Relay.app Is Gone, and So Is Everything It Remembered About Your Work

Relay.app is gone. Free accounts were deleted on August 15, 2026, paying accounts on September 14, and that was the end of it: per the company's own shutdown notice, accounts and all associated data were permanently deleted. Founder Jacob Bank rejoined Google as VP of Product for Chrome, several colleagues went with him, and TechCrunch reported it as a shutdown with a set of hires, not an acquisition.

The migration conversation that followed was all about workflows: export your triggers, rebuild your steps, find a new vendor. That is the easy part. Workflows are logic, and logic can be retyped.

What cannot be retyped is everything Relay.app remembered about how those workflows behaved in the real world. Run history. Which inputs worked and which broke. The corrections applied after a bad run. The patterns your AI steps learned about your customers, your data, your edge cases. All of that accumulated knowledge lived inside the platform, and it was permanently deleted along with the accounts.

This is what run-to-run amnesia looks like at platform scale. You have probably been living a smaller version of it: your scheduled agent wakes up blank every morning and re-learns the same lessons from scratch. A shutdown is that problem wearing a suit. Same disease, bigger funeral.

The most valuable thing on the platform was never the workflows

Think about what an automation actually accumulates over time. The workflow definition is the recipe: trigger on this, send that, branch here. That part fits in a JSON export and is genuinely replaceable.

The valuable part is the ledger of how the recipe behaved. Which runs succeeded and which failed, and why. What you changed after each failure. The decisions an AI step made on ambiguous inputs and whether they turned out right. The patterns in your data that only showed up after a thousand runs. Most of that was never written down anywhere except inside the platform's run history.

Make's migration guide for ex-Relay users confirmed the export included workflows, sequences, run history, tables, and MCP servers as JSON or CSV. But that export was only available while accounts were still active, and it was a one-shot deal during a 60-day wind-down when most operators were focused on rebuilding, not archiving. The default outcome was predictable: people exported the recipe and left the knowledge behind. Then September 14 passed, and the knowledge stopped existing.

This was never a Relay-specific flaw

It is tempting to file this under "choose better vendors," but the structure is everywhere. Your n8n execution history lives inside n8n's walls. Your Zapier task history lives inside Zapier. Your Make scenario logs live inside Make. Wherever your automations run, your agents' memory is your vendor's data, and your vendor's data has an expiry date you do not control.

Relay raised $32.2 million in May 2024 (TechCrunch) and launched in 2021 aiming to become the new Zapier. Five years of funding, talent, and momentum ended with a shutdown notice. The lesson is not that venture-backed platforms are fragile. It is that the average lifespan of an automation tool in your stack is shorter than the lifespan of the knowledge it holds. Companies get acquired, pivot, sunset products, or die. When they do, anything stored only inside them dies with them.

Every migration is a memory wipe

Shutdowns are the dramatic version. The everyday version is every vendor switch, every tool replacement, every "we are moving from Zapier to Make" project. Each one quietly deletes accumulated operational knowledge, even when nothing is technically deleted. The new system knows the steps. It does not know what the old system learned.

Operators who rebuilt on a new platform after Relay discovered this in a quiet way: their rebuilt workflows worked, but they started from zero. The new system did not know the patterns the old one had learned. Every mistake the old system had stopped making came back. That is the cost of storing memory inside a tool instead of alongside it.

Put memory in a layer that survives its tools

The way out is architectural. Memory should not live inside any single tool in your stack. It should live in a layer of its own, one that every tool reads from and writes to, and that survives any individual tool dying, being replaced, or getting acquired. This is the same argument operators have long made for portable databases over vendor-locked datastores, applied to agent context.

A hosted memory service over MCP is one way to get there. Your agents, across n8n, Make, Zapier, or whatever you migrate to next, read and write the same shared memory through an open standard protocol. The memory does not live in any of those tools, so none of them can take it with them when they die.

Vilix AI works this way: cloud-hosted with zero infrastructure for you to manage, the same memory available to every AI tool over MCP, and full conversation history rather than just extracted facts. It is free forever with a 7-day Pro trial that asks for no credit card. Your data stays portable: export everything or delete it anytime. The point is not the product name. The point is the architecture: memory that outlives the tools that use it.

FAQ

Could Relay users have saved their run history?

Technically, yes, and briefly. The shutdown notice gave free users 30 days and paying users 60 days to export, and Make's guide listed run history among the exportable data. In practice, exports during a wind-down get triaged: the recipe gets exported first while the history gets left behind. After September 14, the question became permanently unanswerable.

Is this only a shutdown problem?

No. Shutdowns are just the version with a deadline. Every migration between automation platforms resets the learned layer of your operation, whether or not the old vendor deletes anything. If you cannot take your agents' accumulated knowledge with you, you did not own it; you were renting it.

What belongs in an external memory layer?

Decisions and their outcomes. Corrections after failures. Learned patterns about your data and customers. The full run history that lets an agent understand what "normal" looks like for your operation. The test is simple: if a new tool cannot function without it, it does not belong inside the old tool.

Relay's shutdown is a story about a well-funded product losing to platform gravity. For automation operators, it is also a warning: the memory your workflows depend on should never live in a place that can shut down. Put it in a layer you control, connect it over an open protocol, and the next time a tool in your stack dies, you will lose the tool, not what it knew.

Start free at Vilix AI. Free forever, no credit card.