Are Zapier and Make AI agents?
Zapier is migrating its standalone Agents product into an AI step inside the Zap editor; at the default Premium tier that step uses 5 tasks per model call and per tool call. Make's agent app is in open beta and bills a credit per operation plus tokens. Read 2026-11-06.
Contents
Short answer, from the vendors' own help pages: both sell something they call an agent, and both put it inside their existing automation builder as one step or app among fixed steps. Zapier's standalone Agents product is being folded into a step called AI by Zapier. Make's agent app has been in open beta since February 2026. Neither page reports how often its agent finishes a job correctly. The prices per plan are on the earlier page, No-code AI agent builders; this one covers what an agent run adds to the bill. Builders that can be used free, with each one's licence and catch, are on Free AI agent builders: licence, free tier, catch.
Zapier is moving its standalone Agents product into the Zap editor
Zapier's Agents page promises teammates that work "across 9,000+ apps" and shows customer quotes, which the vendor chose.[1] Its help centre now says: "Zapier is migrating Agents (agents.zapier.com) to AI by Zapier, a more powerful AI experience built directly into the Zap editor."[2] The reason it gives is that Agents "ran as a standalone product", so you "could not combine your agents with the full library of Zap triggers, actions, filters, branching logic, and automation history."[2] The article was read on 2026-11-06 and is marked updated a month earlier; it says "Zapier has not set a date to turn off Agents yet."[2]
The replacement mixes three kinds of step in one Zap: "agentic steps that reason and act autonomously, AI steps for tasks like summarization and classification, and deterministic steps for precise, predictable actions."[2] So in Zapier's own terms a Zap is an agent only where it contains an agentic step; the rest is fixed. Sources such as HubSpot, Asana, Zendesk, Zoom and Jira "are not yet supported in AI by Zapier", with support "planned for Q4 2026".[2]
What a run costs
Zapier bills AI by Zapier in tasks, scaled by model tier: Standard is 1x with no tools, Advanced 3x, Premium 5x, and your own AI account 1x. A new step on a paid plan defaults to Premium.[3] The formula on the page is "Tasks used per run = (1 × model rate) + (number of tool calls × model rate)".[3] A tool call is "any time AI by Zapier successfully uses a tool (an app action or knowledge source)".[3] By that formula, a Premium run that calls four tools uses (1 × 5) + (4 × 5) = 25 tasks. That arithmetic is mine, not a figure from Zapier. A run is limited to 75 tasks by default and can be raised to 500; past the limit it "automatically pauses for your review before continuing."[2] At the Premium rate, 14 tool calls reach 75. On a Free plan, "you cannot add tools or publish a Zap with an AI by Zapier step."[3]
In Make the agent is an app inside a scenario
Make's help pages say: "AI agent refers to Make AI Agent (New), an app that you use to create agentic scenarios."[4] The app was released on February 2, 2026, and "is available in open beta, so product functionality and pricing may change."[4] Make says when it is the right tool: "Use a scenario with an agent for tasks that require flexible reasoning, judgment calls, and variable inputs and outputs", and a scenario with an ordinary AI app "for tasks that combine predefined logic with AI-generated content".[4]
What a run costs
Make counts credits. Running an agent on Make's own AI provider costs "1 credit per operation + credits based on AI tokens"; with your own provider connection it is "1 credit per operation".[5] A chat costs "1 credit per operation + 1 credit per operation from called tools + credits based on AI tokens".[5] The page does not say how many credits a token amounts to, so a run's cost on Make's provider can't be worked out from it. Make's advice includes: "Sub-agents use extra credits to add their own reasoning."[6]
What both pages say about control
Make's best-practice page lists "Prompting a human to validate outputs" among the guardrails to write into an agent's instructions, then adds: "agents could still behave unpredictably and ignore or misinterpret your explicit guardrails. Choose your data with this assumption in mind."[6] Zapier's control is the task limit above. Neither page describes a check that runs outside the model. For what the security literature says about that gap, see AI agent security and guardrails.
Not yet proven
- How often either agent completes a job correctly. Zapier's page has customer quotes chosen by the vendor. Make's pages have no results. No independent test of either is among the sources.
- What a real workflow costs. The task formula is documented, but no source gives tool calls per run for a real job. Make's pages give no credits-per-token rate.
- When Zapier turns Agents off. The help page says no date is set, and it was last updated a month before reading.
- Make's pricing after beta. The page says pricing "may change".
- A side-by-side run. No source runs the same job in both tools.
If you run a business on Zapier or Make and have compared an agent step's bill with the plain version of the same workflow, I would like to compare notes: mackenzie@mackenziebowes.com.
Sources
- Zapier, Agents product page. A vendor's marketing page. Read in Rotunda 2026-11-06. zapier.
com/ agents - Zapier Help Center, Migrating from Agents to AI by Zapier. Read in Rotunda 2026-11-06. help.
zapier. com/ hc/ en- us/ articles/ 47402591569805- Migrating- from- Agents- to- AI- by- Zapier - Zapier Help Center, AI by Zapier model tier pricing. Read in Rotunda 2026-11-06. help.
zapier. com/ hc/ en- us/ articles/ 46425475442829- AI- by- Zapier- model- tier- pricing - Make Help Center, Introduction to Make AI Agent (New). Read in Rotunda 2026-11-06. help.
make. com/ introduction- to- make- ai- agent- new - Make Help Center, Credit usage for AI agents. Read in Rotunda 2026-11-06. help.
make. com/ credit- usage- for- ai- agents - Make Help Center, Make AI Agent (New) best practices. Read in Rotunda 2026-11-06. help.
make. com/ make- ai- agent- new- best- practices