Make.com Alternative: Elaichi vs Make

When is Elaichi the Make.com alternative you want?

Elaichi is the Make.com alternative for a company whose automations and AI clients must answer to one set of rules. Each workflow runs as a named owner, and a person signs off before a risky step.

Make (make.com) is a visual automation platform. You draw scenarios, Make's name for workflows, on a canvas. You connect apps and pay in credits, per its pricing page, checked November 2026.

The picture changes once AI agents do the work. Somebody has to decide which person, role and agent may reach which app. Elaichi puts workflows, approvals, shared tables, dashboards and the Elaichi Agent behind one set of restrictions and one audit log. The same rules govern Claude, ChatGPT, Cursor or any MCP client. MCP (Model Context Protocol) is the standard way an AI assistant calls tools in other apps.

Workflows, collections, dashboards, knowledge bases and the Elaichi Agent come with the Black plan. It lists at $24 per user per month.

How do Elaichi and Make compare side by side?

The two differ most on who a run acts as, how a person signs off, and what the bill counts.

Question	Make	Elaichi
Workflows	Scenarios on a visual canvas, with routers, filters and error handlers	Built as drafts, test-run, published as numbered versions, then turned on
Triggers	App triggers, webhooks and schedules down to 1 minute on the Make Plan	Manual, cron (no more often than every 5 minutes), one-time, webhook, Slack, collection row and another workflow's run
AI agent	Make AI Agents, marked Beta in the pricing table	The Elaichi Agent, in Elaichi and in Slack, and agent steps inside workflows
Tables and dashboards	Data stores; Make Grid is a visual map of scenarios; Analytics Dashboard on Enterprise	Collections (shared tables), and dashboards that chart a collection or live app data
Governance per role	6 organization roles and 6 team roles; custom roles on Enterprise	8 built-in roles over 58 permissions, plus custom roles
Governance per person	Role assignment per team	Allow and restrict rules on a role or one member, per connector or per tool
Audit	Audit logs on Enterprise, kept 12 months	Audit log on every plan, kept 90 days; export to your own Datadog comes with the Black plan
Approvals	Manual approvals for AI agents	Approve steps answered by a person, in Elaichi or in Slack
Hosting	AWS, EU and North America zones	EU, US or APAC, chosen when the workspace is created
Apps	3,500+ apps	650+ connectors
Price	Per credit; unlimited users	Per seat, plus run minutes and storage past the free allowance

The Make column follows Make's pricing page, audit logs, custom roles, teams and AI agents pages, all checked November 2026.

How does Elaichi put connectors, workflows, data and any AI in one place?

Elaichi keeps 650+ connectors, workflows, collections, dashboards and a knowledge base in one workspace. One set of rules covers them all. Your team brings the AI it already uses. Claude, ChatGPT, Cursor or any MCP client connects at https://api.elaichi.ai/mcp as its one address. The Elaichi Agent works beside them, in Elaichi and in Slack.

Here is how the pieces fit:

• A workflow calls connectors and writes rows to a collection, which is a shared table.
• A dashboard reads that collection, or live data through a toolbox's read-only tools. A toolbox is a set of tools picked for one job.
• A knowledge base holds the documents a workflow searches.
• Any of those AIs builds, tests and reads all of it through the same operations.

The public API reference documents workflows, collections, dashboards and knowledge bases too. So a team can build on them with an API token.

The Agent and workflow agent steps run on your own keys, one key per provider. The providers are Anthropic, OpenAI, Fireworks AI, OpenRouter or an OpenAI-compatible gateway. Both pickers offer the same models.

One set of rules covers every path. The same roles and restrictions apply to each connected-tool call. That holds whether the call starts in Claude, in the Agent or in a workflow running as its owner. Calls from the Agent and from MCP clients land in one audit log. Approvals gate writes in the Agent and in workflows. One bill counts seats plus pooled usage.

Make puts its audit logs, single sign-on and custom roles on the Enterprise plan, per its pricing page, checked November 2026.

How does your own AI write and test a workflow before it goes live?

In Elaichi, the AI your team already uses writes the workflow. It checks the workflow and tries it on real data before anything is published.

The loop has five stops:

1. Draft. The AI creates a draft over MCP with the elaichi__automation__ operations, such as draft__create and draft__patch.
2. Validate. draft__validate checks the definition without saving it. The AI can then fix what it reports.
3. Test run. A dry run over MCP, or "Test run (changes nothing)" in the console, runs the latest draft against real data. Tool calls that write are recorded as what would have been called, and never sent. So are collection writes, knowledge writes and state writes. A notify step sends nothing. An approve step reports what it would ask.
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 still in the 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, then turn on. Publishing makes the draft a numbered version. Turning it on is a separate step. An older version can be restored later.

Tools that only read run for real during a test, as the person who starts it. So the AI sees the real records it will work on. A test run costs no run time, so trying a draft again costs nothing. Dashboards follow the same pattern: edits land in a draft until someone publishes.

How does a workflow run in Elaichi?

A workflow in Elaichi runs as the person who owns it. It uses that person's connections, access and restrictions.

You open it from Automate > Workflows. A workflow starts from up to five triggers of seven types:

• a manual run
• cron, a repeating schedule
• schedule once
• a webhook
• a Slack message or reaction
• a collection row change
• another workflow's run

There is no trigger for an event inside a connected app. That app's webhook starts the workflow instead.

A workflow holds up to 50 steps. Steps call tools, transform and filter data, branch, loop, wait and run code. They also write a collection, search a knowledge base, ask an approver or hand a task to an agent step. Every step can carry a condition, a retry policy, a timeout and an error mode.

A step retries up to five attempts. It waits as long as an app's Retry-After asks, which is the app's own request to wait before trying again. A step that writes is never sent twice when its outcome is uncertain. It ends as outcome_unknown, for a person to check.

Restrictions apply while the run goes, not only at the start. So a change reaches a run already under way. If the owner leaves or loses tool:execute, the run fails with owner_invalid. Then the workflow pauses.

Who approves a step before it runs?

In Elaichi, a person approves, and the AI never approves its own work. An approve step asks named people or teams. It also asks the workflow's owner, unless the step leaves the owner out.

Requests reach approvers on the Approvals page, in Slack and by email. The Approvals page has a badge counting requests "waiting on you". A Slack card has Approve and Reject buttons only when no destructive step waits behind it. A request expires after 72 hours by default. The range is five minutes to 30 days. No AI client, no Elaichi Agent and no workflow can answer one.

Make's AI agents page describes manual approvals and stop points for its agents, checked November 2026. Elaichi's approve step is a separate person's sign-off.

Where do the tables and dashboards live?

Elaichi keeps shared tables and live dashboards next to the workflows that fill them. A collection is a shared table, or a set of up to 20 tables, at Automate > Collections.

A collection holds up to 1,000,000 rows and 1 GB. Row history is off until someone turns it on. The switch is Table settings > "Keep a history of row changes". Then any version of a row can be restored.

Dashboards sit at Automate > Dashboards. A widget reads a collection, workflow runs, or a read-only tool from a toolbox. That tool shows live data from a connected app. A collection widget is read for each viewer as that viewer. A public dashboard link shares a snapshot, not live access. Public links stay off until an admin turns them on. Each one expires after 90 days by default.

A knowledge base holds short documents that workflows and the Agent search. A document a workflow writes is saved as untrusted until a person marks it reviewed.

Make data stores are a "built-in database tool to store and manage data" across scenarios. Make Grid is "an automatically generated, real-time visual map" of scenarios, apps, data stores and agents. Make's Analytics Dashboard reports how workflows perform, on Enterprise only, per the pricing page. All checked November 2026.

What does the Elaichi Agent do?

The Elaichi Agent is an AI agent that works inside Elaichi and in Slack. It runs on your organization's own AI provider keys. It opens from the Agent row at the top of the sidebar, or when someone tags @Elaichi in Slack.

The Agent can build, test and publish a workflow, a collection or a dashboard from a chat. Validating and a test run need no approval. Changing a draft or writing rows puts an approval card in front of the person. The card has "Deny", "Allow once" and, when the action can be remembered, "Always allow". A remembered choice or the organization's automatic-writes setting, off by default, skips the card. Publishing or turning on a workflow asks every time and can never be remembered. The Agent cannot answer a workflow approval.

Models come from Settings > AI models. One limit to know: Elaichi does not browse, search or fetch the web for a workflow or the Elaichi Agent.

Make AI Agents run on Make's AI Provider on all plans, at "1 credit per operation" plus credits for AI tokens. Paid plans can use their own provider instead, per Make's credit usage for AI agents, checked November 2026.

How does Elaichi decide who may reach what?

Elaichi governs each role and each person, down to a single tool. RBAC (role-based access control) means a person's permissions come from their role. Elaichi has eight built-in roles over 58 permissions, plus custom roles. On top sit restrictions. These are allow and restrict rules that name whole connectors or single tools. Each one targets a role or one member.

A restriction change takes effect within about two minutes. A member's own rule can only narrow their role's rules, never widen them. No AI surface can create or change a restriction. Only a person decides an access request. Neither an AI client nor the Elaichi Agent can approve or deny one.

People leave, and access follows. In Elaichi, removing or suspending a member revokes every live grant in the same transaction as the membership change. So their AI clients are refused on the next call.

The audit log, a record of who did what, holds one row for each connected-tool call that reaches execution. That covers calls from the Elaichi Agent and from a remote MCP client alike. Each row records the person, the tool, the connection it reached, how it was approved and the outcome. Elaichi keeps 90 days of history, and export to your own Datadog comes with the Black plan.

Make lists six organization roles and six team roles, per its teams page. Its audit logs are "available on the Enterprise plan only" and kept 12 months, checked November 2026.

What do Make and Elaichi cost?

Make charges for steps. The Black plan charges for people plus working time.

Make prices one Make Plan with a credits slider, and every plan has unlimited users. A credit is Make's billing unit, and most module actions cost one credit. With annual billing, the Make Plan starts at $9 a month for 5,000 credits. The rate falls to $0.60 per 1,000 credits at 8,000,000 credits a month. These figures come from the Make pricing page and how features use credits, checked November 2026.

Black lists at $24 per user per month in US dollars, or $18 per user per month billed yearly ($216 a year). Usage beyond a monthly allowance costs extra. The allowance is 1,000 run minutes and 1 GB-month of storage, pooled across the workspace. Past it, a run minute costs $0.02 and a GB-month costs $5. These are US-dollar list prices. /pricing/ shows the price for your country.

Only working time counts. Waiting for an approval, in a wait step or on a timer is free, and test runs cost nothing. AI apps never take a seat, and neither do Guests, Billing Admins or Auditors.

Every new workspace starts on the Black plan itself, as a two-week trial. Every Black feature is on, workflows, collections, dashboards, knowledge bases and the Elaichi Agent included. The one limit: the free monthly usage is a hard stop until you buy Black.

Where does Make really win?

Make wins on building speed, breadth and price per step. A team that needs those should pick Make. These claims come from Make's own pages, such as its pricing page, checked November 2026.

• The visual builder. Scenarios are drawn on a drag-and-drop canvas. Routers, filters and real-time run monitoring come on every plan, Free included (pricing).
• Price per step, with no seat fee. Every plan has unlimited users, and the per-1,000-credit rate falls as volume grows (pricing).
• Data transformation. Make has built-in text, math, date and array functions, and the Make Code app for JavaScript or Python. Custom functions come on Enterprise (functions, pricing).
• App breadth. Make lists 3,500+ apps. Custom apps cover "virtually any application or service that provides a public API" (apps page).
• App triggers and fast schedules. Scenarios start when an app sends data. The Make Plan schedules down to 1 minute (pricing).
• Run tooling. Make has Skip, Retry, Resume, Commit and Rollback handlers, and incomplete executions for retries. It keeps 60 days of saved versions (error handlers, version control).

Make also runs an MCP server that lets Claude or ChatGPT call scenarios as tools. It comes on all plans at no extra cost, checked November 2026. For one builder automating a few apps, Make is the shorter path.

Switch to Elaichi if

Switch when the question is control over what people and agents may do, not the number of apps. Elaichi fits if most of these are true:

• Each workflow must run as a named owner, under that owner's restrictions, and stop when the owner leaves.
• A person, not the AI, must approve a write, a payment or a destructive step, in Elaichi or in Slack.
• Your rules must reach one tool for one role or one member, and change within about two minutes.
• Claude, ChatGPT, Cursor or any MCP client should reach your connectors at https://api.elaichi.ai/mcp, and the Elaichi Agent should share the same catalog.
• You want waiting time free and the price per seat, with AI apps never counted as users.

Every new workspace starts on a 14-day Black trial, no credit card to start. Buying Gold or Black during the trial keeps its remaining days rather than granting a fresh 14, so it is one continuous trial, and the seats are first charged when it ends. A workspace that buys nothing moves to the free Personal plan. Nothing is deleted. The trial is the Black plan itself, with workflows, collections, dashboards, knowledge bases and the Elaichi Agent all on. The free usage of 1,000 run minutes and 1 GB is a hard stop until you buy Black. So a test cannot run up a bill.

For the full plan, read what the Black plan includes. For triggers, steps and runs in depth, read how Elaichi workflows work. For the idea behind one governed address for every AI client, start at the MCP control plane. To check which apps you can reach, browse the connector catalog. Compare plans on the pricing page.