Moderne

The trait that decides it is where the human effort goes. You review one change and it lands identically everywhere it applies, instead of reviewing three hundred pull requests that each differ slightly because a model improvised. For a platform team that has run a dependency migration by hand, that inversion is the whole product.

There is an IntelliJ plugin and a command line if you want to work locally, but this is bought by a department, not by a person. Pick it at estate scale. Pick Claude Code for a handful of repos.

The limit is the front end. Six language families are listed and the documentation is explicit that Java is where the support runs deepest, which means the promise degrades quietly as your estate diversifies. Nothing warns you that a service in a thinner language received a shallower transformation; it simply is not in the result set, and absence looks like success.

What it does right is the interface. The command line is modelled on version control, so engineers already know what a checkout, a run and a diff mean here before they read anything.

1. This is the architecturally serious answer to the hallucination problem. The model does not rewrite text; a deterministic recipe transforms a parsed tree that retains formatting and type information, and the language model's role is selecting and shaping the recipe. 2. The output is therefore reproducible, which almost nothing else on this board can claim.

2. No benchmark is published, and none is needed for determinism, though the selection step, where the model actually decides, is the part left unmeasured.

The commercial architecture is the reason to take this seriously. An open transformation library builds a standard and a community; the proprietary platform monetises the only place that library is hard, which is running it across an estate nobody can hold in one machine. Switching cost is the recipe corpus a customer accumulates, and that compounds.

Enterprise-only pricing signals large contracts and long cycles rather than land-and-expand. Likely acquirer: a code host or a static-analysis vendor that wants the parsers. Position: a durable asset, priced like one.

Contact sales as the only pricing is my least favourite sentence, and here it costs me a quarter before I can compare anything. What buys back the meeting is deployment: an air-gapped edition exists, which means the source never leaves our perimeter and the security review becomes a network diagram rather than a negotiation.

Runs are scriptable, so this becomes a scheduled job rather than a habit sixty people have to form, and that is the correct shape for a migration tool. Approved with conditions: the air-gapped edition, a fixed-term contract, and a measured pilot on our two largest repositories.

There is exactly one thing here I can use on my terms. It acts as a server, so tools like its search can feed an agent I wrote, and that inversion is genuinely useful because I get the structured context without adopting the client. Everything else is shut: proprietary licence, no key of mine, and no inference on my hardware.

A fork is impossible for the platform and unnecessary for the recipes, which live in a project I can already read. Grudging respect for shipping the server half.