First, the part nobody's saying
If you're reading this, you probably got the July 16 email. That's a bad week, and we're not going to pretend otherwise.
Here's the other thing we're not going to pretend: Relay got something right that most of this category still hasn't. It put a person in the middle of the automation. A step where a human checks the work before it goes out. If you picked Relay, you picked it at least partly for that — you wanted automation you could trust because you could see it and stop it.
That instinct was correct. Keep it.
And picking a tool that later shut down wasn't a judgment error, either. 5,600 AI startups closed between January 2025 and March 2026. You didn't choose badly. You chose in a category that's still shaking out.
So here's our claim, and it's a narrow one: the instinct that made you choose Relay — see the work, approve the work — is the thing Sterling is built around. And we're really good at execution. Sterling doesn't hand you a canvas to rebuild your flows on. It does the job itself, in your Slack, and checks with you before anything leaves your business or can't be undone.
The rest of this page is a moving plan.
The two deadlines
You have two dates.
August 15 — Relay's free plans stop working. That's 27 days from today. September 14 — paid plans stop working. That's 57 days.
After that, every recurring job Relay was running becomes your job again. The Monday report. The follow-up chasers. The lead that's supposed to land in the CRM without you touching it. Each one quietly returns to your plate on a Sunday in September.
And there's a second, sneakier cost: the rebuild. The obvious move is to recreate every workflow in another builder — and anyone who's done that once knows where it goes. "i could have done the entire task i'm trying to automate with the time i've spent trying to get zapier" (a Trustpilot reviewer). You'd spend August wiring flows in a new canvas, learning a new tool's quirks, re-breaking things a field rename will break again.
A Reddit owner said the quiet part: "small businesses do not want to have an automation platform at all they want a system that keeps on running by itself."
You don't want to migrate workflows. You want the work to keep happening.
The move isn't to another builder
That's the reframe: don't replace your workflow tool. Replace what the workflows were for.
Sterling is an AI coworker that lives in your Slack. There is no canvas. There are no flows to rebuild, because there are no flows — there are jobs. You type the job in plain English: "Every Monday at 8, pull Shopify and Klaviyo, build the weekly report, and draft the win-back campaign." Sterling connects to your stack, does the work, and posts the finished output in the thread.
Then the part you'll recognize: the internal jobs you delegate — build the doc, log the record — Sterling just does, then reports back. But anything that leaves your business or can't be undone — a send to customers, an external post, a delete — waits for your go. You reply "go" in the thread. It executes. It's logged.
Sterling does everything up to the send. The send is yours.
Week three is where automations go to die. Sterling doesn't have flows to break — it has a job.
Where your Relay workflows go
If Relay ran it, you can say it to Sterling in one Slack message. The common moves:
| Your Relay workflow did this | You tell Sterling this |
|---|---|
| Scheduled report, assembled and sent weekly | "Build the Monday report from Shopify and Klaviyo, every Monday at 8." |
| Drafted an email, waited for a person to approve it | "Draft the win-back campaign in our voice. I'll say go before it sends." |
| Captured leads and pushed them to the CRM | "Pull the new leads, clean them up, and log them in the CRM." |
| Synced data between two apps and flagged mismatches | "Cross-check Shopify against QuickBooks and flag anything that doesn't line up." |
| Chased people for follow-ups on a schedule | "Turn this thread into action items and chase the follow-ups until they're done." |
Every row ends the same way: the finished work shows up in Slack, and anything that gets sent to your customers, charged, or deleted waits for your go. Recurring jobs run on schedule — and that gate applies every single time.
Relay.app and Sterling, side by side
| Relay.app | Sterling | |
|---|---|---|
| Status today | Shutting down — announced July 16, 2026; new signups disabled | Live and taking new workspaces today |
| Service end date | Aug 15, 2026 (free plans) / Sep 14, 2026 (paid plans) | No contracts — you decide when it ends, from the dashboard |
| What it is | Workflows you build in its own web app | A coworker you delegate to inside the Slack you already have open |
| Migration path | Workflows end with the service | Type the job in plain English — no flows to rebuild |
| The approval step | Approval steps you added into a workflow | Nothing gets sent, charged, or deleted without your go, in the thread where the work happened — plus off/view/full access per person, per tool |
| Ecommerce jobs | General workflow automation | Shopify/Klaviyo audits, ROAS comparisons, win-back drafts, QuickBooks reconciliation — running today |
| Setup | — | Add to Slack, connect your apps, delegate. Minutes, no code |
| Pricing | New plans unavailable — signups are closed | $50/month = 20,000 credits for the whole workspace, no per-seat charges; $25 tops up 10,000 more |
What you get
Nothing to rebuild — the migration is a sentence
Other landing spots hand you an empty canvas and a two-week learning curve. Sterling takes a Slack message. The job Relay ran as a twelve-step workflow, you describe in plain English once — Sterling connects to your apps and runs it. Moving your recurring work is an afternoon, not an August.
The approval instinct you already have, sharpened
You chose a tool with a human check for a reason. A Reddit owner set the bar for the whole category: "if there's no exception queue, audit log, and big red pause button, it's not automation, it's just a faster way to lose money." Sterling is built as exactly that: the job runs in a Slack thread you can see, nothing gets sent to customers, charged, or deleted without your go, every action lands in the audit log, and pause is a dashboard button. Plus something Relay didn't shape this way — each teammate's access is off, view, or full, per tool, set by you.
Recurring work that keeps running — with the gate every time
Schedule the Monday report once and it shows up every Monday, built. The win-back draft, the CRM updates, the reconciliation — on autopilot. But autopilot never means unsupervised: nothing gets sent to your customers, charged, or deleted without your go, run after run. "The best automation is the one people can actually keep using without calling someone every week" (a Reddit owner). No consultant. No canvas. A coworker with a job.
A price you can reason about
$50 a month gets your whole team a coworker — 20,000 credits that cover the jobs you actually delegate: the weekly audit, the campaign drafts, the CRM updates. No per-seat charges. Run low? Sterling pings you in Slack at 80% — before it matters. A top-up is $25 for 10,000 credits — decided by you, not discovered on an invoice. And if credits run out, Sterling pauses and tells you. It never surprise-bills you.
The migrate-this-week plan
Three steps. Do the first one today; you have until September 14, and you should use every day of the overlap.
1. Today — add Sterling to your Slack and connect your apps. Minutes, no code, no IT contractor. Shopify, Klaviyo, your ads accounts, QuickBooks, your CRM — 3,000+ apps. 2. This week — hand over your most important Relay job first. The Monday report is the best test, because you know exactly what Relay produced and exactly how long it takes by hand. Watch the finished version arrive. Say go. 3. Before September 14 — move the rest while Relay still runs. Keep Relay live and run Sterling next to it. Compare outputs side by side. Cancel nothing until Sterling has done every job at least once. The overlap costs you $0 — that's what the 20,000 free credits are for.
Frequently Asked Questions
How fast can I be moved? Setup is minutes: add to Slack, connect apps, type the first job. Most Relay setups map to a handful of delegations — plan an afternoon for the handover and a week of running both tools side by side. You have 57 days of paid-plan runway. That's more than enough, if you start this week.
Your first 20,000 credits are free — sized to cover the transition.
That's enough to hand Sterling your recurring Relay jobs, run them next to Relay's output, and judge the finished work on your own stack — not a demo store. If Sterling's Monday report isn't worth $50 a month to you, stop there. You've spent nothing, and you've lost nothing but an afternoon you'd otherwise spend evaluating workflow builders.
After the trial: $50/month = 20,000 credits, whole workspace, no contracts, cancel from the dashboard. No sales call, no scoping doc, no onboarding project.
Worst case, you got a free second opinion on your own numbers before September 14. Best case, the shutdown email turns out to be the reason your Mondays got shorter.
P.S.
The deadlines don't move: August 15 for free plans, September 14 for paid. Everything else about this decision is reversible — Sterling is month to month, the trial is free, and Relay keeps running while you compare. Add Sterling to your Slack today, hand it the Monday report, and keep your one job: the go.