What are the best Relay.app alternatives now?

Why are Relay.app alternatives suddenly urgent?

Relay.app alternatives are suddenly urgent because Relay is shutting down, and paying workspaces lose access on September 14, 2026.

Relay's own shutdown notice says free accounts have already reached their cutoff, while paying customers keep access through the final wind-down date. New sign-ups and upgrades have been disabled since the shutdown announcement.

That changes the question completely. Earlier, switching away from Relay was optional. A team could compare Make, Zapier or n8n, decide Relay still worked better, and stay where it was. Existing users now have to move.

Relay has at least made the exit cleaner than many SaaS shutdowns. Users can export workflows, Sequences and MCP servers as JSON, AI prompts as Markdown, run history and Tables as CSV files. The export also contains JSON for individual runs. According to Relay's shutdown documentation, stored credentials and authentication tokens disappear when the account is deleted.

So we are judging Relay alternatives under unusual conditions. Product quality matters, but migration speed, approval handling, pricing at the same workflow volume and the chances of facing another forced move all deserve more weight than they normally would.

Is there a true drop-in replacement for Relay.app?

Relay.app has no true drop-in replacement today; Make comes closest overall, while Zapier now comes surprisingly close on human review.

Relay mixed several things that usually live in separate products. It had the visual simplicity of a mainstream automation builder, AI steps from multiple model providers, structured data, reusable workflows and unusually polished human-in-the-loop controls.

Make covers the visual workflow side very well. Zapier now has far stronger native approval features than it had when Relay first became popular. n8n gives users much more control over complex logic and AI agents. Gumloop goes further into agentic work. Activepieces offers attractive run-based pricing and self-hosting.

Each alternative therefore asks former Relay users to give something up or change how they build.

A team using Relay mainly for predictable SaaS automations will probably find Make familiar. A team built around approvals may prefer Zapier. Technical users can gain more than they lose by moving to n8n. Someone who was already turning Relay workflows into semi-autonomous AI workers may find Gumloop more exciting than a literal Relay clone would have been.

The replacement market has improved a lot. We no longer need one product to reproduce every Relay feature perfectly. We need the product that reproduces the part of Relay our workflows actually depended on.

What did Relay.app actually do better than Zapier and Make?

Relay.app was unusually good at making human judgment a normal part of an automation, and that remains the hardest part of Relay to replace cleanly.

Relay offered four dedicated human-in-the-loop steps: approval, data input, tasks and path selection. When a workflow reached one of those steps, it paused until the assigned person acted. Relay could notify that person through email or Slack, set due dates, send reminders and even escalate overdue actions.

The important detail was how deep this went.

A manager could approve a workflow. Someone could fill in missing structured data. An employee could complete something that happened outside the software, such as calling a customer. A person could decide which path the workflow should take.

Relay also built human review directly into AI generation. Its documentation gave reviewers four choices after an AI step: approve the output, end the run, manually edit the output or re-prompt the AI and review the new version again.

That fit real business AI surprisingly well. A model could draft a sales email, extract information from a document or classify a customer request, while a person remained available when judgment actually mattered.

Zapier has closed a lot of that gap lately. n8n is strong when human approval controls an AI tool call. Lindy and Gumloop also let people supervise consequential agent actions.

Still, Relay set a high bar here. We should compare alternatives against that experience rather than simply counting how many integrations or AI models each product advertises.

Is Make the best Relay.app alternative for most teams?

Make is currently the best Relay.app alternative for most teams, especially now that Make has built a dedicated Relay migrator instead of leaving users to rebuild everything by hand.

That new migration tooling materially changes our judgment.

Make's Relay → Make Migrator can take a Relay workspace export and create paused Make scenarios plus new Data Stores. The current migrator automatically handles several common pieces of Relay workflows, including manual triggers, common schedules, HTTP requests, if/else logic, conditions, fixed delays and Relay Tables. It also supports a set of Gmail actions and Airtable's record-changed trigger.

Unsupported pieces remain visible as TODO modules. AI steps, many other connected-app actions, webhooks, forms, RSS triggers, loops, sequence calls and more advanced delays can still require manual work. That makes this a partial converter rather than magic, but a partial converter is already much better than rebuilding every flow from a blank canvas.

Make also has the breadth needed after migration. Its current site lists more than 3,000 prebuilt apps, and its Relay migration guide puts the figure above 3,500. The company says more than 400,000 organizations use Make across more than 200 countries.

Pricing remains easy to understand. At 10,000 monthly credits, Make currently lists Core at $12 per month, Pro at $21 and Teams at $38. A module action generally consumes one credit.

AI is becoming a bigger part of the product too. Make now combines ordinary scenarios with AI Agents, hundreds of AI-related app integrations and Maia, its conversational building assistant.

Human approvals remain the main compromise. Make can build serious approval flows, but the experience usually involves more explicit workflow construction than Relay's dedicated approval primitives.

For the average Relay workspace, we still think Make gives up the least while adding a much larger ecosystem. The new Relay migrator strengthens that lead considerably.

Is n8n better for complex Relay.app workflows?

n8n is the better Relay.app replacement when workflows are long, API-heavy, AI-heavy or technically complex.

The biggest difference is freedom. n8n combines a visual builder with JavaScript, Python, HTTP requests, databases, custom APIs, advanced branching and self-hosting. A Relay workflow that had started bumping into the edge of what a no-code builder could comfortably express may actually become easier to maintain after moving to n8n.

Its pricing model also suits long workflows. n8n's current Starter cloud plan costs €20 per month when billed annually for 2,500 workflow executions. Pro costs €50 for 10,000 executions. Those executions can contain unlimited steps.

That distinction becomes huge once an automation contains ten, twenty or fifty actions. n8n counts the workflow execution while Make generally counts module actions and Zapier generally counts successful billable tasks.

n8n has also become much more serious about AI oversight. Its documentation now includes human review for AI tool calls, and individual integrations can pause an agent before a sensitive action runs. Gmail, for example, supports sending a message and waiting for approval, while more complicated approval logic can be built with n8n's Wait node and branching tools.

Company risk has fallen sharply too. In May 2026, n8n announced a strategic investment from SAP at a $5.2 billion valuation. At the same time, n8n said it had reached 1.7 million monthly active builders and more than 1,400 enterprise customers. SAP is also embedding n8n inside Joule Studio.

Pipedream belongs in this same technical bucket. It offers JavaScript, Python, custom APIs and thousands of connectors, with billing based on compute time rather than the number of steps. Pipedream is even more developer-oriented, so we would choose it mainly when APIs and code are the real workflow rather than occasional escape hatches.

For an operations team that wants something immediately friendly, Make wins. For a technical team with difficult Relay workflows, we would take n8n.

Is Gumloop better if Relay.app was mostly for AI?

Gumloop is the better Relay.app alternative when the real goal is to move from fixed workflows toward AI agents.

Gumloop has made that direction very explicit lately. Its current product and pricing presentation gives Agents top billing while still supporting conventional Workflows.

The Pro plan currently starts at $37 per month with 20,000 credits, unlimited agents, more than 35 models, unlimited seats and teams, shared credentials and bring-your-own API keys.

Gumloop agents can use connected apps, workflows, custom MCP servers, a code sandbox, web search and web fetching. A conventional Gumloop workflow can itself become a tool that an agent decides to call. That creates an interesting migration path: deterministic pieces of an old Relay system can survive as workflows while an agent decides when to use them.

Gumloop has also built much stronger controls around those agents. App Rules can restrict individual tool calls, such as preventing an agent from creating a Linear ticket without required labels. Teams can control which apps and specific agent tools different users may access.

The company has gained more credibility too. Gumloop raised a $50 million Series B led by Benchmark in March 2026 and says companies including Shopify, Ramp, Gusto, Samsara, Instacart and Opendoor use the platform.

The trade-off is product direction. Gumloop is pushing toward agentic automation, so teams mainly looking for a conventional deterministic workflow builder should think carefully about whether that is where they want to go.

For Relay users who wanted more autonomy from their AI, that direction is a feature. For teams that mostly want reliable trigger-action business processes, Make, Zapier or n8n feels safer.

Is Zapier safer for Relay.app integrations and approvals?

Zapier is the safer Relay.app choice when integration coverage and straightforward approvals matter more than workflow economics.

Zapier's biggest advantage is still enormous. Its current directory has more than 9,000 apps, compared with more than 3,000 on Make and roughly 700-plus on Activepieces. For a company with dozens of niche SaaS tools, that difference can eliminate custom API work.

Zapier has also fixed one of the reasons Relay once felt better.

Human in the Loop is now a native Zapier tool on paid plans. A Zap can pause, ask someone to approve or decline a request, allow that reviewer to edit the submitted fields and then use the edited values in later steps.

Zapier expanded the feature again in 2026 by allowing requests to go to people outside the Zapier account. That gets much closer to the way real approval workflows operate with clients, managers and external partners.

AI workflows are broader too. Zapier now combines normal Zaps, AI steps, Agents, MCP and thousands of integrations in the same ecosystem.

The problem is still task economics. Zapier Professional currently starts at $19.99 per month, and each successful billable action generally consumes a task. Five billable actions executed 10,000 times can therefore mean roughly 50,000 tasks.

Many built-in steps such as Filters, Formatter, Paths, Delay and Tables do not count as tasks, so the exact number depends on the flow. Even so, long workflows tend to expose the difference between Zapier's task model and run-based products such as n8n or Activepieces.

We would choose Zapier when a missing integration could break the migration or when simple human approvals are central. Otherwise, Make gives the average Relay user more attractive workflow economics.

Is Lindy better for email and assistant workflows?

Lindy is the better Relay.app replacement when the automation behaves more like an executive assistant than an operations pipeline.

This distinction is easy to miss because Lindy now has a proper workflow builder. Users can create triggers, actions, conditions and agent steps for repeatable jobs such as lead enrichment, document processing, email handling and follow-up.

Where Lindy becomes interesting is the point where fixed logic ends.

Agent steps let the AI decide what action to take next instead of forcing us to predefine every branch. Lindy's own documentation is refreshingly clear about the trade-off: agent steps cost more and can be less reliable than standard actions, so Lindy recommends using them when the next step is genuinely uncertain.

Human oversight is built in as well. Actions with outside consequences, such as sending an email, changing a record or creating a calendar event, can ask for confirmation. The agent pauses and waits for the human decision.

That combination fits inbox management particularly well. An agent can understand an incoming email, find information, prepare a reply and wait before sending. The same pattern works for meeting preparation, scheduling and routine assistant work.

Current pricing starts at $29.99 per user per month for Plus with 3,000 credits. Pro costs $99.99 with 15,000 credits and Max costs $199.99 with 35,000.

We would avoid picking Lindy simply because a Relay workflow contains some AI. Make, n8n and Zapier remain better general workflow platforms. Lindy becomes compelling when the thing we are building would naturally be described as an assistant.

Is Activepieces the best cheap Relay.app alternative?

Activepieces is the strongest cheap Relay.app alternative we found for long workflows, and its current pricing is unusually favorable at high run counts.

The key is how Activepieces counts usage.

Its Plus plan currently costs $16 per month when billed yearly and includes 10,000 credits. One complete flow run consumes one credit. Every ordinary step inside that flow is free.

That means a flow with two normal steps and a flow with twenty normal steps consume the same base credit when each runs once.

AI uses extra credits. Activepieces currently lists its fast model at two credits, smart model at ten and frontier model at twenty. Bringing your own AI key costs one credit for the AI action.

The integration library is no longer tiny either. The live catalog currently shows more than 700 integrations, with the catalog page exposing more than 760 pieces when we checked it. Activepieces says more than 300,000 builders use the product.

Self-hosting adds another reason to look closely. Activepieces' core is under the MIT license, and the company says the Community Edition can run with no cap on runs, users or flows. Agents, Chat and a number of collaboration and governance features sit outside that free core, so the self-hosted Community Edition does not reproduce the entire paid product.

For a Relay user with a few simple automations, saving $10 or $20 a month is barely relevant. Once thousands of long workflows run every month, Activepieces' pricing model becomes a serious competitive advantage.

Which Relay.app alternative handles human approvals best?

Zapier currently gives former Relay.app users the closest ready-made approval experience, while n8n gives technical teams more control over what happens around the approval.

Relay remains our benchmark because its four human-in-the-loop step types covered more than simple yes-or-no approval. Zapier has now caught up on much of the practical workflow: it can collect missing data, request approval, let reviewers alter the submitted values, pause execution and route notifications through email, Slack or another Zap.

n8n works especially well when the approval controls an AI agent or a complex process. Its Wait logic and human-review tools give builders more freedom to design what happens before, during and after the decision.

Lindy makes approvals very natural around assistant actions. Gumloop is getting increasingly sophisticated at controlling what agents may do, although its model is more about guardrails and permissions than recreating Relay's classic human-task system.

Make remains capable, but approvals still require more assembly than we would like for a product replacing Relay.

Platform	Human-review experience	Best use
Zapier	Native approval and data collection, editable fields, external reviewers	Closest simple Relay replacement
n8n	Approval steps, waits and AI tool-call review	Complex or technical workflows
Lindy	Confirmation before consequential agent actions	Assistant-style workflows
Gumloop	Agent permissions, rules and controlled tool use	Autonomous AI workflows
Make	Flexible approval flows built from scenario logic	Teams already committed to Make
Activepieces	Human input, tasks and approval pieces	Low-cost workflows

Which Relay.app alternative is cheapest at real volume?

Activepieces and n8n are usually the cheapest Relay.app alternatives for long workflows because they charge around complete runs rather than every ordinary step.

Consider 10,000 executions of a workflow with five normal billable actions.

Activepieces consumes about 10,000 base flow credits because the internal ordinary steps are free. Its current Plus plan includes exactly 10,000 monthly credits for $16 when billed yearly.

n8n records 10,000 workflow executions. Its €50 Pro plan currently includes that volume with unlimited steps inside each execution.

Make would generally record around 50,000 module actions in this example. Zapier could also reach roughly 50,000 tasks if all five actions are billable, although some Zapier-native utility steps are free.

At twenty actions per run, the gap becomes much larger: 10,000 runs are still 10,000 base runs on Activepieces or n8n, while an action-metered system can move toward 200,000 actions.

AI complicates every comparison. Activepieces charges separate AI credits, Gumloop uses its own credit system, Lindy's consumption depends on the work being done, and model usage can alter the economics elsewhere too.

So the cheapest platform depends on the shape of the workflow. Still, we found one very clear pattern: longer deterministic workflows favor run-based billing much more than advertised entry prices suggest.

Example: 10,000 runs × 5 ordinary actions	Approximate metered usage	What drives cost
Activepieces	10,000 base flow credits	Complete flow runs
n8n	10,000 workflow executions	Complete workflow runs
Make	About 50,000 credits	Module actions
Zapier	Up to about 50,000 tasks	Successful billable actions
Gumloop	Varies	Credits, AI and agent behavior
Lindy	Varies	Credits and task complexity

Will your Relay.app integrations survive the move?

Most common Relay.app integrations can be rebuilt elsewhere, but identical app logos do not guarantee identical triggers, actions or field behavior.

Zapier has the lowest raw integration risk. Its directory currently contains more than 9,000 apps. Make advertises more than 3,000. Activepieces now has more than 700. Pipedream says developers can connect to thousands of apps and exposes more than 10,000 prebuilt triggers and actions.

n8n has fewer turnkey integrations than Zapier, but its HTTP, code and API flexibility makes the raw directory count less restrictive for technical teams.

Gumloop uses another route increasingly often: MCP. Agents can connect directly to supported MCP integrations, custom MCP servers and workflows that themselves wrap multiple services.

For common products such as Gmail, Slack, Notion, Airtable and major CRMs, we would expect migration to be straightforward on several platforms.

The problems appear deeper inside each integration. One builder may expose a trigger that another lacks. Search operations can return different fields. Pagination can behave differently. A convenient Relay action may require an HTTP request elsewhere.

Credentials also need to be reconnected. Relay's shutdown documentation explicitly says stored credentials and authentication tokens are deleted with the account, so exporting a workflow does not transfer the authenticated connection.

We would therefore audit important workflows at the trigger-and-action level rather than using integration counts alone.

Platform	Current integration reach	Migration implication
Zapier	9,000+ apps	Lowest risk for niche SaaS apps
Make	3,000+ apps	Very broad coverage for normal business workflows
Pipedream	Thousands of apps, 10,000+ triggers/actions	Strong when APIs and code are acceptable
n8n	Large native catalog plus HTTP/code	Stronger than raw app count suggests
Activepieces	700+ integrations	Good mainstream coverage, smaller long tail
Gumloop	Native apps plus MCP and custom MCP servers	Flexible, especially for agent workflows

How hard is it to migrate Relay.app workflows now?

Relay.app migration is much easier now than it was when the shutdown was first announced because Make can directly convert part of a Relay export.

This is probably the most important recent change in the entire Relay alternatives comparison.

Relay exports a useful amount of structure. A workspace export contains workflow JSON, plain-language prompts, run history, individual run data and Table data. That already gives other systems much more information than they would get from a screenshot.

Make now goes one step further with its dedicated Relay migrator. Users upload the Relay export, connect Make, select what they want to migrate and receive paused scenarios plus Data Stores for supported content.

The supported list covers enough common workflow logic to make a real difference: schedules, manual triggers, HTTP requests, branches, conditions, fixed delays, Relay Tables and selected Gmail and Airtable operations.

The unsupported list tells us how far we still are from one-click migration. AI steps, most other connected-app actions, webhook and form triggers, RSS, Gmail filtering details, loops, sequence calls and advanced delays still require work.

Make then points users toward Maia for finishing TODO modules, while its Relay migration program also offers migration support for eligible accounts during the shutdown window.

We would use that tooling before attempting a manual recreation elsewhere, even if Make is only being evaluated. It gives us a concrete benchmark for how much of a Relay workspace can be translated automatically.

For n8n, Activepieces, Gumloop or Zapier, more rebuilding will usually be involved. That can still produce a better final system, particularly when the old Relay workflow had grown messy.

The practical order is simple: preserve the Relay exports first, migrate one representative workflow, compare outputs and failure cases, then use what we learned to rebuild the rest.

Which Relay.app replacement is least likely to disappear?

n8n, Make and Zapier look safest if vendor continuity worries you after Relay.app, while Activepieces gives you the strongest escape hatch through self-hosting.

Zapier has operated since 2012. Its current company materials say 3.4 million businesses use the platform, with more than 9,000 app integrations. Whatever we think about pricing, Zapier is far beyond the stage where a small customer base makes the product fragile.

Make says more than 400,000 organizations use its platform and has been part of Celonis since 2020. That gives Make a very different risk profile from an independent early-stage automation startup.

n8n has recently moved into the same serious-company category. Its SAP investment valued the company at $5.2 billion, and the reported 1.7 million monthly active builders plus 1,400 enterprise customers give us much more evidence of scale than we had a couple of years ago.

Activepieces offers a different kind of protection. Its automation core is MIT licensed and can be self-hosted. A company with the technical capacity to operate it therefore has more control over continuity than it would have with a cloud-only service.

Pipedream has also changed category since Workday completed its acquisition. Workday described Pipedream as an integration platform for AI agents with more than 3,000 prebuilt connectors when discussing the acquisition in its fiscal-year results.

Gumloop is younger, but a $50 million Benchmark-led Series B and customers such as Shopify, Ramp and Gusto reduce the concern considerably.

None of those facts guarantees permanence. Relay's closure is a useful reminder that a working SaaS product can disappear for reasons unrelated to whether customers liked it.

So we would now give portability real weight. Exports, APIs, self-hosting and readable workflow definitions can save more pain than another small feature advantage if a platform changes direction later.

What are the best Relay.app alternatives now?

Make is the best Relay.app alternative for most teams today, with n8n second and Zapier third.

Make wins because the migration problem has changed. Its dedicated Relay migrator now converts supported parts of a Relay export directly into paused Make scenarios, while Make itself gives users more than 3,000 integrations, visual workflows, AI Agents and enough depth to replace both simple and fairly complex Relay setups.

n8n would be our first choice for technical teams. Unlimited steps per workflow execution, strong AI tooling, human review, code, APIs and self-hosting make it particularly strong once workflows become complicated.

Zapier now deserves third place. Its 9,000-plus integrations were always an advantage, but native Human in the Loop has fixed a major weakness for Relay users. Allowing reviewers to approve, decline, edit data and even participate from outside the Zapier account makes Zapier much closer to Relay than it was before.

Gumloop comes next because it offers something more ambitious than straightforward replacement. The product is clearly moving toward agents. We would choose Gumloop when that direction matches what we want rather than simply because it resembles Relay.

Activepieces is the value pick. One credit per complete flow run, 10,000 monthly credits for $16 on Plus, more than 700 integrations and an MIT-licensed self-hosted core give it unusually strong economics.

Lindy works best for assistant-shaped workflows around email, meetings, calendars and follow-up. Pipedream makes more sense for developers who see APIs and custom code as normal parts of automation.

The choice is therefore much clearer than it first appears. A typical Relay team should start with Make. A technical team should start with n8n. A team whose migration depends on obscure integrations or polished approvals should check Zapier. AI-agent-heavy users should test Gumloop. High-volume users should run the numbers on Activepieces before paying action-based prices elsewhere.

Rank	Relay.app alternative	Best for	Our judgment now
1	Make	Most Relay.app users	Best overall replacement, especially with the new Relay migrator
2	n8n	Complex and technical workflows	Best combination of control, AI and run-based economics
3	Zapier	Integrations and human approvals	Much closer to Relay now that Human in the Loop has matured
4	Gumloop	AI agents	Best when the migration is also a move toward autonomous agents
5	Activepieces	Price and self-hosting	Strongest value option for long, high-volume workflows
6	Lindy	Email and AI assistant workflows	Excellent in its niche, less universal as an automation platform
7	Pipedream	Developers and APIs	Powerful technical option, less natural for the typical Relay user