By Arjun Dublish, Head of Smart Reconciliations Premium, Smartstream
Last week we hired a new AI Product Manager. Formerly of Goldman Sachs, recent HEC Paris–London Business School MBA. Somewhere in the interview process, I realised the role she was walking into barely resembles the PM job of a decade ago. That realisation has been reshaping how we hire and how we structure our teams – and I think it has implications that go well beyond product management.
Every role has a floor and a ceiling. The floor is the routine work the role cannot function without: writing specs, documenting requirements, coordinating tickets, waiting for a first build before an assumption can be tested. The ceiling is judgment: knowing which customer complaint is a signal and which is noise, which trade-off to take when the data is ambiguous, when to kill a feature the roadmap says you should ship.
For most of my career, the floor consumed the majority of everyone’s week. We accepted this as the nature of work. A PM spent perhaps 20% of their time on decisions that actually moved the product and 80% on the machinery around those decisions. Only the most senior individual contributors or heads of product ever spent real time at the ceiling. The rest of the team serviced the floor, and we structured organisations accordingly.
That ratio has broken.
Agents Have Absorbed the Machinery
A PM on my team can prototype an idea in hours, put it in front of a reconciliation ops team, and watch where it breaks – before a single engineer is involved. The spec writes itself from the prototype rather than the other way round. The coordination work, the status chasing, the documentation: agents handle it.
The lazy conclusion from this is that AI is coming for the PM role. The accurate conclusion is stranger. The floor of the role has risen, and it keeps rising. The routine work has not disappeared; it has been absorbed. What remains is the part that was always hardest to hire for: judgment – commercial and technical, formed by experience, applied under ambiguity.
I find it worth pausing on what this means for organisational structure, not just individual roles. Consider how much overhead has existed purely to service the floor. At a previous company, there was an entire product operations department. Capable people, but the function was, in honest terms, administration: chasing updates, maintaining trackers, formatting roadmaps, making sure PMs did the routine parts of their own jobs. That function should not exist today. Not because the work stopped mattering, but because agents do it now, and what little remains should sit directly with the person accountable for the outcome. Building a department around that kind of supervision institutionalises the inefficiency.
In a domain like reconciliation – where mistakes are measured in regulatory findings rather than bug reports – that is a cost we should not be paying.
It Changes How We Hire
We no longer screen for whether someone can do the mechanical work competently. We assume the machinery is handled and screen for what someone does above it. Can this PM make a commercial argument a bank’s operations director would respect? Can this engineer challenge a design decision on business grounds, not just technical ones?
These used to be senior-person questions. They are now the entry bar.
That is a meaningful shift. It is not that junior roles have disappeared; it is that the expectation attached to them has moved. The candidate who would have been hired for their ability to manage a well-structured backlog is now being hired – or not – on the basis of whether they can hold a commercial view on a product decision. The tools are there. The judgment is what differentiates.
It Changes How Teams Work
Our engineers are not waiting for specs to land in Jira. They are in design trade-offs and commercial discussions from day one, because the mechanical work no longer fills their week either. Product still owns the decision. The decision is simply better when the people who will build it helped shape it early.
This is a structural change, not a cultural one. It is not about creating a more collaborative atmosphere. It is about the fact that the work which used to justify siloed handoffs – detailed written specs, formal approval gates, documentation cycles – is now generated faster than any human team could review it. The bottleneck has moved. It is no longer output; it is judgment on the output.
The Gap That Can No Longer Hide
The uncomfortable implication of a rising floor is that the gap between people who operate at their ceiling and people who were coasting on floor work becomes impossible to ignore. When routine work filled the week, mediocre judgment could hide inside busy competence. It cannot anymore.
Being busy has never been a proxy for good judgment. But when the busy work can be handled by an agent, that truth becomes visible in a way it simply was not before. The people who were adding genuine value through their thinking were always there. Now the contrast is clearer.
What Compounds from Here
The floor will keep rising. Every capability that becomes routine gets absorbed – that has been the pattern at every stage of automation, and there is no reason to expect it to stop. What compounds in value is the thing that has consistently resisted that process: judgment, built through experience, tested under pressure, applied in conditions where the data is incomplete and the stakes are real.
We hired for the floor because the floor was most of the job, particularly at entry level. We now hire for what someone does above it. The firms that move fastest on this – that reconfigure their hiring, their team structures, and their leadership expectations around that reality – will not just be more efficient. They will be making better decisions, faster, in exactly the domains where better decisions matter most.
At Smartstream, that shift is already visible in how we build and deploy Smart Agents. The agentic work that once required dedicated coordination overhead now runs continuously in the background – freeing the people closest to our clients’ operations to spend their time on the decisions that actually move things.
If you would like to discuss how Smartstream is thinking about AI-driven operations in financial services, get in touch.
