# n8n alternative: Elaichi vs n8n > Elaichi as an n8n alternative: workflows, approvals, shared tables and an AI agent under per-person governance, compared plan by plan. **TL;DR** Elaichi is an n8n alternative that puts connectors, workflows, shared tables, dashboards and the Elaichi Agent under one set of rules. Your own AI, from Claude, ChatGPT, Cursor or any MCP client, drafts each workflow and test-runs it on real data with every write held back. Each run acts as its owner. n8n is the better pick for self-hosting, custom code and per-execution billing. ## Why teams look for an n8n alternative Elaichi is the n8n alternative for teams whose workflows and AI questions reach the same apps. It puts both under one set of rules: every run acts as a named person, and a person approves each risky step. Workflows, collections, dashboards, knowledge bases and the Elaichi Agent come with the Black plan. Here is how it usually starts. Jake Morgan, your operations lead, built forty n8n workflows that sync tickets and chase invoices. Then support starts to ask Claude about escalations. Finance asks ChatGPT about overdue accounts. Now the same apps have a second set of credentials in a second place. A team that searches for an n8n alternative usually wants one thing: the same rules over the workflows and the AI questions. n8n has a different center. It is a workflow builder first, with code steps and a free Community edition you can self-host ([n8n pricing](https://n8n.io/pricing/), checked November 2026). Self-hosting means you run the software on your own servers. Each one wins in different places, and this post names them. ## Connectors, workflows and data in one place, with any AI Elaichi keeps connectors, workflows, AI, collections and dashboards in one workspace. One set of rules covers all of it. The team brings the AI it already uses. The chain is short: - A workflow calls connectors from a catalog of 650+. It writes rows to a collection, which is a shared table. - A dashboard reads that collection, or live data from a connected app. - A knowledge base answers the team's questions. - Claude, ChatGPT, Cursor or any MCP client signs in at https://api.elaichi.ai/mcp and uses the same Elaichi operations. It builds, tests and reads all of it. MCP (Model Context Protocol) is the standard way an AI assistant calls tools in other apps. The Elaichi Agent works beside these clients. It runs in the console, and when someone tags @Elaichi in Slack. It uses the organization's own AI keys, one per provider: Anthropic, OpenAI, Fireworks AI, OpenRouter or an OpenAI-compatible gateway. The Agent and workflow agent steps offer the same models. So model costs stay on your own provider account. Before the Agent writes, it shows an approval card with "Deny", "Allow once" and, when the action can be remembered, "Always allow". Its prompt-injection check looks for hidden instructions in what tools return, and it can stop a write. Neither the Agent nor a workflow browses or fetches the web. One place means one set of answers. A restriction is a rule that limits which connectors and tools a person may use. Restrictions are enforced at every place a member can reach a connector or tool, including listing, connecting, advertising, executing, the final outbound-request check and opening a stored tool file. So a workflow step and a chat question meet the same rules. The same audit log, the record of who did what, records each call. It also records where the call came from: the app, an MCP client, the API or the console. The same people answer approvals. One bill covers seats plus pooled usage. ## Elaichi and n8n side by side Elaichi governs every run by the person behind it. n8n gives a builder code steps, an HTTP Request node and a choice of cloud or self-hosting ([n8n pricing](https://n8n.io/pricing/), checked November 2026). | | Elaichi | n8n | |---|---|---| | Workflows | Up to five triggers of seven types: manual, cron, one-time, webhook, Slack, collection row, another workflow's run | Schedule, webhook and app triggers; JS and Python code steps; HTTP Request node | | AI agent | Agent steps in workflows; the Elaichi Agent in the app and in Slack | AI Agent node with tools; Agents in Preview | | Tables and dashboards | Collections up to 1,000,000 rows and 1 GB each; dashboards over collections, live app data or runs | Data tables, 200 MiB per instance by default; Insights over workflow runs | | Governance per user | Each run acts as its owner, under the owner's restrictions | Fixed credential by default; end-user credentials Enterprise only, in Preview | | Governance per role | One role per member; custom roles and per-tool restrictions from Gold | Project Admin, Editor and Viewer; custom roles Enterprise only | | Audit | Audit log on every plan; export to your own Datadog comes with the Black plan | Audit logging and log streaming Enterprise only | | Approvals | Approve step to named people or teams, in the console or Slack | Human review on AI Agent tools; Slack Send and Wait for Response | | Hosting | Hosted in an EU, US or APAC region, chosen at signup | n8n Cloud in Frankfurt, or self-hosted anywhere | | Connectors | 650+ connectors | 2,000+ integrations, by n8n's count | | Price | Black plan: $24 per user per month (USD list), plus run minutes past 1,000 a month | Per execution; Starter 20 euros a month billed yearly for 2,500 | The n8n column comes from n8n's own [pricing page](https://n8n.io/pricing/), its [integrations catalog](https://n8n.io/integrations/), its [roles docs](https://docs.n8n.io/administer/manage-users-and-access/set-permissions-and-roles-rbac/see-available-roles) and its [data tables docs](https://docs.n8n.io/build/work-with-data/data-tables), checked November 2026. ## Your own AI writes and tests the workflow before it goes live In Elaichi, your team's own AI builds the workflow. It proves the workflow on real data before anyone publishes it. Claude, ChatGPT, Cursor or any MCP client does this through the elaichi__automation__ tools. The Elaichi Agent, in the console or in Slack, gets the same operations. It can draft, validate, test-run, publish and turn on a workflow. It asks the person every time before it publishes or turns one on. The loop has five steps, and each one has its own workflow operation. 1. **Draft.** The AI writes the workflow as a draft with draft__create. It changes the draft with draft__update or draft__patch. A draft saves even while it has validation issues. 2. **Validate.** draft__validate checks the definition without saving it. It names each problem at its step. 3. **Test run.** dry_run runs the latest draft against real data. The console calls it "Test run (changes nothing)", in the Run now dialog. Tools that read run for real, as the person who starts the test. Tool calls that write are only recorded as what would have been sent. So are collection writes, state writes and knowledge writes. A notify step sends nothing. An approve step reports what it would ask and carries on. 4. **Replay a webhook.** A webhook is a message an app sends to a web address when something happens. A webhook trigger has its address while the workflow is still a draft. Deliveries to it are kept as samples, up to five for seven days. A test run can replay one as the trigger's payload. 5. **Publish and turn on.** Publishing turns the draft into a numbered version. It is refused while validation finds errors. Turning the workflow on is a separate step. An older version can be restored later. A test run costs no run time. So the AI can try a draft as many times as the job needs. No workflow operation answers an approval or reveals a webhook secret. So the AI that builds a workflow cannot approve its own steps. Dashboards follow the same pattern. Edits land in a draft, and publishing makes the draft the version everyone sees. ## How do workflows run in Elaichi? A published Elaichi workflow starts from its triggers and acts through connectors. Each run acts as the workflow's owner, with the owner's connections and the owner's restrictions. A workflow starts from up to five triggers. There are seven trigger types: - a manual run - cron, a repeating schedule - a one-time schedule - a webhook - a Slack message or reaction - a row changing in a collection - another workflow's run finishing There is no trigger for an event inside a connected app. A new Salesforce deal starts a workflow only if Salesforce sends a webhook to the workflow's own address. A workflow holds up to 50 steps. Steps call connected tools, branch, loop, wait, write to a collection, run code or hand a task to an agent step. Elaichi does not browse, search or fetch the web for a workflow. So every call reaches an app through a connector. Runs stay correct when people change. The owner's restrictions apply while a run goes, not only when it starts. Elaichi re-checks the owner about every minute. If they can no longer act, the run fails and the workflow pauses. The console says it plainly: "This workflow runs with its owner's access." The [workflows guide](/blog/elaichi-workflows/) walks through a build from trigger to version. Retries are careful. A step can be retried up to five attempts in all. But a step that writes is never sent again when its outcome is uncertain. It ends as outcome_unknown, for a person to check. So a ticket is not created twice. Runs stay in the workflow's Runs tab for 30 days. A failed run can run again from the failing step. ## Who approves a step before it runs? In Elaichi, a named person or team approves. An approve step asks the people it names. It also asks the workflow's owner, unless the step leaves the owner out. It can also bar the person who started the run from approving. The request reaches approvers on the Approvals page in the console, in Slack and by email. A Slack card has Approve and Reject buttons only when no destructive step, such as a delete, waits behind it. Otherwise the approver opens the console. A request expires after 72 hours by default. A rejection cancels the run unless the step is set to carry on. Only a person can answer. No AI client, no Agent and no workflow can approve or reject a step. So the AI never approves its own work. n8n handles approvals from the agent side. Its human review can apply to all tools on an AI Agent node, or to selected ones ([n8n docs](https://docs.n8n.io/build/integrate-ai/ai-examples/human-in-the-loop-for-tools), checked November 2026). Its Slack Send and Wait for Response operation records who responded. But its default link buttons let anyone with the link respond ([n8n docs](https://docs.n8n.io/integrations/builtin/app-nodes/n8n-nodes-base.slack/approvals), checked November 2026). ## Where results live: collections, dashboards and knowledge Elaichi stores what workflows produce in collections and shows it on dashboards. Knowledge bases keep what the team knows. All three sit in the console, under Automate and Connect. A collection is a shared table, or a set of up to 20 tables. It holds up to 1,000,000 rows and 1 GB. Row history can be turned on for a collection. Then any earlier version of a row can be restored. A dashboard charts a collection, live data from a connected app, or workflow runs. A collection widget is read for each viewer as that viewer. A public link shares a copy of a dashboard, not live access. An admin turns public links on first. A knowledge base holds short text documents that workflows and the Agent search. Search is keyword search, not semantic search, which would match by meaning. A document a workflow writes is saved as untrusted. It stays that way until a person in the console marks it reviewed. n8n's data tables serve a lighter job. n8n describes them as suitable for light to moderate storage. All tables in an instance are limited to 200 MiB by default ([n8n docs](https://docs.n8n.io/build/work-with-data/data-tables), checked November 2026). Its Insights dashboard reports production executions, failures and time saved. That is a view of the workflows, not of business data ([n8n docs](https://docs.n8n.io/administer/observe-and-log/track-usage-with-insights), checked November 2026). ## How Black governs each person and each role Elaichi gives every member one role. Restrictions then narrow what each role or person may reach. Gold already includes custom roles, per-tool restrictions, SSO with SAML or OIDC, and SCIM. SSO is one company sign-in. SCIM adds and removes users from your identity provider. Black adds everything on this page on top. A restriction names which connectors and which individual tools a target may reach. It targets a role or one member. A Black workflow obeys its owner's restrictions. So restricting a tool for a role also restricts it in the workflows of people in that role. A restricted tool is left out of the tools Elaichi offers a client and cannot be called. Search still names it, flagged restricted, with no schema. The person can file an access request for an admin to decide. A role or restriction change takes effect within about two minutes. Removing a member works faster: removing or suspending a member revokes every live grant in the same transaction as the membership change. The audit log keeps one entry for each connected-tool call that reaches execution (a call refused earlier writes none). Each row names the acting member, the tool, the connection the call reached and how the call was approved. Elaichi keeps 90 days of audit history. For a longer record, export to your own Datadog comes with the Black plan. For key custody, customer-managed keys in AWS KMS come with the Black plan. n8n puts its per-person tools higher in its price list. End-user credentials let a workflow use the triggering person's account. They are Enterprise only and in Preview. The default Fixed credential uses the same account for everyone ([n8n docs](https://docs.n8n.io/administer/manage-credentials/end-user-credentials), checked November 2026). Custom roles, audit logging and log streaming are Enterprise only. SSO starts on Business ([n8n pricing](https://n8n.io/pricing/), checked November 2026). The [per-user permissions comparison](/blog/n8n-ai-agents-per-user-permissions/) goes deeper on n8n's credential model. ## What the Black plan costs, and how n8n counts the same work Elaichi's Black plan charges per person, plus the minutes workflows spend working. n8n charges per workflow execution, however many steps it has ([n8n pricing](https://n8n.io/pricing/), checked November 2026). Black lists at $24 per user per month, or $18 per user per month billed yearly ($216 a year). These are US-dollar list prices, and [/pricing/](/pricing/) shows your own. Usage past a free monthly allowance costs extra. The allowance is 1,000 run minutes and 1 GB-month of storage, pooled across the workspace. A GB-month is 1 GB kept for a month. Past the allowance, a GB-month costs $5 and a run minute costs $0.02. Only working time counts. Waiting for a person, a timer or a called workflow is free, and test runs cost nothing. Guests, Billing Admins and Auditors take no paid seat. AI apps never count as users. Gold, without workflows, lists at $15 per user per month in USD, or $10 per user per month billed annually. n8n's cloud plans start at Starter: 20 euros a month billed yearly for 2,500 executions. Pro is 50 euros for 10,000. Business is 667 euros a month billed yearly for 40,000 executions, and is self-hosted only. Enterprise is priced by sales. Every plan includes unlimited users and unlimited active workflows. Starter and Pro are capped. Business overages are 4,000 euros per extra 300,000 executions ([n8n pricing](https://n8n.io/pricing/), checked November 2026). The two units reward different shapes of work. n8n's unit favors a long workflow that runs often, because n8n counts one execution however many steps it has ([n8n pricing](https://n8n.io/pricing/), checked November 2026). Black's unit favors a team where many people need governed access. Each seat also covers that person's AI client and the Agent. ## Where n8n is the better choice n8n wins on hosting control, code and breadth, by its own [pricing page](https://n8n.io/pricing/), checked November 2026. If those decide it for you, n8n is the right tool. - **Self-hosting.** The Community edition is free on your own servers. Business and Enterprise self-host too. Data stays wherever you host n8n. Elaichi is hosted only. - **Code and raw HTTP.** Every n8n plan includes JavaScript and Python code steps, and custom HTTP and GraphQL requests. Self-hosted n8n also runs bash scripts and custom nodes. Elaichi has a code step but no web fetch. - **Per-execution billing.** One execution covers a whole workflow with unlimited steps. Every plan includes unlimited users. - **Breadth.** n8n lists 2,000+ integrations, plus a fallback HTTP Request node for anything else. - **Fair-code source.** n8n's source is available under its Sustainable Use License, a fair-code license. n8n itself does not call the project open source. Each point comes from n8n's [pricing page](https://n8n.io/pricing/), [integrations catalog](https://n8n.io/integrations/) and [license page](https://docs.n8n.io/n8n-community-license/community-license), checked November 2026. Say a workflow touches one app, and no person asks an AI about that app. Then a single n8n instance may be all you need. The [automation versus MCP access](/blog/n8n-automation-vs-mcp-access/) breakdown covers that split. ## Switch to Elaichi if Switch when your AI users and your workflows act on the same apps and need one set of rules. These are the hard signals. - Your workflows should act as a named owner, under that owner's restrictions, not as one shared account. - A person, not an AI, must approve a step before it writes, and your approvers work in Slack. - People on Claude, ChatGPT, Cursor or any MCP client already need governed access to the apps your workflows touch. - You want workflow output in shared tables and dashboards that respect each viewer's access. - Long approval waits should cost nothing, because only working minutes are billed. Every new workspace starts on a 14-day Black trial with no credit card. Every Black feature is on: workflows, collections, dashboards, knowledge bases and the Elaichi Agent. During the trial, the free usage allowance is a hard stop, so nothing bills by surprise. A workspace that buys nothing moves to the free Personal plan with nothing deleted. A Gold workspace upgrades to Black in place, from Settings. The [Black plan overview](/blog/elaichi-black-plan/) lists everything the plan includes. Check the apps your workflows need in the [connector catalog](/connectors/), and compare plans on [/pricing/](/pricing/). Try the full Black plan for 14 days, with no credit card. Build a workflow, add an approve step and chart the results. [Start your free trial](https://app.elaichi.ai/signup) ## FAQ ### Is Elaichi a good n8n alternative? Yes, when your workflows touch the same apps your team reaches from Claude, ChatGPT, Cursor or any MCP client. Elaichi puts one set of roles, restrictions and audit entries over both. Each Elaichi workflow runs as its owner, under that person's restrictions. An approve step waits for a named person. n8n fits better when you need to self-host, write custom code nodes or call any HTTP API from a workflow ([n8n pricing](https://n8n.io/pricing/), checked November 2026). ### How does Elaichi pricing compare with n8n pricing? The Black plan lists at $24 per user per month in US dollars, or $18 billed yearly. Usage past a free monthly allowance costs extra. The allowance is 1,000 run minutes and 1 GB-month of storage, pooled across the workspace. Only working time counts, so waiting for an approval is free. n8n charges per workflow execution, with unlimited steps and unlimited users. Its Starter plan is 20 euros a month billed yearly for 2,500 executions ([n8n pricing](https://n8n.io/pricing/), checked November 2026). ### Can n8n workflows run with each person's own account? Only with end-user credentials, and n8n limits them. n8n lists them as Enterprise only, in Preview, OAuth only and limited to team projects. It also advises against relying on them in production ([n8n docs](https://docs.n8n.io/administer/manage-credentials/end-user-credentials), checked November 2026). The default Fixed credential uses one account for everyone who runs the workflow. In Elaichi, a workflow runs as the person who owns it, with that person's connections and restrictions. ### Does Elaichi have a free trial of Black? Yes. Every new Elaichi workspace starts on a 14-day Black trial with no credit card. During the trial, the free usage allowance is a hard stop. Workflows pause at 1,000 run minutes, and shared tables stop growing at 1 GB, until you buy Black. A workspace that buys nothing moves to the free Personal plan, and nothing is deleted. ## Read next - [Automation platform vs MCP access: which job?](/blog/n8n-automation-vs-mcp-access/) — Scheduled workflows and ad hoc assistant calls are two jobs. How to split automation platform vs MCP access, and when one tool covers both. - [n8n AI agents per-user permissions and audit](/blog/n8n-ai-agents-per-user-permissions/) — n8n AI agents per-user permissions come down to whose credential a run holds: what n8n documents, and what Elaichi's MCP endpoint requires. - [Zapier MCP alternative for a whole company](/blog/zapier-mcp-alternative/) — Elaichi is a Zapier MCP alternative: one organization endpoint with role restrictions and an audit log. What each costs, and when Zapier fits.