Adaptive Organizations for AI Success

Recently an AI Podcast made the assertion that companies which best organize to leverage AI will be market leaders. It’s one of those assertions that breaks apart or becomes obvious the more you reflect on it. Companies that can leverage any new tool effectively will be more successful. It should be a law on its own. The law of ‘adapt to thrive’.

But the assertion got me thinking; what would a company that effectively leverages AI ‘look like’. How would engineering organizational structures change, how would engineering adjacent organizational structures change, how would processes change?

My first instinct was to reach for an old favorite — Conway’s law. If organizations build systems which mirror their organizational structures — then ipso facto, Organizations will not change because of AI. Organizations will leverage AI in a way that mirrors or reinforces their existing organization. That’s not a very interesting answer — but it is plausible. Generative AI is a flexible tool and it will be used in lots of use cases.

That answer, while not interesting, simultaneouly fails to really answer the core question. The answer doesn’t address the ‘successful’ part of the question. What would a company that successfully leverages the new tools look like?

What if we look at the ‘Reverse Conway’ — while more of an adage than a research supported ‘law’ — the ‘Reverse Conway’ asserts that companies will modify their structures around core tools. Think SAP or Salesforce. Companies that leverage these large platforms have a specific organizational structure. The organization becomes fit for purpose to support and extend the technology.

I think that gives a thread to pull on. AI tools themselves are very flexible, they don’t have the dedicated support structure of an ERP. But the tools can change our processes. Looking at writing and reviewing code; the asymmetric scaling of code output is necessitating a change to the code review process. A process that has been gradually codified over the 15 years since git was introduced. Our processes will change. Organizations that can modify their structures to optimize this new process will be more successful than others.

So ‘Tool Adoption’ → ‘Process Optimization’ + ‘Organizational Change’ == New Structures

That is also not a satisfying answer. While probably a realistic assertion, it misses the underlying reason why some companies and organizations are better able to make these sorts of changes. If we look at market leader charts for any ‘organizational’ or project success metric — there are always leaders. Some companies rise above the competition. Whether that’s grounded in agile transformations, DORA metrics, SRE — you name it, some places are better than others at adoption, staying ahead of the curve, and optimizing process.

Those companies are the adaptive companies. Most will have some market driven need to change. They will have autonomous teams with results aligned to the team goals. They will seek input from all levels, and iteratively change.

In short, the organizational structure that defines AI success will be the same organizational structure that has driven success across the last 50 years. It will be a structure which creates an environment of directed and coordinated autonomy. The structure will achieve the hard to achieve balance between giving teams authority, communicating what has and hasn’t worked, providing appropriate guardrails, and focusing on customer value. My advice, strengthen the automated guardrails that handle compliance and quality and keep a close eye on business KPI’s. Relax the sprint metrics. When the guardrails do the governance work, measuring velocity just optimizes for points.