Airtop

📣 We’re excited to announce that Airtop Web Agents now run inside Make workflows. Your Make scenarios can now automate the entire web. Make is a powerful platform for connecting and automating thousands of integrated apps. But what if you need to build automations that can click, login, and interact with web pages? That’s where Airtop Web Agents change everything for Make users. Make workflows can finally take real actions on websites that don’t have APIs. Airtop Agents allow you to: 👉 Log into vendor portals 👉 Pull data from gated pages 👉 Fill out multi-step web forms 👉 Scrape structured data from dynamic sites And many more web automations that we haven’t even thought of yet. 🔌 How it works 1️⃣ Add the new Airtop “Run Agent” module inside Make. 2️⃣ Create an Airtop Agent by describing what you want to automate in plain English. For example: “Go to any public Loom URL, extract the transcript, and output structured JSON.” 3️⃣ Connect the Airtop Agent to your Make scenario and pass parameters. You can now drop a real browser agent directly into your Make scenario – one that sees web pages, clicks buttons, types into fields, scrolls, and extracts structured data. And the best part? You can build an Airtop Web Agent in just a few minutes by chatting with Airtop’s AI Agent Builder. It’s as easy as talking to ChatGPT. If your Make workflows ever get blocked by “this site or app doesn’t have an API,” that problem is officially solved with Airtop Agents. 🎥 Full tutorial video + docs in the comments. 💬 Drop a comment if you want an invite to a live walkthrough.

This is BEYOND COOL. Think Limitless pill 💊 for your Zapier AI Agents and workflows to do stuff in the browser. Things I'm doing with it: 🔥 Zapier Agent that uses Airtop to extracts the transcript from a loom video. 🔥 Agent to fills out google forms for events using data it already has about me 🔥 Agent that reviews marketing dinner invites, sees who I know on the list, and RSVPs for me

Your competitors just automated 40 hours of work into 2 hours with browser agents. You're still doing it manually.   Airtop's browser agents launched in September. And they’re already changing the game. Here's what I'm building for clients with Airtop: - Sales: 6-hour lead lists → 15 minutes automated - Growth: 2-day enrichment → Real-time ICP scoring - Marketing: Monthly competitor reports → Live Slack updates - RevOps: Login-blocked data → Seamless extraction The gap is widening. Fast.   Most teams are still wrestling with Captchas and login walls. Smart teams automated past all of it just by chatting with AI.   I built the complete guide: 8 proven browser automations.   The exact system to: → Build powerful lead lists with ICP scoring automatically → Monitor competitor activity and deliver insights to Slack → Extract data from sites that require OAuth, 2FA, or Captcha → Automate web interactions using natural language prompts → Support use cases that were previously impossible   This is your automation advantage with Airtop.   Teams deploying browser agents today will be months ahead by Q2.   Effortless automation isn't an edge anymore. It's the requirement.   I'm giving away the complete automation package for free.   To claim it: - Comment "AIRTOP" - Like this post - Connect with me for the DM   You get: 8 ready-to-use template automation links + 1 month FREE Airtop Premium access.   Your competitors are already automating.   Repost this so your network doesn't get left behind ♻️

Airtop’s Agent Builder just got a lot faster. 3X faster. Upgrading a core model is like swapping an airplane engine mid-flight… but we didn’t crash. In fact, things got noticeably better. This week we changed the Agent Builder’s conversational model to Anthropic’s newly released Haiku 4.5. Speed matters. And this new model is fast. Our users have seen a 3X-5X speed improvement in Agent collaboration compared to the previous model (Sonnet 4), along with a 34% increase in successful agent deployments. We also tested Sonnet 4.5. It’s smart, no doubt, but harder to steer. Think brilliant but over-eager toddler: powerful, but not always doing exactly what you meant. Haiku 4.5 gave us the best balance of speed and control, which translates directly into a better UX for users. There is a lot of trial and error in upgrading to a new set of models, and although we have evals to help guide us, it does feel, at times, part art, as well as science. The conversational model is only a part of our story. We use a complex combination of models (OpenAI, Anthropic and Google Gemini) for our browser interaction APIs, as well as our MCP agents, all of which are orchestrated by our Agent Builder. This careful combination of models, taking advantage of each model’s strengths, while minimizing their weaknesses, is a key component to building successful agentic products. Every upgrade comes with tradeoffs. This one is a clear win. Try it out and tell us what you think.

Airtop は自動化を民主化します 🌏 Automation shouldn’t have a language barrier. Airtop’s Agent Builder can understand (and respond in) dozens of languages. So even if you’re not comfortable with English, you can easily build useful web automations. Our community has built agents in Spanish, Portuguese, French, Japanese…all thanks to the magic of LLMs. Our mission at Airtop is to democratize automation, and having the Agent Builder be language agnostic is a huge part of that. You can even switch languages mid-stream or speak in a mixture of multiple languages that you’re comfortable with, and the Agent Builder will catch your meaning – como si nada. If English isn’t your first language, try it out in your native tongue!

B2B Marketers: Stop Spamming! 🛑 The only people worth reaching out to are the ones already engaging in conversations about the problems you solve. That’s why I built a simple automation that finds warm, high-intent leads from actual online discussions. Here’s what it does: 1️⃣ Reads your target list: Starts with a Google Sheet of thought leaders, competitors, or key voices your audience follows. 2️⃣ Finds relevant posts: Scans each profile’s recent posts and filters only those matching your chosen topics (like “AI agents” or “growth marketing”). 3️⃣ Extracts engaged leads: Pulls profiles of people who comment on those posts - complete with name, title, company, and LinkedIn URL. 4️⃣ Cleans & saves: Removes duplicates and adds new leads to your sheet, ready for outreach. 💡 The result: a constantly updating list of people showing real buying intent. No spam. No guesswork. Just timing that’s right. It’s endlessly tweakable, you can set it up to: - Filter leads by fit. - Find verified email addresses. - Send personalized connection requests automatically. 👉 Try it: Link in the first comment 👉 Drop a comment if you’d like me to show you how I set it up or help you tailor it for your niche.