Your graph, on infrastructure you choose
The short answer
Two arrangements exist beyond the standard plans. A dedicated instance gives your organisation a graph database that holds nothing else. Running on your own infrastructure keeps CodeMesh and the graph inside your network, talking to the model endpoint you already run. Both are set up with us rather than switched on in the console, so the next step on this page is a conversation, not a checkout.
The default. Your graph lives on a database instance alongside other customers, and the separation between them is enforced by CodeMesh rather than by separate infrastructure.
What does not change
Your developers keep the same extension, the same panel and the same agent connection. Nobody learns a different tool because the graph moved, and there is no separate enterprise build of the client to keep in step with the public one.
Administration stays in the same console: repositories, per-member usage, storage, cost and which repositories were indexed. An organisation admin sees those figures for their own organisation and never the contents of the repositories behind them.
Where you land. Whether every repository synced clean, how large the graph has grown, how many seats are in use, and which agents currently hold read access.
Example workspace (Northgate Systems). The reduction shown is the published benchmark's 95.1%.
Which one you want
The question is almost always about the second row of the diagram above, so it is worth being blunt about the difference.
- Dedicated is the right answer when the concern is other customers. On the standard plans your graph shares an instance and the separation is enforced by CodeMesh rather than by separate infrastructure. A dedicated instance removes that sharing without changing anything else.
- Your own infrastructure is the right answer when the concern is the network boundary itself: code that contractually cannot leave, or a model endpoint you are required to use. This is a larger engagement, because we are handing you something to run rather than running it for you.
How access is scoped
OAuth 2.1 with PKCE, read-only by construction, tenancy from the token, and one-click revocation.
What leaves a developer's machine
The full list, including the parts most vendors leave out: commit author emails, uncommitted work, and how long logs are kept.
The standard plans
What Free, Basic and Team include, and the per-seat model Team uses, in case the standard plans are enough.
Can we start on a standard plan and move later?
Do you sign a DPA?
Who do we actually talk to?
Tell us which arrangement you need
Number of developers, and whether the concern is the shared instance or the network boundary. We reply with a real quote.