AI versus Marshy - Research on Autopilot with Zapier Central

If you’re new to Zapier, it’s a tool that helps connect different apps and services to automate tasks without needing to code. I’ve been using Zapier for a while now, and I’ve found it to be incredibly useful. In this article, I want to explore a new feature called Zapier Central, which has the potential to take my newsletter creation process to the next level. It’s a tool that acts like duct tape for your software. You want your email sign-ups to push into your CRM? ​ There’s a Zap for that. You want someone to receive a text adddress them by their name and linking them to the new buyer guide they signed up for on a Facebook ad and then send an email to the Sales team? ​ Yeah can create a Zap for that too. You want to get a Slack alert when someone signs up to your newsletter? ​ There’s a Zap for that and you better believe I do it! Love seeing the newsletter grow! Another way to think about it is a way to do automation without coding. Anyone with developer knowledge can create scripts/code snippets that do these things too - but takes time/effort/attention that they usually don’t have. Because people with developer knowledge are usually working on harder to solve problems within a business. Zapier has released a newer tool called Zapier Central . ​ Free to test and play with This is a new experimental tool that enables you to create a bot with: • Behaviours - what to do when a behaviour is triggered, how to process or summarise data, and when to execute a particular task • Instant action - tasks you’ve asked the bot to do • Data sources - sources of information the bot can analyse, summarise, and answer questions on A long held idea on Marshy’s idea shelf is to start using AI to make the newsletter creation process easier for myself. In theory - I could train an AI to read all my previous newsletters, adopt that style, and then grab news bytes from RSS feeds, newsletter subscriptions, and my saved articles, and produce a first draft of the newsletter pronto. The reality today is I do all my sifting, reading, and research manually (and enjoy that process) but wanted to investigate how easy/hard the above would be to execute. The answer: not very . The screen opens without much steering, but I was able to work out some things: Figure it out! I can then message the research bot behaviours, actions, and data sources I want it to do: Okay now we’re cooking. I was then able to create a behaviour that fires when emails appear in my Gmail newsletter folder - scan them for AI news, and send me a Slack alert. I can also test the behaviour to make sure it’s doing what I needed. ​ It looks similar to the GPT action builder, but is easier to use. And it sort of does it! Not bad. I could ask it to tidy up the formatting and improve its citations, but it seems to be able to skim things and summarise them for me to read in Slack (if I so choose). I can see this being really useful on things that are quite dense. For example - • Set up a Google scholar alert for ” biodiversity ” • Send alerts to inbox • Scan alerts for any mentions of ” microplastics ” • Move paper link to research file The use case I’ve been fantasising about is moving all my previous written newsletter articles into Airtable, pushing the articles into Webflow with canonical URLs ( like a blog), and then regularly scanning the articles for updated information about the content on the Internet. Again - this is possible - and would improve my SEO. But I just get a bit funny about over-teching my writing process. ​ Consultants are an early winner in the AI race Via New York Times . Throughout my career the BCGs, McKinseys, and KPMGs of the world have moved in and out of my periphery. As someone who strategises, reads voraciously, and loves navigating complexity - I see the appeal of a consulting house. Originally appeared in newsletter : AI versus Marshy #54 - Zap Central, why do consultants win, and troubles with Carrd