IntelliCode

The trait that decides it is that there is no decision. It arrives with the editor, it costs nothing, and it makes the completion list less stupid by putting the member you probably wanted at the top instead of alphabetically. Nobody will ever describe this as exciting and nobody has to justify it either.

What it will not do is anything the rest of this board does. Pick it as a baseline that is already running. Pick GitHub Copilot the first time you want a change applied across files.

The honest complaint is that there is nothing to complain about, and that is itself the finding. Repeated-edit suggestions apply within a single file and only in one language, and the vendor directs readers to its subscription product for anything wider. A team that expected more has misread a page rather than been misled by one.

What it does right is scope discipline. No shell, no autonomy, no file writing outside what you accept, so the failure mode is a wrong suggestion, and a wrong suggestion is caught by the person who asked for it.

1. The mechanism deserves note precisely because it is unfashionable. Observing a developer's edits and generalising them is inference over a tiny, high-quality local sample, which produces suggestions grounded in the session rather than in a corpus. 2. It applies within one file and one language, which bounds the claim honestly.

2. No benchmark is published, and the ranking claim, most likely member rather than alphabetical order, is the kind of assertion that could be measured easily and has not been.

There is no business here and there is not meant to be. Shipping it inside the workloads keeps the editor competitive at no marginal cost, and every user who runs into its ceiling becomes a qualified lead for the paid assistant from the same vendor. That structure makes it durable in a way funded startups are not, because ending it would weaken something much larger.

Moat: the editor's installed base, which is not this product's asset but guarantees its survival. Likely path: gradual absorption into the paid tier's onboarding. Position: free infrastructure, not a strategic choice.

This is the shortest procurement note I will write this year. Nothing to purchase, nothing to provision, and edit tracking happens locally, so the residency question that consumes weeks elsewhere does not arise. Sixty developers already have it whether or not I approve it.

What I get for that is small: modest ergonomics, no throughput number I could defend, and nothing that appears in a pipeline or a report. It is a floor rather than a strategy, and it does not compete with the tool I will actually have to buy, which is a separate line in the same budget. Approved.

There is nothing to bend. The model is the vendor's own and cannot be replaced, no provider key applies, and there is no repository to read or fork. My entire configuration surface is a checkbox in an installer.

The one thing I will credit is that it does its work on my own machine rather than shipping every keystroke somewhere, which is more than several paid competitors manage. That is a privacy property, not an ownership one, and it is the only reason this is not the lowest score I have given.