Zapier

Just recorded a webinar on Zapier's new AI Workflow Index. A finding that got a lot of reaction: among our leading adopters, only 18% of the steps in an AI workflow actually run on AI. The other 82% is pedestrian. Rules, logic, app connections, moving data from one place to the next. This is old fashioned automation. When we modeled it, this approach cost 71% less to run than routing every step through a model. Hundreds of companies across our panel are basically converging on the same idea: your AI architecture is your strategy. In the webinar, I go deep with Rebecca Hinds, PhD and Ryan Anderson on what this approach actually looks like inside workflows. And how controls have to change once AI stops drafting messages and starts making decisions. Where's this wrong? If you've built AI into real workflows, I'd love to hear where selective beats all-in and where it doesn't. Recording here: https://lnkd.in/gbyRUYGa

The tokenmaxxing era is over ‼️ Steps in this workflow that actually need AI: 1 Steps you can just automate: 25 The whole thing doesn't need to be agentic...and it'll be ~90% cheaper

Unbelievably grateful to Rob Ayre and Zapier for a special feature about how I run my AI-native legal practice, FTE Legal, a Social Purpose Corporation, which was news to me in more ways than one. When Rob emailed asking to chat about my Zapier usage, my first thought was that I had broken something. Evidently, 12,000+ MCP tasks coming from a lawyer’s account is particularly unusual. Perhaps. But honestly, I was just irritated with doing the same stuff over and over, every single day. Manually entering client requests received by email as tasks in Clio. Downloading PDF redlines from Slack to my downloads folder, converting to docx, revising in Word, saving to Clio, then emailing back to the client. You know the routine. I just kept picking off whichever part of the day was annoying me the most that week and handed it to Claude and Zapier. Apparently, that adds up to a special feature (which, frankly, is pretty cool). The bigger thing I didn’t see coming is what it did to my billing. I’m hourly. For the last several decades, that model has quietly rewarded lawyers who took the longest. Now that the work moves faster, the delta goes to the client instead of into a partner’s second vacation home. And I’d rather be on the client side of it, so I'm staying hourly for the foreseeable future. (Follow me for my thoughts on why the rush to fixed-fee models is spectacularly ill-timed.) So thank you to Rob Ayre and the team at Zapier for the write-up, and for noticing something unique about my badgering Anthropic's Claude to do the stuff that I don't want to do, and do it right. The time it frees up goes into other projects I’ve wanted to pursue but had little time for. Ergane LLC, a legal efficiency consultancy. Ariadni LLC, an AI-powered lawyer oversight tool for non-lawyers. And of course, pursuing my sommelier training and donning my apron to serve the community at Book Society. I also get the time to teach my in-house clients and other solo lawyers what I know (even those lawyers with whom I technically compete) through a free 8-week AI and automation seminar (link in the article). Today's webinar coincidentally features how I use Zapier. Anyways, life is good, thanks to the tools that do the busy work for me. https://lnkd.in/ggYZKp_d And to be safe, the disclaimers: Attorney Advertising. Prior results do not guarantee a similar outcome. Not legal advice.

A shared brain that's answered 1,300 questions. An agent clearing 47,000 tickets a month. Watch these three Zappy-winning builds happen live at ZapConnect 👇 Patreon's Blair Mishleau turned 600+ internal resources into 'Treon, a shared context layer that's answered 1,300 team questions. He joins Eric Mistry to show you how to build your own personal context layer that's just as powerful. Eric McNulty's vendor AI couldn't handle real support tickets, so he built his own: a ~90-step orchestration at Mercari US that now handles 120,000 tickets a month. He'll walk you through the guarded-agent pattern he runs in production. Jeff Hirsch rolled out Zapier at Redis and within weeks, 5 teams were on it (with SSO and SCIM locked in from week one). He'll show you how to stand up enterprise Zapier so teams can move fast without anyone losing control. Join us at ZapConnect on September 23! It's free, virtual, and BYO keyboard, you're building too 👀 Register → https://lnkd.in/gTY-utR8

Does $5,000 in AI spend equal $5,000 in value? At Zapier, it's my job to find out. Much of it is automated already: - Every Zapier employee gets an AI spend limit - Hit 75% and get a DM: tell us what you're working on - If it checks out, more budget is added automatically No tickets, no meetings. Lots of our heaviest AI users internally don't even hit this limit, especially since we started educating more on model routing, reasoning levels, etc. Zapiens know when to use the smart model, and when to use the fast/cheap one. The harder cases are the ones the system flags instead of auto-approving, and that's where I come in. Certain roles will need much higher spend, and we've planned ahead for that. Here's the case that still doesn't have a clean answer: did the $5,000 an engineer spent on code generation last week produce $5,000 of net new functionality, in customer satisfaction, adoption, engagement? Good luck answering that directly (I've tried) Even Zapier's not there yet.

An enterprise customer of ours just won their company's hackathon. He used Zapier to audit hardware from photos. Nick Francoeur at Compucom manages hardware audits across 700+ locations. Reviewing every scanner, pin pad, monitor, etc at a site used to take hours. Now a tech photographs the gear and a Zap runs each photo through AI by Zapier. It pulls the make, model, and serial number, then logs it in a Zapier Table. Early results have it at 90-95% accurate. A daily Zap flags anything the AI's unsure about and pings Nick for human review. He's more than halfway through all 700+ locations, on track to finish by mid-August. Plus, his build just won Compucom's company-wide hackathon. Best part is he's already planning the next version: one that catches its own gaps before a human has to. All with Zapier, stay tuned.

Meta's algorithm gets smarter every time it sees a real conversion. Zapier makes it easy to send those signals straight from your CRM, no code required👇 Together, we're helping advertisers close the loop between their sales data and their ad performance with server-side events But the real unlock is knowing which events to send, when in the funnel, and what data to attach for the highest match rate Join Meta's Rafael Curran and Zapier Solution Partner Jared Weiss as they walk through the strategy live, and show you how to set it up through Zapier in under an hour Thursday, August 13 at 1 PM ET Register here:https://lnkd.in/g_NiywHC

You can probably just automate most of what you're handing to an agent right now 👀

How to build a GTM AI system your sales reps will actually use. (Without replacing your CRM or hiring a data team.) Most GTM AI rollouts fail in one of two ways: → The tool is too centralized. Reps ignore it. → Every rep builds their own thing. It drifts from company policy. Here is how to run both in parallel connected through a shared context layer. The core idea: one Google Sheet holds all definitions, thresholds, and risk signals. A Claude agent reads it automatically before every call. Rep-facing skills read the same sheet on demand. When RevOps updates a threshold, it changes everywhere instantly. Here is how to build it. 1️⃣ Create the Context Layer Build a Google Sheet with 4 tabs: Definitions, Risk Signals, Account Context, Request Log. One rule enforced in code: prompts contain instructions only. No ARR threshold, no risk signal name is ever hardcoded. 2️⃣ Connect via Zapier SDK One npm install. No OAuth to manage. Connect Google Sheets and Gmail in your Zapier account. The SDK handles all reads, writes, and email sends. 3️⃣ Run the Centralized Agent A pre-call prep agent reads the sheet and sends a Claude briefing to the rep's Gmail before every call. No rep action required — it just arrives. Built-in guardrail: the email is blocked if Claude mentions a risk signal that is not active on the deal. 4️⃣ Give Reps Skills On Demand Deal Strategy: rep asks about a stalled deal, gets a structured action plan. Account Planning: generates a 14-day plan with priorities, risks, and one immediate next step. New skills are built from a single template no shared infrastructure to rewrite. 5️⃣ Log and Measure Everything After every skill response the rep rates it with y/n. Every run is logged to the sheet automatically: eval feedback, groundedness check, latency, cost. The KPI dashboard is generated from that log via the same Zapier SDK reads. ↳ What makes this different from a standard AI rollout → One context layer — one update changes every agent and skill instantly → Reps adopt it because it fits how they already work → Every run is instrumented from day one — no blind spots → No new database, no BI tool Google Sheets does the work → The system grows bottom-up: personal skills that work get promoted, not mandated ↳ What you need to start → Node.js 18+ → A Zapier account with Google Sheets and Gmail connected → An Anthropic API key → One Google Sheet with the 4-tab structure The full system runs in under 200 lines of TypeScript per skill. Built with Claude Code + Zapier SDK. #gtm #revOps #zapier #SDK