Make
by Make (Celonis)
Visual automation platform that added reasoning AI agents to workflows
Last reviewed 2026-06-18
Make (formerly Integromat) is a no-code visual automation platform where users build multi-step workflows (called scenarios) that connect thousands of apps. The classic product is deterministic: each scenario follows predefined triggers, routers, filters, and actions, and is not an agent. In 2025 Make added AI Agents, a module that reasons about a task, chooses what to do next, and triggers real workflows and tools rather than following a fixed branch. You give an agent plain-English instructions, attach a knowledge base, and give it tools (such as Make scenarios or app actions), and it decides which to call. Make AI Agents are built in the same visual builder, connect to OpenAI-compatible and other LLMs, and support the Model Context Protocol (MCP) for extending tool access. Humans design the agent, its instructions, and its allowed tools, and approve consequential business decisions; within those bounds the agent decides the steps. This places the agent product at the supervised-agent level even though the underlying automation engine is deterministic.
What it can do
Run reasoning AI Agents inside workflows
SupervisedAn AI Agent module reasons about a task in plain English, chooses what to do next, and triggers connected scenarios and tools rather than following a fixed branch; the human designs the agent and its allowed tools.
sourceGive agents tools and a knowledge base
SupervisedAttach a knowledge base and tools (Make scenarios, app actions) to an agent and reuse one agent across multiple workflows; the agent decides which tools to call.
sourceBuild deterministic multi-app scenarios
AssistantVisually compose triggers, routers, filters, and actions across thousands of app integrations; classic scenarios run a fixed, predefined path and are not agentic.
sourceConnect via MCP
SupervisedMake supports the Model Context Protocol to let agents and MCP-compatible clients reach Make tools and extend tool access.
source
Strengths
- +Mature visual builder with thousands of app integrations, now extended with reasoning AI Agents
- +AI Agents are built no-code in the same canvas and can be reused across workflows with attached tools and knowledge
- +Credit-based pricing with a free tier makes entry cheap; MCP support for extending tool access
Limitations
- −Classic scenarios are deterministic, not agents; only the newer AI Agent module reasons and decides
- −Credit-based metering can get hard to predict as agent and scenario usage scales
- −Agents need human design and approval of consequential decisions, so autonomy is bounded by setup
Overview
Make (formerly Integromat) is a no-code visual automation platform that connects thousands of apps. In 2025 it added AI Agents, moving beyond fixed automations into reasoning, tool-using agents. It is owned by Celonis.
What it does
The classic product builds scenarios: deterministic workflows of triggers, routers, filters, and actions. These are not agents. The newer AI Agent module reasons about a task in plain English, chooses what to do next, and triggers connected scenarios and tools. You attach a knowledge base and tools to an agent and can reuse it across workflows. Humans design the agent and approve consequential decisions, so the agent product operates as a supervised agent.
Integrations & setup
Thousands of app integrations, OpenAI-compatible and other LLMs for the agent's reasoning, and Model Context Protocol support to extend tool access. Agents are built in the same visual canvas as scenarios.
Pricing
Freemium and credit-based: a free tier with monthly credits, with Core paid plans reportedly from about $12/mo billed annually, scaling with credit packs. AI Agents are available on paid plans.
Best for / not for
Best for teams already automating across many apps who want to add adaptive, reasoning steps without code. Less suited to those who need a developer-grade agent framework or fully autonomous, unsupervised operation.
Alternatives
Zapier Agents is the closest comparable (agentic layer on a large automation platform); n8n offers self-hostable workflow automation with AI nodes; Gumloop targets AI-first workflow building.
What people are saying
We aggregate real LinkedIn discussion into sentiment for the agents people search most. Make isn't tracked yet, want it added? Request tracking.
FAQ
Are Make scenarios AI agents?+
No. Classic Make scenarios are deterministic automations that follow a fixed path of triggers, routers, filters, and actions. Make's separate AI Agents feature is the agentic part: an agent reasons about a task, decides what to do next, and triggers scenarios and tools.
How autonomous are Make AI Agents?+
Supervised. A human designs the agent, writes its instructions, and chooses its allowed tools; within those bounds the agent decides which steps and tools to use, while humans approve consequential business decisions.
Sources
- Make (official site) · accessed 2026-06-18
- Make AI Agents: the next step in automation (Help Center) · accessed 2026-06-18
- Make pricing (official) · accessed 2026-06-18
Last reviewed 2026-06-18