Why we built read and write integrations, not read-only, and what the trade-off actually cost us

Every early version of dolv was read-only. Pull the data, surface the insight, recommend the action.

The feedback was consistent: useful but incomplete.

Teams were getting good analysis. Then they were leaving dolv to go execute in five other tools. Then they were coming back to report what happened. The loop was not closing.

The decision to rebuild as read-and-write was not obvious. Write access requires more trust, more permission scoping, more careful handling of credentials. It also means the product can make mistakes that have real consequences.

We made the trade deliberately. Every integration now supports both directions. Reads run immediately. Writes that touch customers go through the approval queue. Credentials are scoped to one workspace and never surfaced in the UI.

The product is more useful and more responsible because of the write capability, not in spite of it.