About
Co-founded Zapier in late 2011 with good friends Mike and Wade, which was admitted in…
Activity
6K followers
-
Bryan Helmig reposted thisBryan Helmig reposted thisForbes asked me what our employees spend on AI. That’s easy to track. ROI, not so much… I hopped on a video chat with Tim Keary to chat token spend and open up Zapier's numbers: - Most employees spend hundreds of dollars a month - Most devs spend several thousands a month - Our top AI builders spend as much as $30,000 a month Keep in mind, those high-spend devs are looping coding agents at scale on greenfield projects, fixing bugs and building new functionality. We’re all trying to solve for ROI here. According to Ramp, the top 1% of companies spent a median of $7,400 per employee on AI in July, up from $2,590 in January. OpenAI's own research found revenue per employee wasn't meaningfully associated with an employee's token output. Our own AI Workflow Index points the same way. Across 1,500 companies on Zapier, the leading AI adopters run AI in only about 18% of their agentic workflow steps. The rest runs on code and logic. So you can be a top AI builder without shredding through tokens, especially when you use AI to write the code and build the automations (which then run cheaper than the AI would) But to be clear - you can also be a top AI builder who uses a ton of tokens. More and more of what we ship gets iterated on by agents running in loops, and that does not come cheap. That’s why it’s some of our most intentional and scrutinized work, with protections and guardrails in place. Everyone they talked to can tell you what they spent but not one of us measures ROI the same way. Full article here: https://lnkd.in/giJ5-83N
-
Bryan Helmig reposted thisBryan Helmig reposted thisGemini 3.5 Flash ranks #1 on Automation Bench (from Zapier), beating every other frontier model at a much lower cost https://lnkd.in/gEbN8RdJ
-
Bryan Helmig reposted thisBryan Helmig reposted thisWe’re setting a new standard at Zapier. 100% of new hires must be fluent in AI. How? By giving 100% of applicants concrete ways to grow their skills. Each candidate now gets access to AI training materials and guidance. Once hired, they’ll experience our revamped onboarding process focused on product mastery + building tangible AI skills. Higher expectations. Better training. Stronger support. We're raising the bar for AI fluency, and everyone deserves a chance to meet it. Want to join Zapier? Check out our open roles 👇
-
Bryan Helmig shared thisWhat? ZAPIER AGENTS is evolving! Agents now work in the background to handle tasks continuously, across your entire business. We've reimagined the automation experience from the ground up: ⚡️ One agent, one job. Each agent has one role, simplifying setup and management ⚡️ Group related agents in "Agent Pods" to organize similar tasks ⚡️ Real-time activity dashboard for total visibility ⚡️ Faster build experience w/ new prompt assistant (and 20+ templates to get you started) We've evolved AI chat into autonomous action at enterprise scale. Can't wait to see how you use it! Btw - already using Zapier? You're already upgraded. Just log in and go. (Link in comments!)
-
Bryan Helmig reposted thisBryan Helmig reposted thisBig MCP news from OpenAI today with Zapier MCP as a launch partner! 🚀 Less than a month ago, OpenAI's CEO said MCP support was coming to the their new flagship API. 👀 That day is today! OpenAI's Responses API can now leverage MCPs as tools without developers needing to build their own MCP Clients. 😎 Zapier MCP is one of the first supported MCP servers bringing in nearly 8,000 Apps that builders can now easily use with the Responses API. 📈 Oh...and did I mention it all works from the OpenAI Playground! 🤯 To the non-developers, this is a significant moment for standardization of AI tooling which will lead to more connected experiences for you across the apps you use! 🧡 I'm linking to OpenAI's official announcement and Zapier's MCP below. 👇
-
Bryan Helmig shared thisNew launch today: Zapier's AI orchestration platform for enterprises! Over 200 million AI tasks automated, and we’re just getting started. We’re doubling down on our mission to make AI orchestration simple, scalable, and accessible for every team. Why? Because when you remove the technical barriers, AI can drive real outcomes: - Revenue recovered. - Hours saved. - Teams empowered. This is what’s possible when you put connected AI agents to work, exactly where your team needs them. Want to learn more about these workflows and their impact? https://lnkd.in/gp2nFXbSAI orchestration: How to scale AI across your businessAI orchestration: How to scale AI across your business
-
Bryan Helmig shared thisCurious how to get started with MCP? You’re not alone. There’s plenty of buzz (for good reason) but I wanted to share a simple way to get started using Zapier Agents. Zapier Agents work with MCP to connect your AI assistant to nearly 8,000 apps and 30,000 actions. Instead of your AI assistant telling you what to do, it can actually go do it. Imagine setting up triggers so your agent automatically responds when something happens, like a new row in Google Sheets, a new event on your calendar, a new lead in your CRM, etc. Better yet, customize the agent so that if a lead seems promising, it can take one action. If not, it takes another. It’s like giving your AI a to-do list AND the power to check things off. It works with ChatGPT, Claude, Windsurf, or whatever platform you’re building on. Plus it’s free to get started. (And check out our ready-to-use template library to kick things off quickly) What’s the first thing you’d have your agent do?
-
Bryan Helmig posted thisStruggling to know where to start with AI workflows? At the Zapier Summit in New Orleans a few weeks back, nearly 600 Zapiens gathered to discuss the future of AI and automation. In our keynote, I shared a breakdown of AI tools and approaches to help people understand where each option fits in. Here's the framework I used: 1. Craft & Career: How can AI give you superpowers in your work? 2. Colleagues & Collaboration: How can AI empower your team? 3. Customers & Capabilities: How can we bring AI's power to customers? And how do you know what tool to use? ➡️ Automate Workflows: AI by Zapier, ChatGPT, and Gemini power automated workflows with AI Agents, chatbots, and AI-powered Zaps. ➡️ Standalone AI Assistants: For general-purpose AI help, tools like ChatGPT Enterprise and Google Gemini are great options we all know and love. ➡️ Prototype & Experimentation: If you're testing and communicating new ideas, tools like Claude Artifacts, Replit Agents, and Bolt.new make rapid prototyping easier. ➡️ AI-Powered Developer Tools: If you're writing code, AI-powered IDEs like Cursor, Windsurf, and Lovable.dev can supercharge your development process. ➡️ Learning & Research: Want to go deeper? ChatGPT Deep Research, NotebookLM, and AI-driven research assistants can expand your knowledge of AI and more.. ➡️ Specialized AI Tools: Apps like Aqua Voice and Superwhisper help with transcription, accelerating your ability to inject context into your AI tools. Which tools do you want to learn more about?
-
Bryan Helmig shared thisSome exciting news today -- we're launching Zapier MCP, giving developers instant access to Zapier's 8,000+ apps and 30,000+ actions for their AI agents. It's dead simple: generate an endpoint, select actions, and connect your AI. Works with Claude, Cursor, Windsurf, or any platform you're building on. Your AI can now send Slack or MS Teams messages, manage Google Calendar events, interact with Github or Gitlab -- or whatever else you need, all securely authenticated. If you're a developer looking to do more with AI and Zapier, I'd love to hear what you build with this! Check it out: zapier.com/l/mcp
-
Bryan Helmig liked thisBryan Helmig liked thisForbes asked me what our employees spend on AI. That’s easy to track. ROI, not so much… I hopped on a video chat with Tim Keary to chat token spend and open up Zapier's numbers: - Most employees spend hundreds of dollars a month - Most devs spend several thousands a month - Our top AI builders spend as much as $30,000 a month Keep in mind, those high-spend devs are looping coding agents at scale on greenfield projects, fixing bugs and building new functionality. We’re all trying to solve for ROI here. According to Ramp, the top 1% of companies spent a median of $7,400 per employee on AI in July, up from $2,590 in January. OpenAI's own research found revenue per employee wasn't meaningfully associated with an employee's token output. Our own AI Workflow Index points the same way. Across 1,500 companies on Zapier, the leading AI adopters run AI in only about 18% of their agentic workflow steps. The rest runs on code and logic. So you can be a top AI builder without shredding through tokens, especially when you use AI to write the code and build the automations (which then run cheaper than the AI would) But to be clear - you can also be a top AI builder who uses a ton of tokens. More and more of what we ship gets iterated on by agents running in loops, and that does not come cheap. That’s why it’s some of our most intentional and scrutinized work, with protections and guardrails in place. Everyone they talked to can tell you what they spent but not one of us measures ROI the same way. Full article here: https://lnkd.in/giJ5-83N
-
Bryan Helmig liked thisBryan Helmig liked thisToday, is the start of a new chapter for me: I’ve joined Ashby as their first VP, Product. I spent over a decade leading Product at Zapier. I'm not one to jump around between jobs, so when I decided it was time for something new, it had to be a company I could bet on for the long term. For me, that bet is Ashby, and I wanted to share more on why I'm making this my next long-term home. https://lnkd.in/g-9mG2Xx
-
Bryan Helmig liked thisBryan Helmig liked thisI'm beta testing code-based Zapier automations and... this is big. Not sponsored!! Traditional Zapier interface WAS revolutionary. I've been using and loving it for over 10 years but times have changed... It works great for humans but for AI agents it's a black box. Case in point: I've been working with a customer for two years now. In that time, I built their automation infra comprised of 40+ Notion automations and Zaps. It's beautiful when it works but dammit it breaks so often🥲 When it breaks, I'm the only one who knows which step in which Zap to check, which part of the history to dig through and re-run and fix. Yah I've written docs for all of it, but nobody has read it (including myself) and made sense of it. And things break for a million reasons; sometimes undocumented edge cases, sometimes steps get updated and deprecated sans warning. No agent can touch it, so I thought Zapier was cooked. Code-based Zaps is the sh*t. The automation IS the code. Code-based Zaps are just code text that's readable, writable, debuggable by any LLM, which means an AI agent can look at your broken zap, understand the logic, find the issue, and fix it. Now you can troubleshoot right from chat, monitor from chat, evolve through chat. They even built an import feature for existing zaps. Big fan. Every ops team deep in Zapier should try this!! If you've got automations that only one person on your team understands, that's the real problem. I'm migrating every Zap I've built for this client to code-base and you should too. P.S. Turns out, Zapier/Make/n8n was never "no code" and the real "no code" is the friends we made along the way on Twitter 😉
-
Bryan Helmig liked thisBryan Helmig liked thisOne giant AI agent running your company is a terrible idea. It sounds futuristic. In practice, it is often just a more expensive, less reliable way to do things software already knows how to do. The real opportunity is not putting AI in every step. It is knowing exactly which step needs AI. Zapier studied how leading adopters are building AI into real workflows, and even in those workflows, only 18% of the steps are AI. The rest are rules, logic, APIs, app actions, data movement, and handoffs. That is not underusing AI. That is good architecture. Let AI handle the messy part that requires interpretation, judgment, or language. Let automation handle everything that should happen predictably every single time. In Zapier’s modeling, that approach was 71% less expensive than routing the entire workflow through a model. And the typical AI workflow still handled more than twice the automated actions of a conventional one. My POV is that the companies that win with AI will not have the biggest, most complicated agents. They will be the best at designing the division of labor between AI, automation, systems, and people. AI writes. AI extracts. AI decides. AI coordinates. Then the workflow takes it from there. Zapier just released a genuinely useful report on where AI belongs, what companies tend to build next, and how governance needs to change as AI moves from producing information to taking action. Also, a huge shoutout to Andre Vanier, who spent months digging through the data, finding the patterns, and turning a mountain of information into something leaders can actually use. Forget AI trend pieces. This is a super practical map for getting real work done with AI. Grab it here: https://lnkd.in/eEM4zrih
-
Bryan Helmig liked thisBryan Helmig liked thisGemini 3.5 Flash ranks #1 on Automation Bench (from Zapier), beating every other frontier model at a much lower cost https://lnkd.in/gEbN8RdJ
-
Bryan Helmig liked thisBryan Helmig liked thisI asked an AI to read 4 years of my text messages with my closest family members to tell me who I really am. Expected: "Loving father, devoted son." Got: "Emotionally efficient. Indispensable but not exposed." Cool cool cool cool. This week I built a small tool that lets Claude read my Apple Messages history. The tool is open source if you want to use it: https://lnkd.in/gu38yNZi Then I asked it to build psychological profiles of my closest relationships. My wife, brother and parents. What started as a hack turned into the most confronting therapy session I've ever had. Here's what the AI found: With my dad: I respond but rarely initiate. I provide tech support but not emotional engagement. The AI asked: "What needs to be said before he dies?" With my mom: I say "love you." I engage. I'm warmer. The contrast with my dad was... visible. With my wife: We've developed a private language so compressed it would be incomprehensible to outsiders. We communicate in pure signal because we don't need context. We're already inside each other's context. The pattern across everyone: I'm the responder, not the initiator. The helper, not the one who needs help. The emotionally efficient one. The AI called it "a life where you're indispensable but not exposed." Brutal. And true. I've been in actual therapy. This hit different. Not because AI is smarter than my therapist. It's not. But because it analyzed behavioral data, not my narrative about myself. It saw patterns across thousands of messages that I couldn't see because I was too close to them. Some takeaways if you try this yourself: 1. Your texts can reveal your attachment patterns. Who you reach toward. Who you keep at arm's length. It's all there. 2. What you DON'T say can be as telling as what you do. 3. AI as mirror is genuinely useful. Not as replacement for human connection, but as a pattern-recognition engine that shows you what you're too embedded to notice. 4. It only works if you're willing to hear it. I asked for "brutal honesty." The AI delivered. Most people don't actually want the honest version. The tool is an MCP server that lets you search, browse, and analyze your iMessage history. Sets up in 30 seconds. Fair warning: you might not like what you find. But you might need to find it anyway.
-
Bryan Helmig liked thisBryan Helmig liked thisYes, the dance moves are back (sans puppets). AI Trainer Trainer Ep. 3 is live : Matrix-themed + orange-pilled. It’s about how Zapier lets you run agents and deterministic workflows. No choosing sides. Just power, all in one place. Take the orange pill. Here we go. What else would you like me to school y'all on in this fun new format?
Experience & Education
Other similar profiles
Explore more posts
-
Malik Dixon M.S.
🔧 Technical Update: SSD Consulting Website + LMS Redesign The new site rebuild is running on Next.js 13+ with modern React patterns, strict TypeScript, and an LMS-ready architecture. Recent code review results: ✅ Code Quality & TS: 9/10 🔐 Security: Hardened (noted improvements around storage & auth) ⚡ Performance: Optimized imports, image handling, and layouts ♿ Accessibility: Semantic HTML + Tailwind-first UX 🧪 Next step: Adding unit/integration tests Approved for production with minor fixes—now focused on security hardening, testing, and scale.
-
Abby Grills
Need web data to power your product? Riveter replaces your entire scraper infrastructure with a single API. Here's a fun example of Cody Watters building a vibe coded app with data pulled by Riveter. It's never been easier to build. What has your experience been building with scrapers?
5 Comments -
Rhonda Coleman Albazie
🧠 Examples of How R&D Tax Credits Reduce HNW Taxable Income : How They Work Federal Income Tax Offset • If your business has a $100,000 federal tax liability, and qualifies for a $50,000 R&D credit, your federal tax owed drops to $50,000. • This is after deductions not instead of them. Payroll Tax Offset (Startups & Small Businesses) • Qualified small businesses can elect to apply the R&D credit against payroll taxes which is essentially immediate cash flow relief. State Tax Credits • Additional credits can offset your state income tax, reducing the overall blended federal + state rate. Carryforward/Carryback • Unused credits can often be carried forward (up to 20 years) or back one year, preserving your savings over time. #ResearchandDevelopment #TaxCredits #AI #Innovation #Robotics #Taxes #IRS #HNW #HighNetWorth #HighIncome
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content