Skip to content
unnudged
brandSurging

Zapier on X

Talk about Zapier splits between people describing Zaps that quietly run their leads, inboxes and CRM syncs, and 13 people who think a prompt or an agent now does the same job. Its defenders' case is simpler: quick app-to-app connections, where n8n and Make are the alternatives named most.

First written 27 Oct · what changed · from 95 posts by 85 people

Tracking posts that say “zapier”

Track this for your market
Posts, last 24h
41
—vs. previous 24h
Posts, last 7 days
147
82 in the last 72h
Velocity
21/day
7-day average
Momentum
+95%
Last 24h vs. a normal day this week

Posts per day, last 30 days

as of Nov 3, 09:00 UTC
100500Oct 20: 20 postsOct 20Oct 21: 22 postsOct 22: 34 postsOct 23: 29 postsOct 24: 51 postsOct 25: 46 postsOct 26: 23 postsOct 26Oct 27: 38 postsOct 28: 0 postsOct 29: 44 postsOct 30: 0 postsOct 31: 45 postsNov 1: 17 postsNov 2: 0 postsNov 2

What changed

as of 27 Oct · checked once a month

First written 27 Oct. When what people say changes, it shows here.

Tone· 66 posts from 60 people· written 27 Oct

How do people feel about Zapier?

The biggest group of posts is people describing Zaps doing real office work — leads, inboxes, CRM syncs — and the hours they say it gives back. The doubts are about relevance, not bugs: 13 people call it dated or replaceable, while 9 say simple app-to-app jobs are still its lane.

The largest group of posts describes a specific job rather than the tool: leads pushed from a form into HubSpot, Gmail and Slack; quote intake pulled out of a sales inbox; inbox summaries drafted for a person to send; overdue onboarding milestones routed to Slack; Xero payments reconciled against Stripe. People put numbers on it — hours a week freed, a two-hour daily inbox routine cut to minutes, response time halved — and several describe a draft-and-approve loop where a human still sends.

The doubts, from 13 people, are about relevance rather than breakage: that agents and vibe-coded workflows make the drag-a-box model dated, that much of what was sold as AI automation was "Zapier with a new label", and that a prompt or a Claude Code build replaces the subscription. 9 people put it the other way: simple app-to-app connections, fast starts and non-technical teams are the lane, "not built for AI agents". The AI releases — Next Gen Zaps, Copilot, approvals per step — draw welcomes from 8, and 8 speak of the product as solid and hard to copy.

  • Zaps do real work24
  • Doubts or frustrations13
  • Broadly positive11
  • Simple work is its lane10
  • Upbeat about the AI features8
See 5 of 66 posts

Running a family construction company, I've started using bots for the admin grind: inbox sorting, drafting follow-ups and pulling together ideas. Marketing : working with ghl, zapier, Canva and our crm. The rule is they draft and I still hit send. Keeps it our voice and the mistakes low. Still learning but so far I’m amazed.

Before: Sales team spent 3 hours daily manually updating CRM fields from email threads. After: Zapier + GPT automates field population in real-time. Result: Same data, zero manual work, 21 hours freed per week.

ELON JUST PUT GROK, GROK BOT, CURSOR AND X PREMIUM IN ONE PLAN, 3X CHEAPER THAN PAYING FOR THEM SEPARATELY. I CANCELLED 12 SUBSCRIPTIONS THE SAME DAY the x subscription went live this month. one plan, one shared usage pool for all four. what i cancelled, and what does that job now: → chatgpt plus: grok → claude pro: grok → perplexity pro: grok with search → github copilot: cursor → midjourney: grok imagine → elevenlabs: grok voice → otter: grok bot takes my meeting notes → superhuman: grok bot sorts my inbox → zapier: grok bot routines → notion ai: grok bot keeps my docs → grammarly: grok → x premium on its own: it's in the plan now twelve bills became one. the tiers: plus $8 · premium $30 · super $100 · ultra $200 the $30 one covers what used to cost about $90 in separate plans. before you switch: - old plans keep their price, the switch is in billing settings - pick a tier that includes grok bot - usage is shared, so a heavy cursor day eats into grok my first job for grok bot on the new plan: "Go through my email receipts from the last 6 months and list every AI and software subscription I pay for, what it costs, and which ones the X Subscription now covers. Don't cancel anything." full setup in the article ↓

Zapier wins on simplicity and integration breadth. 6,000+ app connections. Setup in minutes. Best-in-class support. Non-technical teams running marketing automations or CRM syncs — Zapier is still the right call. It's not dead. It's just not built for AI agents.

I went through eleven AI releases from the last four weeks and gave each one a verdict for a small online business: use it now, wait, or ignore. Six are worth using now, and most of them sit inside something you already pay for. Claude Opus 5.5 is cheaper to run than Opus 5 by Anthropic's own measure, and in the Claude apps the launch raised the usage limits on Pro and Max. Zapier put an AI step inside the Zap editor that can stop for your approval before each action it takes. Meta's Muse reads your ad account beside Shopify, Stripe and Klaviyo. It's free for most uses, in the US and Canada. ChatGPT Slides builds a deck from your notes and an old deck you liked, on Pro, Business and Enterprise. Shopify switched its agentic storefronts on by default for eligible stores, so your products may already be showing up inside ChatGPT. And TinyFish published the weekday workflow that turns each blog post into two drafts and checks every number against the original before anyone edits them. Three can wait. Notion's agents can now work on one task for 24 hours, and nobody knows what one run costs until the new credits start on November 2. HeyGen's new video model is API-only, with no settled price. Sunra only makes sense if you already pay for two or more image or video tools. Two I'd ignore: Salesforce's agent results, which come with no baselines, and a video QA tool built for video made in code. If you try one thing this week, put the Zapier step on your inquiries. The approval means nothing goes out until you've read it.

5 of the 66 posts behind this answer, one per person. The other 61, and every new one as it’s posted, come with tracking. Get early access

Pain points· 22 posts from 21 people· written 27 Oct

What do people complain about with Zapier?

Price is the complaint most people in this section make: 12 of 21 people. It is monthly plans and per-task charges that climb with volume, a free allowance that runs out, and shock at what others pay; the rest is limits and rough edges.

The price complaints are specific: monthly plans with Zapier Professional quoted at $19.99 a user, per-task bills that rise with volume, and the free allowance — one person's scheduled auto-posting stopped when it ran out, and they wired Claude to Buffer's connector instead. People warn about the bill at scale ("pricing won't surprise you at 10,000 runs"), and a quoted $450 a month drew surprise from someone else. One post walks through how the new AI steps are billed by model tier — several tasks per run, plus several per tool call — and calls the tier a pricing decision for anyone building for clients.

The rest is limits rather than breakage: confusing first steps for someone trying to go no-code, systems that can't be reached without APIs or 2FA, a Chrome extension that was removed, and no way to copy a Zap out as JSON. 2 people raise upkeep — errored steps to replay carefully — and a listing that ties a product's growth to Zapier's uptime and policy.

  • Too expensive13
  • Limits and rough edges7
  • Breakage and uptime worries2
See 5 of 22 posts

Zapier's help docs say new AI by Zapier steps on paid plans default to the Premium model tier, which bills 5 tasks per run plus 5 per tool call. A Premium step that calls two tools uses 15 tasks a run. On the Advanced tier the same step uses 9, and on your own API key it uses 3, with the model cost moving to your provider bill. If you build automations for clients, the model tier is a pricing decision. Set it deliberately for each step and show the client the task math, because a workflow quoted at one task per step can burn several times that on the default.

july 2025: 2,382 posts in r/n8n. last month: 471. that's an 80% drop in 14 months. google searches for "n8n" are down 86% from their peak. but before you call it dead, look at who fell with it: - zapier. searches down 82% from its peak, which came the same week. - make (.com) rounds to zero on google trends now. - claude code. searched 5x more than n8n today. in august 2025 it was the other way around: n8n got 1.8x more searches. n8n didn't lose to a competitor. the whole "drag boxes around to connect apps" category lost to "type what you want." the 2025 wave was mostly people learning n8n so they could sell "ai automations" to small businesses. that pitch is the thing that died. the client can open claude and describe the workflow themselves now. what's left is the part nobody made youtube videos about. read the replies from people who still say it's worth it and almost every one is about fixing things: a google token that expired on day 7, a sheets node that wrote empty cells after someone renamed a column, a webhook payload that changed shape at 2am. ai made building a workflow free. it didn't make running one free. the money moved from whoever builds the automation to whoever gets the call when it breaks. and a team where 5 people touch the same automations still wants logs, access control and a picture of the flow they can all read. that's where n8n keeps its place. n8n isn't dead. the n8n tutorial economy is.

5 of the 22 posts behind this answer, one per person. The other 17, and every new one as it’s posted, come with tracking. Get early access

Alternatives· 35 posts from 34 people· written 27 Oct

What are the alternatives to Zapier?

n8n and Make are the named alternatives, level at 8 and 8 people, with building it yourself close behind at 7. Beyond those, the switching is to Grok Bot (4 people) and a few Claude or MCP setups.

n8n is argued for on self-hosting and cost: its free Community Edition is said to cost nothing and to take code when a node is missing — though one person says Zapier is still easier to hand over to a client. Make is argued for on visual multi-step builds and price, better for complex non-AI flows at around $12 a month against $19.99.

The self-build argument is that the same job can be coded, vibe coded or run on a homelab instead of paid for, and one person asks whether the whole category is still used. The most-read post here is a list of cancelled subscriptions with Zapier replaced by Grok Bot routines; a few others name Claude, Claude Code or an MCP connector. Not everyone leaves: comparison posts keep sending the quick app-to-app job to Zapier, and one agency owner doing $300k a month is described as still on it.

  • n8n8
  • Make8
  • Build it yourself7
  • Grok Bot4
  • Claude / Claude Code3
  • ChatGPT1
See 5 of 35 posts

Make wins on visual power and value for complex non-AI workflows. The scenario builder handles multi-step conditional logic better than Zapier. At ~$9/month for 10k operations, the price-to-power ratio is hard to beat — if you're not building AI agents.

it's becoming fucking dangerous. Chinese Dev dropped kimi k2 at full open weights, so i wired 48 used gpus into one homemade lab with a single rule: "if a chinese model fits across the rack, runpod never gets another invoice" six rigs in a pine-rack frame cooled by one duct and an open window, each with its own job: rigs 1 and 2 pool over rpc into 192gb of vram and serve kimi k2 for every claude api call my sdk still thinks it's making rig 3 is the lora trainer, fine-tuning on 40k support tickets in 3 hours instead of the $2,308 runpod quoted for the same run rig 4 runs 120 agents that port cron jobs, review prs, fix n+1 queries and draft 60 cold emails before i'm awake → a model is too big for one rig: it gets sharded over rpc, so vram scales horizontally and runpod gets canceled → a fine-tune costs more than a day of electricity: it stays on rig 3, so lora training stops being a cloud expense → a task repeats more than twice a week: it gets handed to the 120 agents on rig 4, so zapier dies with it a homemade lab can out-ship a cloud subscription if the room has one open window and the willingness to run cables Kimi serves. Lora trains. Agents ship. every call has to show: > which rig served it > which chinese weight was running > which cloud invoice it just made disappear paste this into your startup script: "Pool rig 1 and rig 2 over rpc into one 192gb endpoint. Serve kimi k2 as the openai-compatible api. Keep rig 3 for lora fine-tunes on local data only. Give rig 4 to 120 agents with write access to code, email, and the crm. Rate limit: none. Open the window" start with the one api bill that funds the biggest chunk of your month, not the biggest headline model full setup below

ELON JUST PUT GROK, GROK BOT, CURSOR AND X PREMIUM IN ONE PLAN, 3X CHEAPER THAN PAYING FOR THEM SEPARATELY. I CANCELLED 12 SUBSCRIPTIONS THE SAME DAY the x subscription went live this month. one plan, one shared usage pool for all four. what i cancelled, and what does that job now: → chatgpt plus: grok → claude pro: grok → perplexity pro: grok with search → github copilot: cursor → midjourney: grok imagine → elevenlabs: grok voice → otter: grok bot takes my meeting notes → superhuman: grok bot sorts my inbox → zapier: grok bot routines → notion ai: grok bot keeps my docs → grammarly: grok → x premium on its own: it's in the plan now twelve bills became one. the tiers: plus $8 · premium $30 · super $100 · ultra $200 the $30 one covers what used to cost about $90 in separate plans. before you switch: - old plans keep their price, the switch is in billing settings - pick a tier that includes grok bot - usage is shared, so a heavy cursor day eats into grok my first job for grok bot on the new plan: "Go through my email receipts from the last 6 months and list every AI and software subscription I pay for, what it costs, and which ones the X Subscription now covers. Don't cancel anything." full setup in the article ↓

Je suis certain que ça tourne toujours, nous les solopreneurs on à zéro friction a changer et encore, quand je suis passé de zapier à Claude en vrai déjà à mon niveau j'ai mis des mois parce que j'avais tellement d'interconnexions que le moindre changement faisait tout capoter

5 of the 35 posts behind this answer, one per person. The other 30, and every new one as it’s posted, come with tracking. Get early access

Questions· 10 posts· written 27 Oct

What do people ask about Zapier?

The biggest question is whether Zapier is still needed now that agents and vibe-coded workflows exist — 4 of 10 people ask some version of it. The rest is practical: what people use Zaps for, whether the Zapier MCP works, and one request for Agents invites.

People ask whether anyone still uses Zapier and n8n, say they moved on because workflows "can just be vibe coded now", and wonder whether agents are coming for no-code platforms or are just custom connectors. A vaguer version — "What happened to Zapier?" — sits alongside them.

The rest is hands-on. 3 people ask what others use Zapier for, ask whether someone knows about Zapier triggers, and share a tip on replaying errored steps from Zap history rather than rerunning the whole Zap. 2 people ask about the Zapier MCP or plugin — what it is, whether anyone is using it, whether they like it and what they don't. One person asks whether anyone has Zapier Agents invites to hand out.

  • Is Zapier still needed?4
  • Usage questions and tips3
  • Does the Zapier MCP work?2
  • Getting access to Zapier Agents1
See 5 of 10 posts

Anthropic just debuted Claude Duet, entering the background agent space alongside Muse and Dots. Are these AI agents coming for traditional #NoCode automation platforms (Zapier, Make), or are they just glorified custom connectors? Late to the party, or a game-changer for citizen developer? #NoCode #LowCode #Automation #ClaudeDuet #AI link

Zapier tip: replay errored steps from Zap history to avoid rerunning successful steps. An entire-Zap replay creates a new run and repeats them. Check prior outcomes before retrying anything consequential. #Zapier #AITools

5 of the 10 posts behind this answer, one per person. The other 5, and every new one as it’s posted, come with tracking. Get early access

Wishes· 3 posts· written 27 Oct

What do people want from Zapier?

Three wishes, each from one person and none repeated: export a Zap as JSON to edit with an agent and paste back, add an MCP so agents can build and drive Zaps, and a shared support session.

One person asks how it can be that they can't copy a Zap out as JSON, have an agent change it, and paste it back — the wish is for a Zap to be an editable file rather than something locked in the editor. One wants Zapier to add an MCP, calling it a huge opportunity and saying they are surprised it took this long. One wants a shared support session so users are spared the "please explain your issue again" ritual.

Read together, they all point the same way — handing Zapier's building blocks to agents and to other people: an editable format, a machine-readable interface, and support that carries context between sessions. No wish here is repeated by a second person, so none of them is a pattern yet.

  • Export a Zap as JSON1
  • Add an MCP1
  • Shared support session1
See the 3 posts

In their words

How do people talk about Zapier?

The phrases people use most, counted once per person.

  1. 1“n8n or zapier”4
  2. 2“ai agents”4
  3. 3“automation tool”3
  4. 4“zapier make n8n”2
  5. 5“app-to-app connections”2
  6. 6“repetitive work”2
  7. 7“zapier mcp”2

Posts are shown the way X shows them. One of them yours? Here’s why it’s shown, and how to take it off.

Also in Workflow automation

Get every post about Zapier, filtered and labelled

This page is the summary. Tracking gives you the stream behind it for any brand, keyword or @mention: each new pain point, request and question as it’s posted, with the spam and promotions already taken out.