Relay

What it is

Relay is a source-agnostic migration substrate. It reads the system a customer is leaving, maps it onto the system they're adopting, rehearses the move against a simulated copy, and only then touches anything real. The same engine works regardless of which system is on either end.

It is built for the two people who feel the pain: the vendor trying to land an enterprise customer without burning a year of forward-deployed engineering, and the enterprise that wants to switch without betting the company on a hand-built data project.

How it works

1. 01 Scout
   An agent reads the source system the way a new teammate would and reports back what's actually there — objects, relationships, custom fields, and which of them are still in use. Read-only, and it verifies its own coverage so nothing is silently missed.
2. 02 Plan
   It maps the source onto the destination field by field, and surfaces every gap it could not place with confidence instead of guessing. You approve the plan before anything moves.
3. 03 Simulate
   The migration runs first against a generated copy of the source, so the failures happen somewhere harmless. Same inputs always produce the same rehearsal.
4. 04 Cut over
   The approved plan executes in batches against the real destination, with checks along the way and a reversible trail behind it.

Built for the parts that scare people

• Human approval gates. A person signs off on what changes. Nothing happens as an opaque black box.
• A complete audit trail. Every decision and every action is recorded and verifiable — which matters most exactly where the data is most regulated.
• Nothing guessed. Anything the system can't place with confidence is escalated to you, not quietly assumed.