AI in the enterprise

For AI to be effective, gains need to be realized at an organizational level, not an individual one. But how do you actually do that? After all, organizations are really individuals and the systems and processes they build. And, to make things even more challenging, how do we do it when the technology landscape changes so quickly? Existing models of technology adoption, governance and training need to change.

As Chief Technology Officer and Chief Talent Officer we’re all too aware of these issues. Leaders have a significant part to play: they need to build the right mindset and cultivate a culture that not only adapts to and evolves with AI, but does so with a commitment to long-term goals.

Phantom productivity wins and bottlenecks

Almost as soon as ChatGPT was released in November 2022 the conversation turned to potential productivity gains. Some of this talk was pure melodrama and marketing, achieving little more than aiding AI vendors’ marketing efforts. But some of it was true — AI can cut waste, remove toil and accelerate productivity. 

Unfortunately, though, that’s not the end of the story. Gains are ultimately meaningless if they aren’t properly distributed; you’ll simply end up with a bottleneck at another point in a workflow. While some bottlenecks are to be expected when you’re experimenting with a new technology or approach, without the right strategic oversight, this can turn into a game of productivity whack-a-mole, trying to tackle bottlenecks at different points in a given process — only to find a new one crops up elsewhere.

Rethinking value streams

Clearly a holistic approach is needed. One way of doing this is to pay close attention to value streams: specific flows of work that produce particular outcomes. To be clear, this isn’t just about mapping how teams and functions work together. If anything, it requires looking beneath your organizational structure and at precisely what work needs to be done and when in order for value to be properly realized.

In fact, your existing organizational structure might even be part of the problem. As a kind of mental model, it can lead us to assume that all we need to do is plug AI in and encourage adoption across all functions. However organization structure is invariably trumped by communication structures: that means we can too mistakenly try and optimize things that aren’t working as we thought they did — if they’re working at all..

In turn, this plays into the bottleneck problem: functions unlocking their own pockets of productivity, impact, value (insert your chosen word) in a siloed manner. However, by focusing on value streams — even those that might not conform to the official picture of your organization — it becomes easier to not only consider how your organization should interact with AI but also allow your organization to evolve with it.

Individuals, teams and organizations

It’s by rethinking your value streams and thinking beyond existing org structures that we can begin to see how AI can be an enabler for every level of an organization: micro to macro. Instead of simply asking how might AI be used by individual teams we can instead explore how our organizational structures can be reconfigured and improved through AI-enablement.

Think about it this way: what’s the point of trying to unlock gains from AI when your existing processes and structures are hampering you? From an individual perspective, talk of AI could well sound like window dressing on what everyone knows to be organizational dysfunction. By thinking through the challenges faced by individuals and teams alongside AI the advantages will be far greater than trying to plug AI into discrete functions.

Bottom up vs. top down

Empowerment is about more than giving someone a tool — it’s about offering people the space and mechanisms to influence and shape decisions. 

Of course, creating these mechanisms and spaces never happens on its own: it requires leadership. These things need to be actively implemented. Some of this will come from culture — if knowledge sharing and collaboration is fundamental to how people and teams work together then you’re already starting from a good place. But beyond that you also need to build those rituals and ceremonies into how you operate. 

At Thoughtworks we ran our inaugural AI for software development festival. It featured talks from Thoughtworkers all over the world, offering a huge range of insights on how they’ve been using AI on projects. 

This was primarily a means of sharing knowledge and amplifying the great work being done across the company, but the further benefit was it showed how new ideas were spreading from people’s peers. In other words, while the festival itself required leadership and organizational resources, the content and ideas weren’t things being imposed from above: they were coming out of people’s work and driven by their creativity as technologists.

Leadership and institutional knowledge

The question of bottom up vs. top down approaches to learning and innovation are, perhaps unsurprisingly, about balance. Having the right guidelines and guardrails are vital, and essential if you’re serious about managing risk. But you do need some freedom in the system; individuals need to feel able to do things. If they don’t, those learnings you want to surface and then radiate throughout your organization simply won’t appear.

One way to think about this as a leader is to pay attention to the structure of learning and knowledge sharing more than the specifics of the content. It may be tempting to try and ensure that everyone’s using this or that tool or technique, but in reality things change so quickly that you’ll never be on top of it all. In truth, those on the ground are often closer to the changes than leaders. Good leadership, then, is about being curious and listening to those actually innovating and experimenting to solve day-to-day challenges.

The overall aim for leaders should be to increase and reinforce institutional knowledge. This involves not only ensuring knowledge is captured and shared but also that it becomes embedded in practices. This isn’t something that’s fixed and unchanging; it will continually evolve and adapt as new technologies emerge and as your people find new ways to leverage them successfully.

Strategy isn't a program; it’s a cycle of learning

There are undoubtedly parallels between the challenges practitioners face today and leaders: constant change and cognitive overload are now the norm. They’re no longer epiphenomenal — they’re the central part of our professional lives. 

In such an environment it can be tempting to want to simplify everything; to build a program that outlines what needs to be done and how. This is a mistake: leaders in every team need to recognize that strategy isn’t a program, it’s a cycle of learning. Acknowledging that and setting yourself and those around you to adapt and evolve is crucial.

We need to be able to leverage the surprising and the unpredictable — not let it throw us off course. Good leadership helps teams and organizations navigate precisely that challenge.