Function points in the age of agentic AI as a coding tool
Coding agents change execution speed. They can generate screens, APIs, tests or scripts in minutes. This acceleration doesn’t remove an older question: what do we really need to build, and what functional size does it represent for the business?
Counting code becomes less central
In a classic software project, we often discussed development days, lines of code or team velocity. With a coding agent, these benchmarks lose part of their meaning. The same feature can be produced faster, then fixed, rewritten or tested with AI’s help.
The volume of generated code is therefore not a good indicator of value. What matters more is the service delivered: data consulted, rules applied, transactions created, exports produced, checks performed, users served.
Function points speak the language of scope
The function point method has a quiet advantage: it looks at the software from the outside. It focuses on inputs, outputs, inquiries, logical files and interfaces with other systems. In other words, it measures what the system must do for the user and for the organisation.
This reading stays relevant when AI writes part of the code. The agent can speed up production, but it doesn’t decide on its own whether a business rule is simple, whether a flow must be audited, whether a piece of data must be historised, or whether an external interface makes the scope riskier.
A scoping tool, not a quoting machine
The point isn’t to turn every idea into a heavy calculation. For an SME, function points can serve as a light grid: how many user inputs, how many outputs, how many inquiries, how much internal data, how many external connections?
This grid helps compare two options. A very visible prototype can be functionally small. A less spectacular back-office can be denser because it contains many rules, exceptions and data exchanges.
What agentic AI really changes
The coding agent shifts the effort. It reduces part of the time spent writing, looking up syntax, creating a file structure or producing basic tests. In return, it raises the importance of scoping, review, acceptance testing and the quality of the instructions.
Function points can help prepare this work. They give a map of the scope before asking the agent to code. They also make review easier: every identified input, output or interface must have an expected behaviour, error cases and validation criteria.
Next step
Turn this reference point into a concrete project
If this topic resonates with a situation in your organisation, a short diagnostic lets us look at the process, the available data, the risks and the right initial scope.