Introduction

I did want to just remind you that that was what was happening at Salesforce.

And there's a little, over here on the left, there's the CTO of HubSpot.

He says, where we used to think that there's an app for that, we're going to soon be talking about there's an agent for that.

So there's, you know, apps and agents are starting to become very synonymous.

So that's the new form.

Okay, cool.

Emerging Trends in AI Agents

Mapping the Landscape

Wanted to also share with you some super, let me see if I can make this bigger, super interesting couple of maps.

So these, let me see.

Let me make sure that, yeah, I'll push that off to the left.

Okay, so you've got vertical agents, you've got horizontal agents, you've got companies springing up all over the place.

Who has heard of Zapier?

They're down here, they're an AI assistant

group.

We've got to get chat for us up here on one of these VC market maps soon.

Lindy, someone that is building an automated kind of assistant calendaring booking system.

So you can go to Lindy and check that out.

You've got the various browser agents over here.

I encourage you to go find these if you're interested and to start to look at the various companies that are building in this space because it really is exploding.

And they do talk about the fact that there's different kinds, decisioning agents and general agents.

Market Insights

Again, another view of this that I saw this morning came out of another venture group and they're

They're showing how there's an existing player in the space and how there are other AI companies that are coming into virtually every industry and every software space there is, okay?

So I don't know if anyone wanted to take a picture of that, but you can look it up or talk to me after.

So we've talked about our solo agents and one, okay.

I may have lost my microphone here.

Dead battery.

Okay.

I'll talk loud until we get the battery up and running.

Oh, there's another one here.

Perfect.

Okay.

Is that good?

All right.

New AI Tools and Applications

Notebook LM by Google

So show of hands, because we're going to do this very quick.

Who has seen Notebook LM?

A few of you have, so small number.

Pretty interesting.

Notebooklm.google.com is where you find this.

little app from Google, and it's really taking the world by storm.

And if you're on xTwitter or any of these other places, this is all anyone wanted to talk about the last couple of days.

So every one of these, basically you can create a notebook,

I'll do the quick thing here.

1So I can come in here and choose from a variety of sources and they keep adding to them.

Obviously Google provides Google Drive and YouTube and other things and you can cut and paste, you can copy text in here.

What's super interesting about this, and I'm going to go back to my notebooks here, is that I'm going to take a look at this one, for example.

So situational awareness was a long essay done by one of the previous or ex-OpenAI employees talking about how we're going to get super intelligence in the next few years, five years-ish, and the impact it's going to have on our

really on our

economies and on everything we do.

So when I uploaded that PDF to my notebook, and I can upload, as you noticed, I'm just gonna choose some notebooks that are already open for the purposes of this demo.

I can put in all kinds of different sources, put them all together, and then we get this, we get this ability to chat.

So I can say, you know, what are the, you know, what are ooms?

So what are OOMs?

This is something that is talked about in the situational awareness PDF.

Look it up online.

It's done by a fellow named Dash and Brenner.

Booms are orders of magnitude, and we're going to be flying through orders of magnitude in terms of the capabilities of these large language models over the next few years.

So basically, I can now talk to the PDF, and I can query the PDF kind of on the fly, right?

And I have a whole bunch of other content in here.

Super interesting, however, is...

Experimental Features

This part, so Google calls this experimental, and this is the kind of fun, practical side of things.

This is, call it an agent, the background, because you've given it a task.

In this particular case, the task is to create a podcast of two people talking about the content that you've had here.

Okay, so I'm going to press play.

Welcome back.

Today, we're diving into AGI.

Artificial General Intelligence.

Yeah, exactly.

The kind of AI that could, you know, potentially think like us.

Not just like book our flights or write our emails for us.

Exactly.

We're talking serious brain power here.

And to help unpack all of this, we're joined by... Oh, it's great to be back.

So we're diving into Leopold Aschenbrenner's situational awareness...

The decade ahead.

And he gets into some pretty mind-blowing predictions about where AI is going.

How soon.

How soon, yeah.

Which is like the biggest question, right?

That's what I want to know.

I mean, we've all... So...

1With the click of a button, with the click of a button, this new experimental feature takes whatever content you give it, and this was a very long PDF, converts it into a podcast, which in this case is 17 minutes.

I gave it the manuscript of my entire book and then asked it to just, you know, I pressed the button and, you know, it's here somewhere.

Just press the button and there it was.

There it was.

Here we go.

This is, you go to notebooklm.google.com, sign in with your Google credentials, your Gmail credentials, your Google credentials, and you can play with this right now.

Can you download the audio?

Oh, yeah, you can download the audio, too.

I did that.

So I downloaded the audio, put it into another little thing, added some custom kind of graphics.

Dead battery.

Okay, I can talk loud.

So you can download it there.

Okay, so buckle up, because today we're diving headfirst into the future.

Get this.

It's a future powered by AI.

You know, it's funny you should say that because it seems like AI is everywhere these days.

It really is.

But what we're talking about today with this book, Agentware, by Alan Wunsch, is a little different.

It's not just about the technology itself.

I did nothing.

I just gave it the PDF.

And the impact it's going to have on all of us.

I pressed a button.

It's kind of like he's pulling back the curtain on this whole AI thing.

Exactly.

Okay.

So...

Advanced Voice Interaction

Sometimes we do a little more practical stuff, like build an app on the fly.

I talked to Josh about this yesterday, and he said, this is blowing up all over the internet.

And Google's product managers are now considering bringing agents into this, creating new forms of

marketing based on what you can populate out of here, or they're also considering multi-voices.

So you'll be able to pick from a hundred different voices and have two of them talking.

You know, it kind of worlds your oyster.

So they've got a whole bunch of ideas that they're putting out in the public too.

And so people are offering their ideas as to what Google's going to do with this new little experimental feature.

Yes, a question.

They do not have an API for this yet that I know of.

But by the way, has anyone tried the advanced voice on ChatGPT?

Yes, yes, yes, a handful of you.

What it is is basically something I wish

I wish I didn't have a dead microphone because really in my pocket now I've got ChatGPT with advanced voice mode and it's so seamless that I can speak to something and I go on long walks these days and I press the button and I start talking to it and I say, okay, I want to brainstorm my presentation.

I want to brainstorm an idea with you.

and I got my headphones on, I'm going for a long walk and 45 minutes later, I've had this conversation with the AI that's literally seamless and I can interrupt.

didn't used to be able to interrupt the voice engine, and this is the advanced voice mode that they've launched last week.

Today, at the Dev Day at OpenAI, literally today, this afternoon, they announced that it's now available real-time API.

People are going to be building voice apps like crazy using this real-time API.

That's part of the advanced thing.

So when you asked about the APIs, they don't have one here, but OpenAI just upgraded their APIs with the real-time voice.

So you're going to have all kinds of customer service agents are now going to be possible.

Anything that you can imagine with voice.

Economic and Business Implications

Market Commoditization

There's a question back there or a comment.

The agentic flow seems interesting.

My main question is, there's multiple different companies working on this, Google, of course, HubSpot.

There's probably already also open source solutions.

Tons of them, like Crew AI I've played with and have shown here a few months ago.

Yeah, so on that point, from an economic business standpoint, it seems like there's going to be some kind of price commoditized around this.

Totally, yeah.

So what are your thoughts on where the value proposition is going to be in 5, maybe 6?

I mean, that's a long time in this space right now, but maybe four or five years down the road.

Challenges and Opportunities

I'll tell you what, I'm going to answer it very quickly, then bring Jay up, because I want to make sure that he's got his time.

The answer is, it is really hard right now to think of a startup, new venture, and such, and to have a moat around it.

Because there are many, many, many, as you can see, companies springing up, but there are open source frameworks for these startups.

for these AI agents.

And yes, I'm pretty sure that the agent.ai one for HubSpot that Dharmesh built was using Krew AI for a while.

Then he used LangGraph as well.

So there's LangGraph and LangChain that have been out there for those of you that are following kind of the bouncing ball.

So these are open source frameworks that you could start using right away.

If it answers your question sort of very briefly, it's incredibly difficult, even for the big players, to think about what kind of moat they're going to have for their services.

So when you go back and if you do, you find it on YouTube or you can find it on my LinkedIn, the Salesforce CEO's

pitch was super hypey like he says this is the future you know he was like all of our salesforce customers we're going to bring them on he's got a moat though because they're already salesforce customers they already have their data behind their firewalls in the salesforce platform right now just expose it to the these new agents that's a moat that they can still work with

But otherwise, you know, it is challenging.

Hopefully that answers your question very quickly.

But I want to bring Jay up, and then I will be around for some networking, all kinds of questions.

So we'll come up at the end.