One foot in the AI-forward world, one foot in the before times. Most engineering organizations are stuck somewhere between the two, and getting them unstuck is the job.
A specific problem with an outcome we agree on before it starts: a review process that is not holding up, a stalled adoption effort, an assessment before a decision gets made.
Advisory retainerStanding time each month for the work that does not end: sitting with leadership as decisions come up, coaching managers, keeping a change from unwinding after the first quarter.
When an agent produces the wrong output while satisfying every written requirement, the failure is usually not the model. It is the intent that never made it into the spec.
Seeing what agents do is a necessary first layer. The product is the layer that tells an organization whether that activity is correct, effective, and improving.
The scarce resource in an agent-enabled engineering day is synchronous, high-bandwidth thinking. Everything downstream of that conversation is execution that can be delegated live or overnight.
Someone asked me recently: where do you get so much work to kick off? I did not understand the question in the moment, and it kept bothering me the next day. Engineering has to be developing the work. That does not mean making up work. There are tiers of work engineering should already own.
Agents run autonomously for hours only where verification is deterministic. That constraint, more than model intelligence, defines what agentic work actually functions today. It also explains a split that keeps repeating, where "agent ships your code" compounds and "agent builds your business plan" disappoints.
I have spent my career inside engineering organizations rather than around them, which means I show up as a working operator, not a slide deck. I work with organizations that know they are behind on AI and are not sure what to do about it, and I start from where they actually are rather than where the hype cycle says they should be.