Critical Propulsion
← Back to Insights
Delivery3 min read

What Changes When Agents Do More of the Delivery Work?

Some of what used to fill a delivery lead's week is starting to take less time, and the job is moving toward the decisions agents can't make. Here is what left the calendar, and what I would rather be held accountable to instead.

I've been noticing a change in our delivery work lately. Some of the things that used to take up a lot of a delivery lead's time are starting to take less.

Status reporting is an easy example. If the work is already visible, the artifacts are current, and everyone can see where things stand, rebuilding all of that into a deck every week doesn't add much, so we're doing less of it.

The same thing is happening with routine reviews. More of the work can be checked against a spec, tests, and evals without a senior person looking at every detail. I still want experienced people reviewing the work, but I want them focused on the things the system can't answer on its own: whether we asked for the right thing, whether an assumption still holds up, and whether the result actually makes sense for the client.

Coordination is changing too. We've all been in recurring meetings that exist because the handoff between two teams isn't clear enough. When the handoff is structured and the artifacts carry the context, some of those meetings just stop being useful.

Estimation is starting to move in the same direction. We've spent years using story points and planning poker to put a number around uncertain work. Now probabilistic banding lets us work with that uncertainty more directly. We use ranges and what we're actually seeing in delivery instead of negotiating toward a number that's probably going to change anyway.

None of this feels like a big process change. It's happening gradually, as we find places where the old activity isn't helping as much as it used to. What it adds up to is that a delivery lead's calendar can get considerably lighter while the engagement gets better.

Some Delivery Activity Was Filling Gaps

A lot of the way we've traditionally run delivery exists because information is hard to keep current and hard to move between people.

We use meetings to keep teams aligned, status reports because the current state of the work isn't always obvious, and handoff documents because context gets lost between people. Reviews are another version of the same thing, a chance to catch mistakes and assumptions before they travel too far. Those things all had a reason to exist, and plenty of them still do.

But if the information is already available, the handoff is clear, or the routine check can happen automatically, I don't see much value in keeping the activity around just because it's familiar.

Where the Job Is Moving

What interests me is what delivery leads do with the time and attention that gives back. These days we're spending more of it on the decisions around the work and less on managing every step of the work itself.

I'd rather use senior people where their experience changes something. That might mean catching a poorly framed problem before a team starts building against it, or recognizing that two stakeholders appear to agree but are actually asking for different things. Sometimes it means knowing when the client needs to make a decision instead of letting the team keep working around it, or noticing that the business has changed since the spec was written. And sometimes it means staying out of the way and letting the work move, which is harder than it sounds and matters more than it looks.

None of that is new to delivery. What's new is that agents put more pressure on us to get it right, because once they have direction they can move very quickly.

That also changes what I care about in a review. I don't need a senior lead personally checking every routine piece of work, I need them to know where judgment is required. A system can tell us whether something matches the criteria we gave it, run tests, surface options, and flag inconsistencies. The harder part is knowing when the criteria themselves are wrong, when the client's priorities have moved but the work hasn't, or when there simply isn't an objectively correct answer and someone needs to make the tradeoff.

What I Expect from Delivery Leads

I'm starting to care less about how much of the process a senior lead personally touches and more about whether they know where their involvement changes the outcome. Just as important, I want them to know when they're no longer needed, because if every routine decision still has to pass through the senior person, we haven't changed the delivery model very much.

Clients may need to adjust to this too. We're used to seeing seniority through activity, where the senior person is in more meetings, reviews more work, gives more updates, and gets pulled into more problems. I'm not sure that's a very useful measure anymore, so here is what I would rather be held to instead. Bad assumptions get caught earlier. Client decisions get made when they need to be made. The team keeps moving without waiting for a senior person to approve every step.

As agents take on more of the execution, I think we're going to get a much clearer view of which parts of delivery actually need experience and judgment, and that's what I'm watching right now.

ShareLinkedInX

Stop Paying for Seats. Start Paying for Outcomes.

See how Critical Propulsion's AI Swarm model delivers enterprise-grade software at a fraction of traditional offshore cost, with zero timezone friction.