Self-hosted node

Fully self-hosting an application means owning its upgrades, its migrations and its outages. Running only the work layer is a much smaller commitment: one service, on one machine, that the hosted control plane talks to.

The division is also economically honest. What a vendor genuinely pays for is compute, sending capacity, data and model tokens. A workspace supplying all four consumes almost nothing, which is why this arrangement can reasonably be free rather than discounted.

Mail leaves your mailbox over your infrastructure, so the reputation you build belongs to your domain. Crawling and enrichment run from your address space. Model calls use your key. And the data that would otherwise sit in a shared pipeline sits on your machine.

The trade is operational: you keep the node reachable and patched, and a node that is down is a queue that is not moving. It is a real cost, and it is a much smaller one than running the whole platform.

Does the free tier exist because the work moved to you, or because features were removed? The first is a coherent arrangement; the second is a trial with a longer expiry.

And can you leave with your data — a full export, in a format something else can read — without asking anyone? Data ownership that requires a support request is not ownership.

SalesShift's Self-Hosted plan is free and works exactly this way: you run the node, connect your own mailboxes and supply your own model key. The whole CRM is included — pipeline, contacts, tasks, activity history, sequences, quotes, invoices and e-signature — with unlimited stored contacts.