Qoder

The trait that decides it is surface coverage. The same product exists as a desktop application, a command line and plugins for the editors your team already opened this morning, so adopting it does not require anyone to change where they work. In practice that is the difference between a tool a team tries and a tool a team keeps.

Breadth also means each surface is less finished than a dedicated one. Pick it when your team is split across editors. Pick Windsurf when everyone already agreed on one.

The exposure grows with the runtime. A mode designed for extended autonomous work, combined with repository write access and no recorded containment, means the window in which something can go wrong is measured in hours rather than in prompts. Nothing published describes a checkpoint, a budget or a stopping rule for a quest that has stopped making progress.

What it does right is externalising state. Rules and memory are declared artefacts a team can read and edit, rather than an opaque profile the product accumulates about you.

1. The context strategy is worth naming. Instead of searching the codebase per request, the product builds a persistent description of it and consults that, which trades freshness for consistency and makes token cost predictable across sessions. 2. It is also a cache, so staleness is the failure mode nobody documents.

2. No benchmark is published and no measurement compares this approach with retrieval, so the design choice is defensible in principle and undemonstrated in practice.

Renaming an existing assistant into a new agentic brand is a cheap way to convert an installed base, and 2.6 million installs is a starting position, not a growth target. Behind it sits a cloud business that gains from every hour of inference this product schedules, which means the parent does not need this line to be profitable to justify continuing it.

Moat: distribution plus first-party models. Likely path: tighter binding to the parent's cloud rather than an exit. Position: durable, and geographically constrained in ways your procurement team will raise before you do.

Individual tiers are published at $20, $60 and $200, which multiplies out to between $1,200 and $12,000 a month for my headcount, but every one of them is metered in Credits, so the tier is a floor and not a forecast. Team and enterprise terms are quoted separately, which means a call before I can compare anything.

There is no unattended execution, so this never becomes a pipeline step and its value stays whatever sixty individuals get from it. Not yet: I need a quoted cap and a written data handling answer first.

Two things stop me dismissing this. My own key works on the free plan rather than being reserved for an enterprise contract, which inverts the usual gating and means I can drive it on an account I already hold. MCP servers plug in, so the tools I maintain are reachable from inside a product I cannot read.

Everything else is shut. Proprietary licence, no fork, and the row records no local inference, so the weights are never mine. Rope, not ownership.