Relay.app is shutting down. Free users lose access on August 15, 2026, and paying customers get until September 14, 2026. Export everything now: credentials, tokens, and anything else you haven't pulled out get deleted permanently once each wind-down window closes.
Relay.app is shutting down: the dates you actually need
Free-tier accounts lose access on August 15, 2026. Paying customers get an extra month, until September 14, 2026. Both dates come straight from Relay's own shutdown announcement on relay.app. A LinkedIn post from founder Jacob Bank the same week says the same thing, and a Hacker News commenter posted both dates on July 17, 2026.
Get the order right, because several recap posts have it backwards. If you read one claiming paying customers lose access first, ignore it and go with relay.app.
Why is Relay shutting down? It hasn't said. The announcement opens with “We have difficult news to share” and names no cause: no funding problem, no acquisition, nothing. Anyone telling you the real reason is guessing.
One thing to get straight before you go further: none of the four candidate destinations, not Zapier, not Make, not n8n, not Gumloop, has a native Relay importer. Every migration guide you find, this one included, is really a manual-rebuild guide. Some come with an AI co-builder attached. A couple come with a consulting upsell.
- Zapier
- Make
- n8n
- Gumloop
What to export before your access ends
Start with the export, not the research. Relay's docs confirm you can pull workflows, Sequences (Relay's term for a multi-step chain of connected automations), and MCP servers (Relay's connectors that let AI models call outside tools), all as JSON with AI prompts included. Run history exports separately. Tables (Relay's built-in data storage) export as CSV.
A handful of mechanics matter here, per docs.relay.app as of July 28, 2026. A full workspace export can take up to 24 hours for large workspaces. The emailed download link expires after 48 hours. Run-history CSV exports cap at 100,000 rows per export. And screenshots are a separate action, not bundled into the standard export.
(Export mechanics per docs.relay.app; refund timing per Relay's own announcement. Both vendor self-reported, as of July 28, 2026.)
The part people miss is the account deletion. Whatever you don't export is gone for good once the wind-down window closes, and Relay's stored credentials and tokens for every connected app get deleted at the same point, regardless of tier. That means re-authenticating each app from scratch, wherever you land next.
Paying customers get a real perk here: an extra 25,000 steps and 10,000 AI credits a month during the transition window, plus prorated refunds within five business days on annual plans. Both figures are vendor self-reported (relay.app, as of July 28, 2026) and restated, not independently verified, by Zapier and rills.ai. Worth knowing. Not worth building a decision around.
Do you actually need to migrate right now?
Exporting is step one. Whether you need to rebuild anything this week is a separate question, and no competitor covering this story will give you the honest answer, since every one of them is selling a destination. If you're a free-tier user with a couple of lightweight workflows, or a paying customer with 60 days on the clock, you don't have to commit to anything this week. Exporting today and deciding calmly over the next few weeks is a legitimate plan, not procrastination.
That changes if a workflow touches revenue or a customer-facing process. An invoice-reconciliation flow or a lead-routing pipeline that breaks silently shouldn't wait until the last week of August or September; rebuild those first.
Either way, the same rule applies: export today regardless of timeline, since it costs nothing and the credential-deletion clock is fixed no matter what you decide. Then pick a destination once, deliberately, not by clicking the first “free migration” offer that lands in your inbox.
How to pick where your workflows actually belong
Once you're ready to pick, know this going in: none of the four platforms worth considering, Zapier, Make, n8n, and Gumloop, is a like-for-like Relay replacement. Based on documented capabilities and verified user reviews, not a first-hand test, here's the rough shape of each.
Zapier has the broadest app-integration count, over 9,000 per its own figures, plus a Copilot feature that assists a rebuild rather than importing one outright. It's also the only one of the four with a named, built-in “Human in the loop” step, per Zapier's own comparison table. That's worth flagging, since Relay's approval steps (pausing a workflow for a person to review) were genuinely popular with users. Make and n8n reportedly need a manual workaround for the same effect, per a third party's characterization, so file that under a competitor's framing rather than settled fact. For a closer look at how it stacks up against Make specifically, see Zapier compared with Make for AI agent workflows.
Make offers a visual, logic-heavy canvas that several guides call the closest structural cousin to Relay's builder. n8n is self-hostable and more configurable, a good fit if you're comfortable going a layer deeper and don't want to hand your workflows to a single vendor again; our n8n vs Zapier comparison digs into that tradeoff in more depth. Gumloop leans into workflow and agent-building with native MCP server hosting, worth a look if your Sequences lean more on AI prompts than simple app-to-app hops.
Map the choice to what your workflows actually do. Simple linear automations don't need Make's branching logic. Real conditional branches probably want it. AI-prompt-heavy Sequences point toward Gumloop, and Gumloop against n8n for non-coders is worth reading if you're torn between those two. Avoiding single-vendor lock-in points toward n8n's self-hosted option. There's no winner to crown here, just a tool that fits the job in front of you.
Relay.app alternatives compared
| Platform | Closest fit to Relay's model | Migration path from Relay | Pricing model | Best for |
|---|---|---|---|---|
| Zapier | Broadest app coverage, Copilot-assisted rebuild, named human-in-the-loop step | Manual rebuild; no native import (“Not automatically,” per Zapier's own FAQ) | Task-based; Professional from $19.99/mo (zapier.com/pricing, as of July 28, 2026) | Widest connector library |
| Make | Visual, branching-logic canvas, closest structural cousin per several guides | Manual rebuild; no native import; free white-glove migration for “eligible” accounts through Sep 14, 2026 | Credit-based; migration page lists $16/$29 per month (Pro/Teams), but Make's main pricing page shows $21/$38 for the same 10,000-credit tier monthly. Unresolved discrepancy, check both | Real conditional branching |
| n8n | Self-hostable, more technical, deeper configuration control | Manual rebuild; more setup, no automatic refunds on cancellation | Per-execution; Starter $20/mo billed annually, 2,500 executions and 2,300 AI credits/month (n8n.io/pricing, as of July 28, 2026) | Avoiding single-vendor lock-in again |
| Gumloop | Workflow/agent-builder positioning, native MCP server hosting | Manual rebuild; no dedicated Relay migration guide exists yet | Credit-based; Pro $37/mo, 20,000+ credits, unlimited seats (gumloop.com/pricing, as of July 28, 2026) | AI-prompt-heavy Sequences, MCP-dependent workflows |
Note on Make's router and filter steps: Zapier's blog states these cost one credit each, while Relay's own blog calls the equivalent feature a free, built-in exception. The two disagree, so treat that cost as an open question until Make's own docs confirm it.
For the deeper mechanics behind tasks versus operations versus credits, that breakdown already exists elsewhere on the site.
Migrating without breaking things (a practical checklist)
Whichever platform you land on, the rebuild itself looks the same. No destination here imports a Relay workflow directly, and it's worth hearing that from three separate vendors instead of one. Relay's own docs say the JSON export “won't work through their native importers.” Zapier's own FAQ says it more bluntly: “Not automatically.” viaSocket's FAQ says “in most cases, not as a one-click import.” Same admission, three times over. Budget rebuild time, not import time.
This is a synthesis of documented migration practice, not a migration this team ran itself, but the order matters.
Export everything first
Workflows, Sequences and MCP servers as JSON, Tables as CSV, run history separately. Do it before you shortlist anything.
Triage by blast radius, not by effort
Rank workflows by how much damage a failure would cause, and rebuild the highest-risk ones first, not the easiest.
Test each rebuild in parallel
Run the new version alongside the Relay original while the wind-down window still allows it.
Re-authenticate every connected app
None of Relay's stored tokens carry over, so each integration needs a fresh connection in the new platform.
Delete nothing in Relay until the rebuild has run
Keep the original in place until the replacement has completed a real run successfully at least once.
Frequently asked questions
When exactly does Relay.app shut down for free users versus paying customers?
Free accounts lose access August 15, 2026. Paying customers lose access September 14, 2026. Both dates come from Relay's own announcement, corroborated by a Hacker News thread and a LinkedIn post from Relay's founder the same week. Competitor site alltomate.com adds a cutoff time of 11:59 p.m. Pacific on both dates, not stated on Relay's own page as fetched.
What does the Relay.app data export actually include?
Workflows, Sequences, and MCP servers export as JSON with AI prompts included. Run history exports separately, and Tables export as CSV. A full export can take up to 24 hours for large accounts, the download link expires after 48 hours, and run-history CSV exports cap at 100,000 rows.
What happens to my connected app credentials and tokens when Relay.app shuts down?
They're deleted at the end of your tier's wind-down window, along with the rest of your account. That's true no matter which platform you move to, so you'll need to re-authenticate every connected app from scratch.
Do I need to migrate to a new tool immediately, or can I wait?
Not necessarily. If your workflows are light or non-critical, exporting today and deciding over the next few weeks is reasonable. Workflows tied to revenue or customer-facing processes are the exception; rebuild those sooner.
Should I move my workflows to Zapier, Make, or n8n?
Depends on what your workflows actually do, not on which platform wrote the loudest migration post. Simple automations, branching logic, and self-hosting needs each point toward a different one of the three; see the framework and table above rather than a single crowned winner.
Do I need to know how to code to replace Relay.app?
No. Zapier, Make, and Gumloop are no-code rebuild paths, with Zapier's Copilot and Gumloop's builder assisting the process. n8n is more technical, but plenty of non-coders run it for workflows that don't need deep customization.
Will my Relay.app workflows keep running until the shutdown date?
Per Relay's own wind-down announcement, existing workflows keep running through your tier's deadline, with no earlier cutoff mentioned for automations already active.
Do paying Relay.app customers get a refund?
Yes, for annual plans, prorated and issued within five business days, per Relay's own announcement as of July 28, 2026.
For a deeper walkthrough on weighing a decision like this, there's a longer decision framework for picking any AI agent or automation tool elsewhere on the site. And if you'd rather not find out about the next platform shutdown from a panicked Reddit thread, a short newsletter on exactly this kind of change is a low-effort way to get ahead of it next time.
The real lesson: this is a vendor lock-in problem, not just a Relay problem
Zoom out for a second. Strip away the specific dates and this can happen to any single-vendor no-code platform you build on. Relay was, by most accounts, good at what it did. In the Hacker News thread on the announcement, one commenter called it “probably the best designed No-Code workflow builder out there,” and another wrote they were “gutted for the team, they really made an amazing product.” Being well-liked didn't stop the shutdown clock.
“Probably the best designed No-Code workflow builder out there.”
Which tool replaces Relay matters less than the habit you build from this: regular exports, and avoiding formats with no export path, so you're covered the next time a vendor closes, gets acquired, or pivots without warning. Before committing deeply to any platform again, ask yourself whether the company is likely to exist in its current form a year from now.
Export what you have today, take your time deciding where it goes next, and if you'd like a heads-up before the next shutdown announcement instead of after, that's what a short newsletter like this one is for.