Replacing Relay.app: Rebuild the Job Before Picking Another Tool
Relay.app is no longer a live option. Its official notice says it shut down in September 2026 and permanently deleted accounts, workflows and run history. If you need a Relay.app alternative for a Slack workflow, start with the job and records your company still has. Do not build a migration plan around exporting a deleted account.

TL;DR
6 min read- Reconstruct what the job should do from retained company records and the people responsible for it.
- Check unfinished work against the destination. An approval message is not proof that its action happened.
- Separate the old backlog from new work before switching a replacement on.
- Rebuild the smallest useful path first. Do not replace a deterministic step with an AI decision unless the job needs one.
Use a workflow tool for known routing and execution steps. Evaluate Mio for the company-context preparation that leads up to a decision, such as a draft account report in Slack. It is not an importer for Relay or a way to recover deleted execution history.
Sources checked November 18, 2026. The records and workflow below are fictional examples, not a customer recovery or a tested migration product.
First, separate the workflow from its missing history
A remembered prompt is not a complete workflow definition. Ask the current owner to identify the event that started the job, the records it read, the person who reviewed it and the system where the final change appeared.
Use material the company legitimately retained: an internal runbook, source records, approval messages, destination records or an earlier export already in your possession. Do not assume any of these exists or is complete. Mark what cannot be established.
The immediate output should be a short reconstruction:
When a project update was ready, the workflow gathered the approved facts, asked the account lead to review the proposed client note, then posted the accepted version to the agreed destination. We know the source and reviewer. We do not yet know whether two older notes were delivered.
That last sentence is useful. Replacing it with "the old workflow handled these" would hide the work that still needs checking.
Reconcile three items before replaying anything
Imagine a fictional approval queue with three client notes. The company still has a project tracker and selected approval messages, but no Relay run history.
| Item | Evidence still available | Safe next step |
|---|---|---|
| A: project 104, note v2 | The destination contains note v2 with the matching project reference and recipient. | Record that exact note as delivered. Do not send it again. |
| B: project 108, note v3 | The approval refers to note v3. The only destination receipt found is for note v2. | Do not mark v3 delivered from the older receipt. Have the owner check whether v3 was sent and whether it is still the version to use. |
| C: project 112 | The tracker says a note was planned. No draft or approval can be established. | Leave the state unknown. Ask the owner whether a note is still needed before recreating it. |
Keep one row per intended action, with the project reference, destination, draft version, approval reference and delivery receipt. A receipt for an earlier version answers a different question. If the retained evidence cannot distinguish the versions, record the gap rather than matching on similar wording.
This is a proposed reconciliation method, not automatic duplicate prevention. Match records using stable identifiers where you have them. Similar wording or a nearby date alone may not be enough.
If an action could change a customer record, spend money or contact someone, unresolved state is a reason for the responsible person to investigate. It is not an invitation to rerun everything and clean up later.
Keep that backlog separate from the new workflow. An old approval may refer to a different version, recipient or business situation. Do not silently carry it into a rebuilt action.
Then choose a replacement for each kind of work
A replacement need not reproduce every old step in one product.
Known triggers and routing
If the job is a form, a fixed branch and a notification, start by checking a workflow tool. Slack's Workflow Builder guide describes workflows, templates and connector steps. Check the actual steps, plan and account permissions needed for your job.
If you need to own a more explicit agent-and-tool workflow, n8n documents human review before selected AI tool calls, including review through Slack. That is a candidate capability to test, not a promise that an old Relay workflow imports unchanged.
Contextual preparation
If the time-consuming part is reading project discussion, finding the latest facts and drafting a reviewable update, evaluate an AI employee for that portion.
Mio's weekly recap workflow prepares a Slack draft from channel history and connected sources. You could use that pattern to prepare a project review from the records you have selected. It does not establish arbitrary Relay feature parity or recover anything deleted by Relay.
A bounded starting request:
From these approved project records and the receipts we retained, draft an internal review. Separate delivered work, approved work with unverified delivery, and items whose state is unknown. Link each classification to the evidence. Do not send messages, change records or infer a missing approval.
The owner must check those classifications. The draft helps organize the reconstruction; it does not make the evidence complete.
Restart with a clear boundary between old and new
Pick a cutoff and record which items remain under manual review. Test the rebuilt process on harmless new material first, using a controlled destination and an authorized reviewer.
Verify one end-to-end result: the correct input, the proposed output, the human decision where needed, and the actual destination receipt. Test denial as well as approval if the new setup includes a review step. A draft-only process should not gain permission to send merely because its first draft looks good.
Before enabling the normal trigger, check that another automation has not already taken over the same job. Name the person who owns exceptions. Put the accepted setup and completion evidence in a recurring-workflow runbook.
The goal is not to recreate an unavailable screen. It is to restore a useful job with an owner who can tell what happened.
Try Mio in Slack if the missing part is preparing the company-context draft. Use a workflow platform for the execution steps that need one, and keep uncertain historical actions out of automatic replay.
FAQ
Keep exploring
Related articles

Comparison
Mio vs Airtable Omni: Do You Need a Work App or a Shared Coworker?
Evaluate Airtable Omni when the job is to build or operate a structured app around your business data. Evaluate Mio when the job is to bring existing company context and recurring team work into Slack. The useful distinction is the system you want to maintain, not whether either product can use AI.

Comparison
Mio vs n8n: Delegate the Job or Own the Workflow?
The useful distinction is who builds, maintains and recovers the workflow. Both AI agents and human approvals can exist on either side.

Guide
Will Your Slack AI Brief Still Run When the Installer Leaves?
A Slack app can stay installed after the person who added it leaves. A passing run before offboarding is only a rehearsal. To trust the handover, verify a normal scheduled run after the old access has ended, using a fresh fact and the intended reviewer's receipt.
Mio is the Slack-native AI employee that already knows your company, connects to 3,000+ tools, and turns shared context into work. Just @mio, it's handled.