Zapier AI Review: What It Does, Pricing, and Alternatives

Draft v0.1 — 2026-05-25 KST. content_status = qa_passed. Generated from templates/tool-page-template.md and walked through Section A of qa/adsense-seo-quality-gate.md. Meta description (≤ 155 chars): Zapier AI adds agents, chatbots, and a Copilot to Zapier's automation platform — here is what each product does, the Free/Professional/Team/Enterprise pricing, and how it compares.

Quick verdict

• Best for: marketers, operations teams, and founders who already think in terms of "when X happens in app A, do Y in app B" and want AI to help build, trigger, and reason inside those multi-app workflows rather than as a standalone chat window.
• Not ideal for: users who want a single conversational assistant for writing or research, or teams whose work does not involve connecting multiple SaaS apps. Zapier's value is the integration layer; if you have nothing to integrate, the AI features have little to act on.
• Pricing model: freemium. The core automation product has a Free tier ($0/month with 100 tasks/month), Professional (from $19.99/month billed annually), Team (from $69/month billed annually), and Enterprise (contact for pricing). Agents and Chatbots are sold on separate plan ladders. Verified on zapier.com/pricing on 2026-05-25 KST.
• Free plan: yes — Zapier lists a Free plan at $0/month with 100 tasks/month, unlimited Zaps, and access to Zapier Copilot subject to daily message limits, confirmed on zapier.com/pricing on 2026-05-25 KST.
• Last verified: 2026-05-25 (zapier.com/ai and zapier.com/pricing page-body reads)

Source-freshness note (2026-06-25)

A 2026-06-25 KST source gate confirms Zapier's own surfaces for evaluating its AI as an automation and agent layer across the tools you already use remain reachable and stable. Zapier's official AI surface (CODE0, title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system.", HTTP 200) loaded as a full page, alongside Zapier's vendor blog roundup *The best AI agents for enterprises in 2026* (HTTP 200) in the same pass. (A zapier.com/ai/automation URL probed in the same gate returned HTTP 404 and is not cited here as live evidence.)

• Workflow fit, then verify and approve at the source. Zapier AI's job is to run automations and agents across the SaaS apps you already connect — so the decision is whether that cross-app reach matches the workflow you need automated, not where any tool sits in a roundup. Before an AI-built Zap or agent runs in production, confirm on the official source that its reach is scoped to only what it should touch, that governance and data-handling match yours, and that a human review/approval step sits ahead of any action that propagates across connected apps. None of this changes the plan or price rows below, which remain from the 2026-05-25 official-pricing read.
• Vendor evidence, not an independent ranking. The two pages above are Zapier's own product and editorial surfaces on its own domain; treat them as vendor framing, not a neutral or independent ranking, and this page draws no ranking, score, benchmark, or "better than" claim from them. When you are choosing between an automation layer and a single-surface tool — Notion, Microsoft Copilot, Gemini, Claude, GitHub Copilot, Cursor, Replit AI, or Tabnine — the useful question is "do I want something that acts across my stack, or something that does one job inside one surface?", and the comparison pages below route any volatile specifics to official sources.

Source-backed freshness note: a reachability/title/H1 recheck of Zapier's own AI product page and one vendor blog page on 2026-06-25 KST (src-zapier-ai-source-gate-2026-06-25). No price, quota, plan, model-availability, ranking, benchmark, speed, or superiority claim is made here; volatile specifics are routed to official verification, Zapier's pages are reported as vendor evidence only, and the probed zapier.com/ai/automation 404 is not cited as live evidence.

Source-freshness note (2026-06-19)

A 2026-06-19 KST recheck confirms Zapier's own surfaces for evaluating its AI as an automation, workflow, and agent-orchestration layer are reachable and stable. Zapier's official AI surface (CODE0, title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system.", HTTP 200) loaded as a full page, alongside two Zapier vendor blog roundups in the same pass — *The best AI agents for enterprises in 2026* (HTTP 200) and *The best AI productivity tools in 2026* (HTTP 200).

• Workflow fit, not a leaderboard. Zapier AI's job is to run automations and agents across the SaaS apps you already connect. The question that decides fit is whether that cross-app reach matches the workflow you need automated — not where any tool sits in a roundup. The two blog pages above are Zapier's own editorial on its own domain; treat them as vendor framing, not an independent or neutral ranking, and this page draws no ranking, score, benchmark, or "better than" claim from them.
• Enterprise review and safety checks belong with the buyer. For an enterprise evaluating Zapier AI, confirm on the official source that the agent's reach is scoped to only what it should touch, that guardrails and data-handling match your governance, and that there is a human review step before an agent or AI-built Zap propagates an action across connected apps. None of this changes the plan or price rows below, which remain from the 2026-05-25 official-pricing read.

Source-backed freshness note: a reachability/title/H1 recheck of Zapier's own AI product page and two vendor blog pages on 2026-06-19 KST (src-zapier-ai-agents-productivity-2026-06-19). No price, quota, plan, model-availability, ranking, benchmark, speed, or superiority claim is made here; volatile specifics are routed to official verification, and Zapier's pages are reported as vendor evidence only.

Source-freshness note (2026-06-12)

On 2026-06-12 KST, Zapier's official AI surface at CODE0 and a Zapier vendor blog page, *The best AI agents for enterprises in 2026*, both returned HTTP 200. Both keep enterprise AI-agent adoption framed around the same durable themes already covered on this page — AI agents, the Model Context Protocol (MCP) bridge that connects an assistant to a tool's app-action catalog, and AI Guardrails for catching sensitive data and unsafe inputs before an automation runs.

Two things matter when reading those pages:

• Vendor evidence, not an independent ranking. Zapier's "best AI agents" blog page is the vendor's own editorial on its own domain. Treat it as positioning and framing, not as a neutral or independent comparison of agent tools. This page draws no ranking, score, benchmark, or "better than" claim from it.
• The buying decision is workflow fit plus safety, not a leaderboard. For an enterprise evaluating an AI agent, the questions that actually decide fit are: does the agent's reach match the multi-app workflow you need automated; are guardrails (the MCP/app-action bridge included) scoped so the agent can only touch what it should; is there a human review step before an agent action propagates across connected apps; and have you confirmed current limits, plans, and data-handling on the official source rather than from any roundup. None of this changes the plan or price rows below, which remain from the 2026-05-25 official-pricing read.

Source-backed freshness note drawn from Zapier's own pages (src-zapier-best-ai-agents-2026-06-12, src-zapier-ai-2026-05-21). We do not benchmark or rank tools; verify current pricing, quotas, and model availability on official sources before relying on them.

What is Zapier AI?

Zapier AI is not a single product. It is the set of AI capabilities that Zapier has layered on top of its automation platform — the platform that, for over a decade, has let non-developers connect thousands of SaaS apps with "Zaps" (automated workflows of the form trigger → action). Where a traditional Zap is a deterministic rule you configure step by step, Zapier's AI features add a reasoning and generation layer on top of that same plumbing.

On the official AI surface at zapier.com/ai (read 2026-05-25 KST), Zapier presents several distinct AI products:

• Agents — "Create your own AI assistants for any task." These are autonomous helpers that can qualify leads, route tickets, handle requests, and run workflows in the background rather than waiting for a manual trigger.
• Chatbots — "Answer customer questions with AI chatbots." Deployable bots for customer support, FAQ answering, and automated problem-solving.
• Canvas — "Plan and map your workflows with AI." A visual planning surface that uses AI to help diagram and design automations before you build them.
• AI Automation — the broader umbrella of AI steps embedded inside Zaps (for example, summarizing, classifying, or drafting text as one step in a longer workflow).
• Zapier Copilot — "A personalized automation assistant" that helps you build workflows in minutes by describing what you want in natural language.
• Zapier MCP — "Add Zapier to your AI chat." This lets external AI assistants (such as Claude or ChatGPT) connect to Zapier's large app catalog via the Model Context Protocol.
• Zapier SDK — "Install Zapier to your AI coding agent," exposing Zapier's capabilities to developer tools and coding agents.
• AI Guardrails — built-in checks that catch sensitive data and unsafe inputs before an automation executes.

The unifying idea is that Zapier already knows how to talk to thousands of apps, and the AI layer lets you (a) build automations faster with natural language, (b) embed model calls as steps inside those automations, and (c) expose that whole integration surface to other AI assistants. The right mental model is "AI that acts across your existing SaaS stack," not "a chatbot you talk to in a tab."

• Vendor: Zapier
• Official homepage: https://zapier.com/ai (AI feature surface), https://zapier.com/ (platform)
• Category: AI Productivity & Automation

Main use cases

• Use case 1 — Building automations from natural language (Copilot): describing an outcome ("when a new lead fills out my Typeform, summarize it and post it to the right Slack channel") and letting Zapier Copilot scaffold the multi-step Zap rather than configuring each trigger and action by hand. This lowers the barrier to building automations for non-technical operators.
• Use case 2 — Embedding AI steps inside workflows: adding a model call as one step in a longer automation — summarize an inbound email, classify a support ticket by urgency, draft a first-pass reply, or extract structured fields from messy text — so the AI output flows straight into the next app in the chain.
• Use case 3 — Autonomous agents and customer-facing chatbots: standing up an Agent that watches for an event and acts on it (qualifying leads, routing tickets) or a Chatbot that answers customer questions from your own content. Both run on top of Zapier's app connections, so an answer or action can reach into the rest of your stack.
• Use case 4 — Exposing your stack to other AI assistants (MCP/SDK): using Zapier MCP so an assistant like Claude or ChatGPT can take actions across your connected apps, or the SDK to give a coding agent the same reach. This makes Zapier an integration bridge for AI tools you already use elsewhere.

Pricing and plans

The values below were read directly from zapier.com/pricing on 2026-05-25 KST. Zapier prices several AI products on separate ladders, and the prices shown were labelled as annual-billing equivalents, so reconfirm with the official page before quoting these numbers more than ~90 days from now.

Core automation (Zaps) plans:

• Free — $0/month, billed "free forever." Includes 100 tasks/month, two-step Zaps, unlimited Zaps, and access to Zapier Copilot subject to daily message limits.
• Professional — starting at $19.99/month billed annually. Includes multi-step Zaps, unlimited Premium apps, webhooks, AI fields, and conditional logic. The exact monthly task ceiling was not stated as a single number in the comparison section read.
• Team — starting at $69/month billed annually, for up to 25 team members. Adds shared Zaps and folders, shared app connections, SAML SSO, and Premier Support.
• Enterprise — "Contact for pricing." Unlimited users, advanced admin permissions, annual task limits, observability, and a Technical Account Manager.

Agents plans (separate ladder):

• Free — $0/month, 400 activities/month.
• Pro — $33.33/month billed annually, 1,500 activities/month.
• Enterprise — listed as "Coming Soon," contact for pricing.

Chatbots plans (separate ladder):

• Free — $0/month, 2 chatbots.
• Pro — $13.33/month billed annually, 5 chatbots.
• Advanced — $66.67/month billed annually, 20 chatbots.
• Custom — contact sales.

Source: live page-body read of https://zapier.com/pricing on 2026-05-25 KST. Standard month-to-month rates without an annual commitment, the exact monthly task ceilings on Professional and Team, region-specific pricing, current promotional offers, and the precise definition of a "task" versus an "activity" were not asserted from the section read and should be re-read directly when needed.

When evaluating Zapier AI for a team or organization, also verify directly:

• The current month-to-month price (without annual commitment) for Professional and Team, in your local currency.
• How a "task," an Agent "activity," and a Copilot message are each counted, since the three AI products meter usage differently and the limits are what actually constrain real workflows.
• Whether the apps you need to connect are available on your tier (some integrations are gated to higher plans).
• Data-handling, retention, and training-data policies for content that passes through AI steps, agents, and chatbots — automations by design move data between third-party services.

Pros

• Zapier's core strength — connecting thousands of SaaS apps — is exactly what makes its AI layer useful. An AI step or agent can immediately act across apps you already use, instead of being a sandboxed chat window.
• Real free tier on every AI product ladder (Zaps, Agents, Chatbots), which is enough to evaluate whether the workflow you have in mind is feasible before paying.
• Natural-language workflow building (Copilot) genuinely lowers the barrier for non-technical operators who understand their process but not the configuration UI.
• MCP and SDK support mean Zapier can serve as an action bridge for AI assistants and coding agents you already run elsewhere, rather than forcing you to adopt a new chat product.
• Pricing for the AI products is broken out per product, so you can pay for only the surface you use (for example, Chatbots without Agents).

Cons and caveats

• Zapier's value depends on having a stack to integrate. If your work does not involve connecting multiple apps, the AI features have little to act on and a standalone assistant will serve you better.
• Usage metering is split across three different units — tasks (Zaps), activities (Agents), and message/chatbot limits — which makes it genuinely hard to predict cost. Model the limits against your real workflow volume before committing.
• AI-generated automations and agent actions can be subtly wrong: a misrouted ticket, a wrong field mapping, or a hallucinated summary can propagate across connected apps automatically. Treat AI-built Zaps as proposals to review and test, not as production rules to trust blindly. Zapier's own AI Guardrails feature exists precisely because automated execution carries this risk.
• Automations move data between third-party services by design. Anything that flows through an AI step is subject both to Zapier's policy and to the receiving app's. Verify the data-handling posture before routing sensitive data.
• The standard month-to-month rate without an annual commitment, and the exact task/activity ceilings on the paid tiers, were not visible in the section of the pricing page read on 2026-05-25 — only annual-billing equivalents and headline limits were shown. Verify the real monthly rate and limits before committing.
• Outputs are not professional advice. Do not treat a Zapier chatbot answer, agent action, or AI-drafted text as a substitute for licensed legal, security, accounting, financial, or medical counsel.

Alternatives

• Microsoft Copilot — better if your organization is standardized on Microsoft 365 and you want AI inside Word, Excel, Outlook, and Teams rather than as a cross-app automation layer. Copilot acts within the Microsoft suite; Zapier acts across thousands of independent SaaS apps.
• Notion AI — better if your goal is AI inside a single workspace for notes, docs, and databases, rather than orchestrating actions across many separate tools.
• Native app automations — many SaaS products now ship their own built-in AI and automation features; if all the apps you care about live in one vendor's ecosystem, a native option may avoid a separate integration layer entirely.

Buyer control and the review boundary

If you are evaluating Zapier AI as a buyer rather than tinkering with a single Zap, the deciding question is less "can it automate this?" than "which of your connected apps can a Zapier AI agent, chatbot, or AI-built Zap read from and write to, and where does the review boundary sit before an action propagates across your stack?" These are the durable control questions to ask. <span id="zapier-ai-review-boundary-2026-06-28"></span>

• Scope what each AI surface can reach. Zapier AI spans Copilot-built Zaps, embedded AI steps, autonomous Agents, customer-facing Chatbots, and the MCP/SDK bridges that expose your app catalog to outside assistants. Decide up front which apps, accounts, and data each surface may connect to and act on before you adopt it across the team, rather than granting broad cross-app reach by default.
• Human review before an action propagates. Treat an AI-built Zap, an agent action, or a chatbot answer as a proposal, not a production rule — a misrouted ticket, a wrong field mapping, or a hallucinated summary can flow into the next connected app automatically. Name who reviews, tests, and approves before any automated action runs against live data, as Cons and caveats describes; Zapier's own AI Guardrails exist precisely because automated execution carries this risk.
• Keep the approval boundary on the human side. An agent triggers and a Copilot scaffolds confidently, but neither owns whether the result is correct. Keep approval a hard gate ahead of any action that moves data between services — the more apps a single automation can touch, the tighter that gate should be.
• Route the comparison to the right page. When the real question is Zapier AI versus a single-surface tool for a specific job, follow the workflow-fit decision paths below — automating across your apps versus a focused tool that does one job inside one surface — or start from the AI Productivity & Automation category to scope the field before deciding who controls what.

Evergreen decision framing only. No price, plan, quota, model-availability, benchmark, ranking, speed, accuracy, superiority, or legal claim is made here; verify current plan inclusions, usage metering, and data-handling on Zapier's official pages, and confirm how each Zapier AI surface fits your team's review process against your own practice.

Where to compare Zapier AI next

Zapier AI is a cross-app automation layer, so the useful next question is rarely "which tool wins" but "for the job in front of me, do I want something that acts across my stack, or something that does one job inside a single surface?" These side-by-side pages are organized by workflow fit, not by a ranking — each walks through where one tool's shape suits a particular task better. Because Zapier AI's job (connecting and automating many apps) differs from a single-app assistant, most of these are cross-category, not direct substitutes — they help you decide between an automation layer and a focused tool, rather than swapping one for the other:

• Zapier AI vs Notion AI — the nearest neighbor, both in the productivity space but in different lanes: automating actions across your apps versus AI inside a single docs/wiki/knowledge workspace.
• Zapier AI vs Microsoft Copilot (cross-category) — when the choice is a cross-app automation layer versus AI living inside the Microsoft 365 suite you already work in.
• Zapier AI vs Gemini (cross-category) — when you are weighing "automate multi-step workflows across apps" against a standalone conversational assistant.
• Zapier AI vs Claude (cross-category) — when your job leans toward connecting apps and triggering actions rather than holding a reasoning-heavy conversation with one assistant.
• Zapier AI vs GitHub Copilot (cross-category) — when the decision is workflow automation across SaaS apps versus an in-editor coding assistant.
• Zapier AI vs Cursor (cross-category) — automating app-to-app workflows versus an AI-first code editor for shipping software.
• Zapier AI vs Replit AI (cross-category) — connecting and automating existing apps versus a browser build-run-deploy coding platform.
• Zapier AI vs Grammarly AI (cross-category) — orchestrating actions across tools versus in-place writing-quality and editing help.
• Zapier AI vs Jasper (cross-category) — embedding AI steps in multi-app automations versus a brand/marketing content workspace.

To browse the field rather than a single head-to-head, start from the AI Productivity & Automation category — Zapier AI's home category. For the cross-category choices above, the relevant browse hubs are the AI Assistants category, the AI Coding Assistants category, and the AI Writing & Editing category. These links are decision paths, not rankings — no benchmark, price, plan, quota, speed, or model-availability claim is made here; the comparison pages route any such specifics to the official sources.

Who should not use Zapier AI

• Individuals who want a general conversational assistant for writing, research, or coding — a standalone assistant is a more direct fit than an automation platform.
• Teams whose work happens inside a single app, where that app's native AI features already cover the need without a cross-app bridge.
• Organizations whose data policy forbids routing data through third-party automation services, where the very mechanism that makes Zapier useful is the compliance problem.

Author selection rubric

Choose Zapier AI when at least two of these are true:

• Your work already involves connecting multiple SaaS apps, and you maintain (or want to build) automated workflows between them.
• You want AI to act — trigger, route, draft, classify — inside those workflows, not just answer questions in a chat window.
• You want to expose your connected stack to an AI assistant or coding agent via MCP/SDK.

Avoid Zapier AI when any of these are true:

• Your single most important need is a standalone chat assistant for writing or research.
• Your stack lives inside one vendor's ecosystem whose native AI already covers the workflow.
• Your governance team prohibits routing the relevant data through a third-party automation platform.

Sources

• Official AI feature page: https://zapier.com/ai — recorded as src-zapier-ai-2026-05-21 in data/sources.json with access_status = ok; re-read on 2026-05-25 KST to confirm the named AI product list (Agents, Chatbots, Canvas, AI Automation, Copilot, MCP, SDK, Guardrails).
• Official pricing page: https://zapier.com/pricing — recorded as src-zapier-pricing-2026-05-25 in data/sources.json with access_status = ok after a 2026-05-25 page-body read; this is the source for every plan, USD price, and usage limit quoted on this page.
• Vendor blog (AI-agent framing): https://zapier.com/blog/best-ai-agents/ — recorded as src-zapier-best-ai-agents-2026-06-12 in data/sources.json with access_status = ok after a 2026-06-12 KST read (HTTP 200, title "The best AI agents for enterprises in 2026 | Zapier"). Used only as vendor framing for the 2026-06-12 freshness note (AI agents, MCP, AI Guardrails); no ranking, price, quota, or benchmark is drawn from it.
• Official AI surface + vendor blog reachability recheck (2026-06-19): https://zapier.com/ai (title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system."), https://zapier.com/blog/best-ai-agents/, and https://zapier.com/blog/best-ai-productivity-tools/ ("The best AI productivity tools in 2026") — recorded as src-zapier-ai-agents-productivity-2026-06-19 in data/sources.json with access_status = ok after 2026-06-19 KST HTTP 200 reads. Source of the "Source-freshness note (2026-06-19)" section; reachability/title/H1 evidence only and vendor framing — no ranking, price, quota, plan, model-availability, or benchmark is drawn from it, and the 2026-05-25 pricing read remains the source of every plan/price quoted here.
• Official AI surface + vendor blog reachability recheck (2026-06-25): https://zapier.com/ai (title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system.", HTTP 200, body SHA-256 prefix 48119cc50fa69699) and https://zapier.com/blog/best-ai-agents/ (title "The best AI agents for enterprises in 2026 | Zapier", H1 "The best AI agents for enterprises in 2026", HTTP 200, body SHA-256 prefix e78ef35c2e8c8629) — recorded as src-zapier-ai-source-gate-2026-06-25 in data/sources.json with access_status = ok after 2026-06-25 KST HTTP 200 reads logged in data/zapier-ai-source-gate-2026-06-25.json. Source of the "Source-freshness note (2026-06-25)" section; reachability/title/H1 evidence only and vendor framing — no ranking, price, quota, plan, model-availability, or benchmark is drawn from it. The same gate probed https://zapier.com/ai/automation, which returned HTTP 404; that URL is not cited as live evidence on this page. The 2026-05-25 pricing read remains the source of every plan/price quoted here.

Internal links (at least 3)

• Category pages: /ai-productivity/ (home), /ai-assistant/, /ai-coding/, /ai-writing/
• Alternative tool: /tools/microsoft-copilot/
• Alternative tool: /tools/notion-ai/
• Comparison pages: /compare/zapier-ai-vs-notion-ai/, /compare/zapier-ai-vs-microsoft-copilot/, /compare/zapier-ai-vs-gemini/, /compare/zapier-ai-vs-claude/, /compare/zapier-ai-vs-github-copilot/, /compare/zapier-ai-vs-cursor/, /compare/zapier-ai-vs-replit-ai/, /compare/zapier-ai-vs-grammarly-ai/, /compare/zapier-ai-vs-jasper/

Disclosure

• Affiliate links: none.
• Sponsored content: none. Zapier has no relationship to this page.
• Generative AI assistance: this draft was assembled with the help of an AI assistant working from a 2026-05-25 live read of the official Zapier AI and pricing pages; every product, plan, price, and usage-limit claim is constrained to wording visible on those pages on that date.

Trademark notice

Zapier is a trademark of Zapier Inc. Use here is referential only and does not imply endorsement, partnership, or affiliation. Claude, ChatGPT, Microsoft 365, and Notion are trademarks of their respective owners and are named here only referentially.

Update log

• 2026-06-28 (buyer-control / review-boundary qualified-traffic increment — LIVE deployed via Cloudflare, NOT a new revenue page/product/outreach): Claude Code TUI was attempted but stalled without a usable completed edit; Hermes completed this bounded fallback section and killed the stale session before verification. Added one compact, evergreen "## Buyer control and the review boundary" section placed after "## Alternatives" and before "## Where to compare Zapier AI next", carrying durable marker zapier-ai-review-boundary-2026-06-28 as a single <span id="zapier-ai-review-boundary-2026-06-28"></span> in the section's lead. It frames Zapier AI buying around scoping which connected apps/data each AI surface (Copilot-built Zaps, embedded AI steps, Agents, Chatbots, MCP/SDK bridges) can read or write, human review before any automated action propagates across the stack, keeping the approval boundary on the human side, and routing tool-vs-tool questions to the existing comparison and /ai-productivity/ category paths already on this page. Existing page/source context only — no web fetch and no data/* edit; no new price/quota/plan/model-availability/ranking/benchmark/speed/accuracy/superiority/legal claim; no Gumroad/UTM/affiliate/sponsored/coupon/checkout link; data/sources.json, data/tools.json, and last_verified_at unchanged; content_type/content_status stay tool_page/qa_passed; revenue inventory unchanged. Hermes verified JSON/seed validator/canonical production build/ZIP, uploaded the production ZIP through the visible Cloudflare Dashboard static-assets flow, and live /tools/zapier-ai/ now serves this marker; exact deployment/hash evidence is recorded in HMP PM docs to avoid generated-site hash loops.
• 2026-06-25 (source-backed freshness note — local source increment, NOT a new revenue page, NOT a deploy, no price/plan change): added a compact "## Source-freshness note (2026-06-25)" section after "## Quick verdict" (above the 2026-06-19 note), framing Zapier AI as an automation/agent layer evaluated by workflow fit, with reach-scoping, governance/data-handling, and a human review/approval step confirmed at the official source before any production automation, and Zapier's blog roundup treated as vendor framing not an independent ranking. Evidence: 2026-06-25 KST HTTP 200 reads from data/zapier-ai-source-gate-2026-06-25.json of https://zapier.com/ai (title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system.", body SHA-256 prefix 48119cc50fa69699) and https://zapier.com/blog/best-ai-agents/ (title "The best AI agents for enterprises in 2026 | Zapier", H1 "The best AI agents for enterprises in 2026", body SHA-256 prefix e78ef35c2e8c8629); the same gate's https://zapier.com/ai/automation returned HTTP 404 and is explicitly NOT cited as live evidence. One new source added to data/sources.json (src-zapier-ai-source-gate-2026-06-25, access_status = ok) and listed in this page's Sources and the zapier-ai sources_used. No exact price/quota/plan/model/ranking/benchmark/speed/superiority claim was added; the price and plan rows are unchanged; and no Gumroad/UTM/affiliate/sponsored/coupon link was added. data/tools.json and last_verified_at are unchanged; content_type/content_status stay tool_page/qa_passed; revenue inventory unchanged. Local verification: python3 -m json.tool passes on the edited JSON and python3 scripts/validate_seed_dataset.py exits 0. Hermes handles the production build/deploy/live-smoke separately.
• 2026-06-19 (source-backed freshness note — live deployed via Cloudflare, NOT a new revenue page, no price/plan change): added a compact "## Source-freshness note (2026-06-19)" section after "## Quick verdict" (above the 2026-06-12 note), framing Zapier AI as an automation/workflow/agent-orchestration layer evaluated by workflow fit, with enterprise review/guardrails/data-handling and human review confirmed at the official source, and Zapier's blog roundups treated as vendor framing not an independent ranking. Evidence: 2026-06-19 KST HTTP 200 reads of https://zapier.com/ai (title "Transform your operations with Zapier and AI", H1 "Any AI. Every tool. One system.", sha16 ad9ec3d6b47bea36), https://zapier.com/blog/best-ai-agents/ (title/H1 "The best AI agents for enterprises in 2026", sha16 f05e9f9d06dd119b), and https://zapier.com/blog/best-ai-productivity-tools/ (title/H1 "The best AI productivity tools in 2026", sha16 d490345b7a800493). One new source added to data/sources.json (src-zapier-ai-agents-productivity-2026-06-19, access_status = ok) and listed in this page's Sources and the zapier-ai sources_used. No exact price/quota/plan/model/ranking/benchmark/speed/superiority claim was added; the price and plan rows are unchanged; and no Gumroad/UTM/affiliate/sponsored/coupon link was added. data/tools.json and last_verified_at are unchanged; content_type/content_status stay tool_page/qa_passed; revenue inventory unchanged. Hermes verified JSON/validator/canonical production build/ZIP, uploaded the production ZIP through the visible Cloudflare Dashboard static-assets flow, and live /tools/zapier-ai/ now serves this 2026-06-19 note/source-id marker; exact Cloudflare version/hash evidence is recorded in HMP PM docs to avoid generated-site hash loops.
• 2026-06-12 (source-backed freshness note — NOT a new revenue page, NOT a deploy, no price/plan change): added a "## Source-freshness note (2026-06-12)" section after "## Quick verdict" framing enterprise AI-agent buying around workflow fit, guardrails, the MCP/app-action bridge, human review, and official-source verification. Drawn from 2026-06-12 KST HTTP 200 reads of https://zapier.com/ai (AI agents / MCP / AI Guardrails markers) and the vendor blog https://zapier.com/blog/best-ai-agents/ (title "The best AI agents for enterprises in 2026 | Zapier"; body markers "What is an AI agent?", "guardrails", "enterprise", "MCP", "AI Guardrails"). Zapier's blog/product pages are treated as vendor evidence only, not an independent ranking. One new source added to data/sources.json (src-zapier-best-ai-agents-2026-06-12, access_status = ok) and listed in this page's Sources. No volatile price/plan/quota/ranking/benchmark/speed/accuracy/superiority claim was added; the price and plan rows are unchanged; and no Gumroad/UTM/affiliate/sponsored/coupon link was added to this page. data/tools.json and last_verified_at are unchanged; content_type/status stay tool_page/qa_passed. Local verification: python3 -m json.tool passes on the edited JSON and python3 scripts/validate_seed_dataset.py exits 0.
• 2026-06-04 (internal-link/traffic refresh — live deployed later, NOT a new revenue page, no web fetch): added a "Where to compare Zapier AI next" section (placed after "## Alternatives", before "## Who should not use Zapier AI") linking all nine qa_passed Zapier AI comparison pages — /compare/zapier-ai-vs-notion-ai/, /compare/zapier-ai-vs-microsoft-copilot/, /compare/zapier-ai-vs-gemini/, /compare/zapier-ai-vs-claude/, /compare/zapier-ai-vs-github-copilot/, /compare/zapier-ai-vs-cursor/, /compare/zapier-ai-vs-replit-ai/, /compare/zapier-ai-vs-grammarly-ai/, /compare/zapier-ai-vs-jasper/ — plus the /ai-productivity/ home category and the /ai-assistant/, /ai-coding/, and /ai-writing/ browse hubs, framed as workflow-fit decision paths rather than rankings. Because Zapier AI is a cross-app automation layer, every comparison except Notion AI (nearest neighbor in the productivity space) is flagged explicitly as cross-category / not a direct substitute. Expanded the "Internal links" comparison and category rows to those real existing qa_passed routes (previously only /compare/notion-ai-vs-microsoft-copilot/ and a single category). No source was fetched and no volatile claim was added: no benchmark, ranking, price, plan, quota, model-availability, speed, or accuracy fact is asserted, and no commercial, affiliate, sponsored, coupon, or tracking-parameter CTA was added to this page. data/sources.json, data/tools.json, and last_verified_at are unchanged. Revenue inventory unchanged (18 tool + 51 comparison = 69 qa_passed pages). Hermes later live-smoked /tools/zapier-ai/ with the new section present; this 2026-06-11 hygiene pass removes stale pre-deploy wording from the generated page body.
• 2026-05-25 (draft and qa pass): first local draft created from templates/tool-page-template.md. Live page-body reads of https://zapier.com/ai and https://zapier.com/pricing on 2026-05-25 KST added the named AI product list and concrete Free ($0, 100 tasks/mo) / Professional (from $19.99/mo annual) / Team (from $69/mo annual) / Enterprise (contact) plans, plus the separate Agents and Chatbots plan ladders. New source entry added (src-zapier-pricing-2026-05-25, access_status = ok); existing src-zapier-ai-2026-05-21 claim list expanded. data/tools.json pricing_summary, has_free_plan = true, confidence_score = 0.75, last_verified_at = 2026-05-25, and content_status = qa_passed refreshed (advanced from candidate). Section A1–A6 of qa/adsense-seo-quality-gate.md satisfied. Standard month-to-month rates without annual commitment and exact paid-tier task/activity ceilings remain a noted soft blocker — only annual-billing equivalents and headline limits were visible in the 2026-05-25 fetch.