Zapier
Also known as: Zapier MCP, Zapier Agents, Zaps, Zapier Central, Zapier SDK, Zapier Tables
Workflow automation platform connecting 9,000+ apps, with goal-directed Zapier Agents, an MCP server, and enterprise governance controls.
Zapier is a no-code automation platform connecting thousands of web applications so that work flows between them without manual effort. Its original and still central idea is the Zap, a workflow with a trigger, such as a new form submission, and one or more actions, such as adding a row to a spreadsheet or sending a Slack message. With connectors to roughly nine thousand apps, plus webhooks and custom code steps, Zapier lets non-technical teams wire their tools together quickly, and run logs and automatic retries help keep those workflows reliable.
Zapier has expanded from a workflow connector into a broader automation and AI orchestration platform. Alongside Zaps it now bundles Tables, an automation first database. Interfaces and Forms build simple apps and collect data, Canvas diagrams how automations relate, and Chatbots handle customer facing conversations.
Zapier Agents are goal oriented AI teammates. Instead of following fixed rules, you give an agent an objective in plain English, such as qualifying a new lead and updating the CRM, and it reasons about which of the connected apps and tools to use and takes the multi step actions needed, adapting as it goes. Agents are grounded through knowledge sources, uploaded files or connected apps that resync every 24 hours.
Enterprise controls include AI Guardrails, which scan for personal data, prompt injection and harmful content before outputs reach systems like a CRM or email. A Needs action state pauses an agent at decision points the builder chooses, and reasoning can run through your own models or your own provider account.
Tying it together is Copilot, a natural-language builder that turns a plain-English description into a working Zap or agent, mapping fields and handling errors, and can also generate tables, interfaces, and chatbots. Zapier also offers an MCP surface that lets external AI tools reach its thousands of apps and tens of thousands of actions. Governance covers admin controls, an immutable audit trail and managed credentials.
Vendor details
Canonical URL
https://zapier.com/
Category
Agent builder
Subcategory
App automation platform with goal-directed agents and a governed MCP surface
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Zapier connects roughly 9,000 apps, and Zapier MCP exposes 66,000+ triggers and actions. Agents are granted specific triggers and actions per app rather than inheriting the whole catalog, added through an Insert tools surface, with administrators able to allowlist apps per team and restrict which specific actions are available within them. Zapier MCP exposes the action catalog to external AI clients including Claude, the Anthropic and OpenAI APIs, Python and TypeScript, with managed connections, domain restrictions and Bring Your Own Model routing. A Zapier SDK installs Zapier into AI coding agents, Functions allows workflows to be written as code in a web IDE, and a developer platform supports building and embedding custom integrations. Knowledge sources connect from Airtable, Asana, Box, Confluence, Dropbox, Google Docs, Drive and Sheets, HubSpot, Jira, Microsoft Excel and SharePoint, Notion, Zapier Tables, Zendesk and Zoom. First-party surfaces alongside Zaps are Tables, Forms, Interfaces, Canvas, Chatbots and Agents.
In practice
Every new form submission should create a task and ping the right channel, with no one remembering to do it. Zapier's trigger-action Zaps run that in the background across your apps without any code.
You want a lead qualified and the CRM updated, but the exact steps vary by lead. A Zapier Agent takes that goal in plain English, decides which connected tools to use, and works through the multi-step task itself.
You're nervous about an AI agent leaking personal data into your CRM. Zapier's AI Guardrails scan agent outputs for PII, prompt injection, and harmful content first, blocking or escalating anything risky.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Zapier connects to a catalog of 9,000+ apps, and Zapier MCP exposes 66,000+ triggers and actions. Agents are given specific apps and actions, administrators allowlist apps and restrict individual actions per team, and live data sources such as Google Drive, Box, Notion and Asana are set up separately from action tools. Sourcezapier.com MCP page, Agents guide and security and compliance pageread 2026-09-30 |
|
| Workflow Orchestration | Full |
|
Agent to agent calling lets agents delegate work to each other, agents can be started from a Zap, and Zap workflows supply deterministic steps around agent steps, with version control on agent instructions. Sourcezapier.com Agents guide (updated 13 November 2025)read 2026-09-30 |
|
| Knowledge Grounding & RAG | Full |
|
Knowledge sources are uploaded files of up to 100 MB in CSV, PDF or ODT format, or as PPTX, DOCX or TXT, or connected apps including Airtable, Asana, Box, Confluence, Dropbox, Google Docs, Drive and Sheets, HubSpot, Jira, Excel, SharePoint, Notion, Zapier Tables, Zendesk and Zoom, synced every 24 hours and within an hour of new records, up to 75,000 records, and indexed so agents can query and analyze them. Sourcehelp.zapier.com Add your own data to an agentread 2026-09-30 |
|
| Human Oversight & Guardrails | Full |
|
The agent activity dashboard's Needs action section flags when an agent needs input before continuing, including when the builder has instructed it to pause at a decision point, and administrators allowlist or blocklist apps and restrict specific actions per team. Sourcezapier.com Agents guide and security and compliance pageread 2026-09-30 |
|
| Security, Identity & Governance | Full |
|
Zapier holds SOC 2 Type II and SOC 3 with annual audits and complies with GDPR and CCPA. Enterprise SSO over SAML 2.0 and SCIM, app allowlisting and blocklisting, action level restrictions, managed company connections, domain restrictions, audit trails with log streaming, AES-256 encryption at rest and TLS 1.2 or above in transit are offered, and Enterprise customers are opted out of model training automatically. Sourcezapier.com security and compliance pageread 2026-09-30 |
|
| Observability & Auditability | Full |
|
Every execution is logged and every connection tracked, with log streaming to existing monitoring tools; the agent activity view gives a complete history of the tasks each agent performed, and Zapier MCP records every MCP action in History. Sourcezapier.com security and compliance page, Agents guide and MCP pageread 2026-09-30 |
|
| Memory & State Persistence | Not documented |
|
Zapier's best practices for working with agents cover instructions, testing, monitoring, collaboration and context in depth, and they name knowledge sources as the way agents get the context and information they need, with no mention of memory. Zapier also says agents may behave differently between runs and points users to the Activity log to review how agents interacted with data. Recent agent releases include agent versions, Human in the Loop and shareable MCP server tools, with no memory feature among them. Knowledge sources resync every 24 hours, with retry limits, an inactivity cutoff, a deletion window and a ceiling on records per account, and they supply reference material. No memory, state kept across runs or per user context is described. Sourcehelp.zapier.com Agents best practices, data safety and product updates pagesread 2026-09-01 |
|
| Deployment & Data Residency | Not documented |
|
Zapier runs on Amazon Web Services in the United States with no region choice documented, and the deployment option offered is VPC Peering, which connects Zapier's cloud to data sources in the customer's private network. Sourcezapier.com security and compliance pageread 2026-09-30 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
The agent templates hub lists about a hundred ready made agent templates, among them an ABM research based email personalizer, lead enrichment, automated email reply drafts, sales call analysis, HR policy Q&A, lead background research, a GitHub PR auto reviewer and a weekly project status update generator, and the Agents guide points builders to them as a starting point. Sourcezapier.com agent templates and Agents guideread 2026-09-30 |
|
| Triggers & Channel Coverage | Full |
|
Agents are triggered on demand, on a set schedule, from a Zap or from another app across the catalog, and the Chrome extension brings agents into any browser tab as a further invocation surface. Sourcezapier.com Agents guide (updated 13 November 2025)read 2026-09-30 |
|
| Model Flexibility & Routing | Full |
|
Agents run on standard, advanced or premium model tiers, and Bring Your Own Model connects the customer's own model provider account, paid directly to the provider; AI model policies and BYOM configuration are listed among the governance controls. Sourcezapier.com/pricing/rates and security and compliance pageread 2026-09-30 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
The Workflow API creates Zaps (POST /v2/zaps, OAuth 2.0 with zap:write scope) and lists them, and docs.zapier.com carries a Zapier SDK API reference; Zapier MCP separately lets AI clients call actions across the app catalog. Sourcedocs.zapier.com Workflow API and SDK reference, zapier.com MCP pageread 2026-09-30 |
|
| Testing, Debugging & Optimization | Partial |
|
Agent configuration includes a panel for testing behavior before saving, version control preserves instruction changes, and an activity history shows every task an agent performed for review. No evaluation scoring, test cases or regression runs are documented. Sourcezapier.com Agents guideread 2026-09-30 |
|
| Browser & Computer Use | Not documented |
|
Agents can search the web for research and news, and the Chrome extension brings agents into a browser tab to act in connected apps. Neither has the agent operate a web page or a screen, and no browser automation, desktop or computer use capability is described. Sourcezapier.com Agents guideread 2026-09-30 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Version 0.119.0 of the Zapier SDK adds runWorkflow, a beta method that fires a published workflow's trigger and returns the finished run with its output, or with its error if the run failed. Other workflow run methods can now also wait for a run to end instead of returning at once.
Bears on: Workflow orchestration
View sourceMeta's Muse agent can now use Zapier's connector. Once a user authorizes it, Muse can act across more than 9,000 apps through Zapier MCP, limited to the apps and actions the user approves.
Bears on: MCP / tool calling / API
View sourceZapier launched an open beta of its TypeScript SDK and CLI for Node.js. The release provides programmatic access to over 9,000 app integrations and features built-in OAuth handling, automatic token refresh, paginated results, and type-safe APIs. This allows coding agents to execute raw and authenticated API calls directly through the Zapier infrastructure.
Bears on: Integrations
View sourcePricing
From $19.99 a month for Professional, billed annually. A free plan covers 100 tasks a month.
Tasks. Each successfully completed action step in a Zap uses one task, and plans are priced by monthly task volume on a slider. Agents and Chatbots are billed separately.
Included quota
Free costs $0 for 100 tasks a month, with unlimited two step Zaps, Tables, Forms, Copilot, MCP and about 1,400 free apps (no premium connectors), polling every 15 minutes. Professional, from $19.99 a month billed annually for 750 tasks, adds multi step Zaps, webhooks, filters, paths, AI fields, unlimited premium apps and faster polling for a single user. Team, at $69 a month billed annually for 2,000 tasks and up to 25 users, adds a shared workspace, shared Zaps and folders, shared app connections, SAML SSO and priority support. Enterprise is custom, with unlimited users, advanced admin and app controls, annual task limits that do not expire, observability, SCIM, VPC peering and a designated TAM.
What is public
Zapier bills by tasks, not credits, across 9,000+ app integrations and a product set of Zaps, Tables, Forms, Agents, Chatbots and MCP. Free costs $0 for 100 tasks a month, with two step Zaps only. Professional starts at $19.99 a month billed annually for 750 tasks and scales on a task slider up to about $5,999 a month at 2M tasks. Team is $69 a month billed annually ($103.50 monthly) for 2,000 base tasks and up to 25 users, with SAML SSO. Enterprise is custom and adds SCIM, governance, observability, VPC and a TAM. New accounts get a 14 day Professional trial.
Billing mechanics
Metering is by task. A task is each successfully completed action step that moves data, and triggers, polling, filters, paths, Formatter, Delay, Tables and Forms use no tasks. A Zap uses 1 task per completed action step, with the trigger free, so volume multiplies fast. Five actions on 100 runs a day for 30 days come to 15,000 tasks a month. The plan tier (Free, Pro, Team, Enterprise) sets the features, and a task slider within Pro sets the price, in tiers from 750 to 2M tasks. Tasks are pooled across the account, so Zaps, AI steps, code, MCP and the SDK draw from one pool, and each MCP call costs 2 tasks. Annual billing saves 33% against monthly.
Cost watchouts
Billing per action step means cost grows with both workflow complexity and run frequency. A 4 action lead flow at 30 leads a day uses 3,600 tasks a month and exhausts the 750 task Pro base in about 6 days. Overage runs at 1.25x the per task rate up to a hard stop at about 3x the included volume. Premium connectors (Salesforce, HubSpot, Shopify) need a paid plan. AI Agents and Chatbots are separate add-ons. Monthly billing costs more than annual, which saves 33%.
Variable cost rationale
Zapier meters tasks, one for each successfully completed action step, while triggers, filters, paths, Formatter, Delay, Tables and Forms are free. The bill grows with the number of action steps in each workflow multiplied by how often it runs, so a workflow that runs more often costs more. MCP calls draw 2 tasks each from the same pool, and Agents and Chatbots are billed separately. Each plan's task tier caps the included volume. Past it, Zaps keep running on pay per task billing at about 1.25x the plan's per task rate, unless the account turns that off, and at about 3x the included volume they pause until the next usage period. Moving up the task slider raises the base price, and Enterprise buys annual task pools that do not expire.
Additional watchouts
Billing per task step makes complex or high frequency workflows expensive, so each workflow's task count sets the cost at scale. The Free tier is a sandbox, limited to two step Zaps and 100 tasks, and premium apps and AI add-ons cost extra.
Overage / add-ons
Past the monthly task limit, Zaps keep running on pay per task billing at about 1.25x the plan's per task rate, unless the account turns it off. At about 3x the included volume, Zaps pause until the next usage period. Sustained higher volume moves the account up a task tier on the slider, and Enterprise uses annual task pools that do not expire. Agents and Chatbots are separate paid add-ons.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free plan with 100 tasks a month, and a 14 day Professional trial with no card required
Lowest paid plan
Professional from $19.99 per month billed annually, for 750 tasks a month
Commercial notes
Zapier connects 9,000+ apps through a linear builder and serves 3M+ businesses. Its AI products are Agents, Chatbots, Copilot and MCP.
Key ambiguities
Two Professional subscribers can pay very different amounts depending on task volume, set on a slider from 750 to 2M tasks, so no single price holds. Enterprise pricing is custom and unpublished, and whether SCIM, VPC and a TAM are included or sold as add-ons varies by deal. The Agents and Chatbots add-ons are priced separately from the task plans.
Cancellation / refund
Professional and Team are self serve, billed monthly or annually, with 33% off annual. Annual Team and Enterprise plans can pay by invoice, and Enterprise is an annual custom contract. Nonprofits get 15% off paid plans.
Support SLA / resale
24/7 email support, with weekday live chat from the Professional 2,000 task tier. Team has priority support. Enterprise adds screen sharing support and a designated technical account manager (at a threshold or as an add-on), plus advanced deployment and observability.
Missing data
Each tier's price depends on the task slider, so any single figure is partial. Enterprise is custom and unpublished.
Related vendors
- Lindy — Lindy is an AI teammate that lives in Slack, answers from your tools…
- Activepieces — Open source, MIT licensed automation platform where every one of 700…
- Agenta — Agenta is an open source workspace where teams build AI agents by…
- AgentX — No code platform to build, evaluate, and deploy multi agent…
- AI Library — No code platform for building and running AI agents, with a…
- Aigensei — Enterprise agentic workflow platform that runs business processes…
Alternatives to Zapier
The closest documented capability profiles to Zapier among agent builders tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Domo10.5 / 14Adds documented Deployment & Data Residency
- Airtable11.0 / 14Adds documented Memory & State Persistence and Deployment & Data Residency
- Cassidy12.0 / 14Adds documented Memory & State Persistence and Browser & Computer Use
- Chipp12.0 / 14Adds documented Memory & State Persistence
- Cohere North11.0 / 14Adds documented Deployment & Data Residency
- Dify12.0 / 14Adds documented Memory & State Persistence and Deployment & Data Residency
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded