Newsletter

Hello and welcome to another edition of AI versus Marshy!

Another “hullo” to our new subscribers, this is a newsletter about AI, written by a perennial optimist, whose tech and marketing background tells him not to be swept away by hype just yet.

But there ARE a hell of a lot of interesting things to look at and this week is no different.

This week covers:

• copy.ai upgrades its platform with Anthropic - with some side-by-side testing and a dash of nope
• A new, newsletter shoutout to Ai5
• Good, bad, and ugly examples of chat built into an app

I have more things to write about this week than time, so let’s make like a clock and tick on with it ⏰

-Marshy

copy.ai powered by Anthropic - side-by-side testing + some nope

Link to copy.ai announcement.

I was onto copy.ai pretty early.

I met Chris in a marketing community where he was pimping their capabilities.

I’ve followed its trajectory with interest and it now positions itself as a GTM AI platform.

This is very different performance to the “other” big AI copywriting tool - Jasper, who replaced their CEO last year after poor performance.

I hadn’t played with copy.ai for a while, and was due since this announcement.

Loading it up the homepage is a lot cleaner and intuitive.

I decided to give it my brand voice by posting 50-500 words.

I chose a LinkedIn post about my ADHD mishap and another post about my big fast value calls.

I think it did it pretty good job at nailing “my style”:

I use ChatGPT as my default, so I wanted to see what it said with the same posts as training.

The results are here.

It’s accurate too, if a bit more bullet-pointy.

I also HATE the example post it’s done in my style.

But that could also be my pedantry as a writer coming out.

Next challenge, write my newsletter for me!

This is what a newsletter draft looks like:

Nothing sexy right?

Let’s see how copy.ai does (I’m using free version):

And ChatGPT isn't any different/better.

​

To be fair my prompts aren’t much but I think they both suck even for a first draft.

Professional curiosity got me to jump into Gemini to test things as well, and it’s pretty similar - but with more exclamation marks (12!):

Which just reminds of the episode of Seinfeld with Elaine and exclamation points (sorry younger readers!)

I think the meta-point here is that comparisons of platforms for everyday users is a moot point now.

By and large - the models and platforms do the same thing for the broad majority of tasks.

Unless you’re cutting edge with your AI use and applications - you DON’T need to be comparing one model over another and trying to get the BEST one.

A red flag when someone is trying to talk to me about “their AI expertise” is an assertive recommendation of one chat tool over another - please - you don’t know what you’re talking about.

An example of an edge case I’ve heard recently?

I met Firuz through an intro.

His team has built an integration with Vertex into his Google Drive.

With it he can query/reference/use any file stored within an extremely large context window.

Meaning it can reference, infer from, and connect to wildly disparate pieces of information to create new information.

This required engineering knowledge to build - but that’s a far better example of using tools for impact than side-by-sides of the publicly available platforms.

Shoutout to Ai5

Check out Ai5 here.

I met Matt of Rare Connections recently as we’re both using ConvertKit.

He’s also Australia-based and writing about AI.

Matt writes about AI in a really easy to understand way.

Recent editions looked at prompt techniques and a Midjourney photorealistic tokens cheat sheet.

What I like about Matt’s newsletter is that it’s short, easy to read, and actionable.

You can subscribe via this link.

Chat in apps - yes, no, or ohhh gosh?

I didn’t have room last week to go all the way on this topic and want to share what I see on LinkedIn Premium with its AI feature.

Gross right? 🤮

I haven’t come across a single example where the AI conversation prompts have added ANYTHING to my use of LinkedIn.

Yet some genius added it as a “feature” and is being paid to do this.

With actual money.

Are any of you using the AI functionality within Instagram/FB/WhatsApp?

Let me know!

A good example of this kind of prompting is with Zapier.

You can tell it you want to connect one software to another and it will draft up a Zap for you.

I’ve found it to be a bit buggy still, but is a good way to get started with automation if you’re stuck and wondering.

​Airtable have recently had a go at this too:

They call it cobuilding and the jury is out on this one.

I like the idea of building apps with Airtable, but in execution (like recently with TheLeadMagnet.biz) there’s a bit missing.

For example - spinning up a form to send to prospects should be easy (and it is) - but finding where that form is saved is about 3 layers deep right now! 🤣

–

That’s it for this week!

You know how I said I was travelling to a rainy location for my football team?

Well we won!

So that was very fun.

I’ve been moving more and more into coaching with work and will be launching a go-to-market training and coaching package soon.

It’s targeted at early B2B startup founders who have early traction/funding and are exploring what marketing can do.

There’s a gap in the market between being at this stage and going all the way to hiring a supplier/vendor/senior talent.

I want to give founders everything I know so they can up skill rapidly in this area, to make more informed decisions about tech stack, hiring, and to avoid burning through runway.

I don’t know how many people on this list fit that description but feel free to reach out if that piques your interest.

Or check out this recent snippet of a workshop I did recently on the stages of marketing for founders.

​

Thanks!
-Marshy

p.s. I just want to give a bit shout-out to uluu - it’s a deep-tech startup that’s building biodegradable plastic alternatives with hardcore science. Worth following!

​