The receptionist was doing fine. The owner was the bottleneck.
An HR services CEO in Camberwell, a business consultant in Dublin and a software CEO in Delhi say the outcome should decide.
AI can tell you who is rostered on the support desk.
As HR Central's Damien Gooden puts it, "What it cannot do (yet) is actually provide that support".
Meanwhile, a Dublin consultant found that a healthcare practice's bottleneck wasn't the receptionist. It was the owner.
Three leaders, three countries, one reframe: start with the result, then pick the resource.
AI reads the roster. A person still shows up.
Damien Gooden, Group CEO of HR Central (Camberwell, Australia)
“AI is now able to tell our greater team which of our support team is rostered on for a given day, taking into account public holidays and leave. What it cannot do (yet) is actually provide that support”
“The key decision is - is there value in a human delivering this. For example - a request to change a password is transactional and clear cut, so there is no benefit to having a human as part of the process.”
Scheduling went to the machine, and some transactional steps need no human at all. His test for every step is whether a person adds value, and where the answer is yes, the person stays.
Count outcomes, not deleted tasks.
Sahil Kakkar, CEO of RankWatch (Delhi, India)
“The first question should be which outcomes improve, not which tasks disappear.”
“If human review remains unchanged, the product may support a role effectively, but it has not replaced it.”
A software CEO setting the bar for software claims. Fewer tasks is not the same as a role replaced, and he checks whether the human review changed.
The bottleneck wasn't the receptionist.
Gerard Collins, Founder of Business Growth Assessment (Dublin, Ireland)
“I worked with a healthcare practice where the owner repeatedly intervened in reception duties that the receptionist was already performing effectively.”
“The better question is: what is preventing the required result from being achieved?”
The Dublin consultant starts with the result the role must produce, then looks for what blocks it. Sometimes the blocker is the boss, and no tool or new hire can fix that.
Where they split
Gooden: judge each step by whether a human adds value. Routine requests need no person.
Kakkar: judge the tool by outcomes improved. If human review stays the same, the role wasn't replaced.
Collins: find what blocks the result first. Sometimes the blocker is the person in charge.
Where they agree: none of them starts with human versus agent. All three start with the result.
The common thread is a discipline, not a tool: name the result, then ask which resource earns each step.
My take
I like this reframe because it takes ego out of the decision. The tool, the hire and the boss all have to earn the job by the result.
It fits my argument that AI takes pieces of a job, not the whole thing. The outcome tells you which pieces.
So before the next hiring plan or agent pilot, write down the result the role exists to produce.
Your turn
What is preventing the required result from being achieved?
