Flattening the Agentic Engineering Org
And its impact on middle management . . .
The Year of Efficiency
Ever since Mark Zuckerberg brought us the year of efficiency in 2023 it’s been a challenging time to be a middle manager in an engineering org. Many companies have reduced the number of layers, increased the span of control and started to require more hands on coding from their Senior Engineering Managers, Directors and Senior Directors of engineering.
The Upcoming Impact of Agentic Development
One of the major responsibilities of middle management is to align and co-ordinate a large number of people to deliver a substantial workstream. But what happens when you have fewer people and co-ordination is mostly automated?
We’re already seeing that agentic tools can help with alignment, summarization, progress reporting and even provide a rich summary for executives to use during performance reviews. It seems increasingly likely that as companies adopt agentic engineering practices and re-envision their SDLC to be agent first, you’re going to see smaller teams and a flattening of the org structure.
What Might this Look Like?
One model that’s being discussed repeatedly is where a VP has 5-15 triads reporting directly up to them, with most of the 1:1’s handled by a “first among equal” team lead and where the VP could efficiently review and impact the context, framing and high level specs/verifications for their teams so they could have a material impact on the functionality and process for shipping their orgs code base.
How this plays out will be different in each org until we come to a new consensus, but in the meantime:
This is a great time to get hands on with building and understanding agentic tools. A deep understanding of context and intent management, thinly sliced verifications and well structured specifications will be a pre-requisite to leading agentic teams in 2027.
This is also a good time to start to ask what your engineering org might look like in an agentic world. If co-ordination was automated, code was cheap and the entire value delivery pipeline from OKRs to production code was a deterministic workflow with clear agentic and human in the loop steps, how would your org chart look?
What do you think your engineering org is going to look like in 2027 and beyond?