results. On smaller teams, context is pre-built into conversations, as developers work closely with their teammates and have greater visibility into their workflows. This opens up teams for more collaboration, reducing the collaboration tax on employees that often hinders their ability to innovate and work creatively. At Google, Salva notes, “When you're able to have higher bandwidth conversations with you and just a couple of collaborators, that is what allows you to move much more quickly. It allows you to be more agile and more nimble in the way that you react to new information as you're shipping these capabilities.”
Enabling teams for productive AI outcomes
AI is changing the way technical teams operate, making it imperative for leadership to focus on real business outcomes. Instead of relying solely on productivity metrics like lines of code or pull requests, leaders should create an environment that encourages practice and shared learning. To do this, leaders need to view learning and creativity not just as benefiting their employees, but as essential elements of the overall business strategy. Without this, AI adoption will stagnate, and leadership will face hurdles in getting teams to work successfully with AI tools.
According to a McKinsey report on AI in the workplace, nearly half of employees cite AI training as the most important factor for AI adoption. Workshops, paired AI programming sessions, and forums where team members can share tips and experiences are all ways that leaders can help build a culture of learning that facilitates widespread adoption and engagement with AI tools.
To create an environment where learning is not just encouraged but deeply engrained, Christina Dacauaziliqua, Senior Learning Specialist at Morgan Stanley, recommends connecting business successes to the learning and reflection of employees. “I think sometimes success is viewed in isolation as if it's something that comes out of nowhere. And we really need to create that dialogue of yes, it does happen because there is a lot of reflection into lessons learned and wins in a very intentional way.”
By creating an environment of learning and communication linked directly to business success, employees can build confidence and skills, knowing their work is making a significant business impact. Leaders need to actively encourage learning as part of their teams’ daily work. This approach helps spread the benefits of learning across teams and departments, enabling organizations to experience widespread adoption and understanding of tools, along more opportunities for creativity and innovation.
Organizational knowledge helping to drive team creativity
AI’s effectiveness in helping teams become more agile, productive, and strategic depends on the quality of the data it is trained on. For this, leadership must prioritize the curation of accurate, high-quality documentation and knowledge bases. As teams begin to use AI to automate processes in their work, low-quality data will lead to incorrect actions and hallucinations. While AI is able to generate documentation, humans are still responsible for its quality and accuracy to avoid technical debt.
“Documentation is paramount,” Ryan J. Salva, Senior Director of Product at Google, shared on episode 11 of Leaders of Code. “And the quality of that documentation… it's going to compound over time because large language models are brilliant imitators. And so if the documentation is not already exactly what you want it to be and really pristine…it will find that vulnerability, that soft spot in the code, and it will amplify it again and again.”
Leaders need to consider not only the speed at which products are shipped or projects are launched, but also how they are documenting these processes. Good data and well-documented organizational knowledge are at the core of a successful AI strategy—one that does not accrue more technical debt or cognitive load for its employees, but instead facilitates innovation and productivity.
AI for technical teams: Going beyond just code generation and towards innovation
While code generation is often the first thing leaders and employees think of when adding AI into technical workflows, it is really AI’s ability to automate the “work around the work” that is proving to be transformative for organizations. By removing bottlenecks caused by repetitive administrative tasks and testing, AI enables developers to focus on higher-level strategic thinking, fostering a more innovative and collaborative workforce.
As AI becomes more and more ingrained in the workflows of successful companies, team and department structures and processes will change. For some organizations, like Google, this has already led to fundamental shifts towards smaller, more agile units able to address issues quickly. But unlike the work of AI, this success isn’t automatic. It’s dependent on leadership’s buy-in.
Leaders must prioritize a strong learning culture that emphasizes creative thinking, allowing teams to experiment with new ideas and tools. High-quality documentation must also be a priority, as AI is only as effective as the organizational knowledge it is trained on. When technical teams go beyond just code generation and have strong support from leadership, real wins from AI automation happen and organizational transformation can be achieved.