Speak Easy

Andy Abramson
===

[00:00:00]

Tyler Sellhorn: Hello everyone. My name is Tyler Sellhorn, and welcome to another episode of Speak Easy, the podcast where we discuss communication, productivity systems, capturing our thoughts and how we share those thoughts with our future selves and others. We believe that managing our brains is core to 21st century work, and we're here to learn how to do it better than we did yesterday.

Thanks so much for listening. Speak Easy is brought to you by cleft with our cross platform applications. You can simply speak your mind and cleft will collect your thoughts. Cleft enables you to process verbally wherever you are and send your best bits later today. We are blessed to be learning out loud with Andy Abramson.

Andy is a marketing communications expert and virtual agency leader who has driven over 60 company exits, generating billions for investors. He's held CMO roles at several tech companies, including Phiv and Sky Switch, and Andy has advised [00:01:00] Giants like at and t and Nokia. The engagement with Nokia was the first time a Fortune 100 company put their name on a social media program, initiating the first influencer program in existence.

Andy was a pioneer in remote work and he is also a wine enthusiast, sports marketing veteran and active in philanthropic endeavors. He was recently honored for his first of its kind, innovative fan development in pro sports with the Philadelphia Flyers. Andy, thanks so much for allowing us to capture your thoughts.

In fact, you've been doing some thinking out loud, learning out loud about this idea of thought processing. Obviously that's the business we're in, but if you can just listen back to the intro read I had. I'm curious, Andy, what do you mean when you say thought processing?

Andy Abramson: Well, going back to the dark days, and we both have a little gray hair on our head,

Tyler Sellhorn: hashtag gray temples.

Here we are.

Andy Abramson: There was something called word processing and word processing were these big giant machines. Usually a, what should I say, like a mini [00:02:00] computer. You know, you might have a Lang or other brands, all of that. Yeah. And these big, cumbersome computers did one thing. They processed words. You know, somebody would type 'em in.

You'd have a secretary sitting there with a headset. She was listening to a dictation tape that came out of what was known as a Dictaphone, and they processed the words and they cleaned up the executive's language. And the very good executive assistants made their bosses actually look intelligent on paper because they couldn't necessarily write, but they could talk.

They might, a lawyer might give a very long-winded description and the paralegal would clean it up, put it into legal parlance that a judge would appreciate the brief. I was thinking all about that, how word processing went from this big beast of a machine to our desktop laptops and eventually mobile devices like a tablet, iPad, or an iPhone or an Android.

And we can do all that in the palm of our hand today, [00:03:00] but. With all these new applications like cleft, we're doing a lot more. We're actually processing the thinking. We're not just processing the words because you can go back and say, well, that's not exactly what I was thinking when I said it, and you can regenerate the output.

Like one of the great features that Cleft has added that I had suggested to Johnny, by the way, Johnny Cosgrove, shout out to you. You are my hero. You get things pre

Tyler Sellhorn: previous guest a co-founder at Cleft. Thank you for saying our name. Andy, we really appreciate you being a customer and a user and like letting our ideas even influence in your own thoughts.

It's so cool to hear you thinking that way.

Andy Abramson: Well, this is not, if it's free, it's me if I've gotta pay. It's no way. Which is one of the things that Ken Rakowski and I used to say on the World Tech Roundup when we had 300,000 daily listeners from 1998 to 2011. But the reality is that. I can go back in and I can add something to it, which I suggested, and boom, it's there the next day.

Hats [00:04:00] off to the dev team as well. Forget Johnny being the co-founder. Yeah. There's somebody in there actually doing the bits and small coding to make it work.

Tyler Sellhorn: We are co-creating our products with customers just like you, Andy. Right?

Andy Abramson: Oh, it's fabulous. You guys do it better than anybody I've seen, but the idea behind thought processing is taking our ideas.

Bringing them to life and being able to look at them many different ways. And that's the power of AI and what the artificial intelligence brings to it. But we're processing it. Think of processed foods where you're taking the tomatoes, you're taking the vegetables, you're putting 'em all together and now you've got some kind of juice, a, B eight, so to speak, that's processing.

It's bringing things together. It's making them better than the original. And I've really dived into. The whole concept of thought processing, not just cleft, but other applications like Heady ai, granola, a new one I found yesterday. Speller [00:05:00] AI with an A, not an E, and have been having been using this stuff since the earliest days of grain.

And then of course, Dialpad, which I am a shareholder in. So I always like to be transparent in that and have consulted to the Dialpad team multiple occasions. We've been using AI in our voice, calling to take and process and give summaries, but then to give actionable insights around sales calls, tech support calls, et cetera.

This stuff is not new to me. I've been dealing with that for seven years with Dialpad. I've been using Grain for three. I've been playing with Time OS for two and a half. These are. Incredible applications. And people say, Andy, why are you running all these ais on your calls on Dialpad meetings, or Zoom or Google Meet?

Because they all do things slightly differently. And what's great about 'EM is you can add your own little prompt and make 'em deliver something that's absolutely fantastic [00:06:00] because it gives you what you want.

Tyler Sellhorn: I, it's so cool to hear you talking about all of those different role. We're in support of the executive about processing their words.

And we've smooshed that down into the device. And, allowed for things that are even better than just dictation, right? To, have it turn into something that is, able to be read by the judge. Using the paralegal's, knowledge and kinda having a specific prompt for the paralegal to say, okay, take this idea, this business logic, these be business ideas.

Turn that into something that we can actually take and use in another context that is outside of this office or outside of the, actual written words, but in this way that, you're talking about features of cleft, where we're talking about, actually, I meant this, not that, and now I'm gonna reset.

What's gonna happen, with the LLM [00:07:00] reprocessing things. So just again, get behind the curtain a little bit. Cleft is obviously AI enabled, right? And is definitely something that is used in that way. But I'm curious to learn you, Andy, like when do you find yourself using our quote, you've called it the thought processor.

I'm curious, what is it that you are looking for? When are you doing thought processing? What are the prompts for you to even start, using tools like cleft and the other ones that you mentioned. Like what is it that you are looking for those thought processors to do for you?

Andy Abramson: Save time, work smarter, not harder. So, AI's been around in my life from the day I was born. It's called Andy Intelligence and we process. As human beings we have the ability to process, and what I like to do with tools like cleft and others is take what's up here in my brain and almost create a neural network within the [00:08:00] application so that it's thinking, writing, presenting information my way, and that's the key to this.

It's not about a generalization. Of a platform because then 9 million people using cleft will get the exact same output with the exact same topic all the time. 'cause the way the AI is trained and one of the challenges for any company, whether it's cleft or Dialpad or granola or read AI or Time OS or grain or sono or heady or the list goes on and on.

Even something as simple as talk notes, which is great, is. The AI based upon an LLM is wrong. It should be based on an SLM small language model, not a large language model. So for example, a Dialpad, which is a telephony company, voiceover IP meetings. You remember Uber Conference, that's Dialpad meetings.

Now. Their LM is really an [00:09:00] SLM so much that I jokingly wanted to call it the Dialpad, lm, the DLM. That learned from the interactions of their customers all about business communication. So now it's a very finite subset. It's not boiling the ocean. We're not trying to reinvent the wheel, and most of these applications are nothing more than chat GPT with an interface where something like Dialpad is not.

It's got its own. LM that it relies on. It never goes out to the so-called internet. So your information is never leaking. Unlike other companies that run their AI on open AI or Anthropic or Mistral or Llama or, or, or, which means you are training somebody else's asset and you're also leaking your information.

To [00:10:00] this third party, which may inadvertently give a strategy or an outcome to your competitor.

Tyler Sellhorn: I think that's so interesting to hear you zooming into Andy, right? And your life as a intelligent thought processor. And then zooming all the way out to like corporate espionage. obviously it's a little different when we're giving permission to a provider to be able to train on our own data.

But I think that's interesting to say. Okay. Actually let's zoom back in. To Andy to like our specific customer communications, right? To say, okay, we want to play in this space. We want to have this information. We want to limit what we're querying when we're thinking and talking about this information.

We wanna be working inside of the context. Of our company and not other companies. And we don't wanna share that with other companies either, right? Like we, we wanna use the business intelligence that has been [00:11:00] gathered. By our agents. By our representatives, and use that for our benefit. You were gonna say something?

Andy Abramson: Well, loose lips sink ships is a saying it's been around for years. Yeah. Fumbling fingers. Foster Fiascos.

Tyler Sellhorn: Yeah. Yeah we don't want that information to be elsewhere. We want it to be within our own entities. And I think that's something that when we think about cleft, right?

Just for to be transparent about how cleft works, like you mentioned it, but I wanna say it explicitly here that we're not training a special cleft AI on, on your data. We are using what's here, what's on our your thoughts, your speaking is what is being served back to you when you get a summary of your thoughts, right? And that's your,

Andy Abramson: that's where Tyler, that's where crafting your prompt gets you even more personalization.

For years the internet has been about personalization, not mass media, not right. It, we can get in a whole discussion as to why it became a mass medium, but [00:12:00] that was because none of the direct marketers wanted to touch it. Whereas the mass market advertisers said, gee, we can get better at people.

But go bringing this back to the thought processing arena. Your data is your data. You need to know what you can do with your data. The GDPR in Europe and obviously Johnny and Team started this in Ireland. There's a. Importance around privacy, whether you're dealing with the dig the Digital Copyright Act here in the us, whether you're dealing with or Digital Markets Act, or Digital Services Act in the uk, whether you're dealing with the California Consumer Privacy Act from you, but here in the US PCI compliance for financial transactions like credit cards, all of this stems from the idea that your information should be starting from the position of being private.

I'll go the other way. Privacy is out the window. We've already given up. As soon as we logged on with our very first username, we gave up privacy. We might think we didn't, but we [00:13:00] did. And then of course we have people who constantly post on social media where they are, what they ate, who they were with.

When they go, when they do, some people have their whole check-ins oh, I just checked into O'Hare Airport. Oh, I just checked into Vancouver Airport. I just checked into Philadelphia Airport. Who cares? Maybe in the old days of Foursquare, remember Foursquare, you actually, it was important because you were scoring points like where you're going for coffee.

Why go to Starbucks for coffee? You know, I like Starbucks, especially under the new CEO. I think he's doing the right moves. I also like little boutiquey coffee shops, but I don't have to tell everybody. That I'm going to drink coffee at Starbucks when I go. On the other hand, if I'm at Willie's wine bar in Paris and I'm seeing my good friend Mark Williamson, who I've known for over 25 years, who was a groomsman in my wedding in 2007 and helped organize a wine tour, 'cause I was going way overboard.

And my former wife may, she, you know, I love her to death, still said, mark, take charges. Andy's gonna have us going to six [00:14:00] wineries in one day. Yeah, why not? We're on the bus. But anyway, thought processing today, had I had it back then, would've said, Andy, slow it down. Don't have a six stop tour. You got people who've never drunk wine.

'cause I would've said, we have a small group of friends on a bus, some who love wine, some who appreciate wine, some who are nerdy about wine, some who make wine. It would've said, here's your suggested itinerary. Well, we use Mark for that. So that was a, my ex-wife and Mark did some co-creation thought processing together.

Fast forward to today, I could have put all that information into some AI tool and then put it into something like cleft or heady or granola. Got,

Tyler Sellhorn: or even when you had the conversation with your spouse, with your friend. Right. It would've been a better conversation because you've already, allowed, we were talking about an executive assistant, we were talking about, you know, [00:15:00] people taking dictation and like having these ways to get your thought into a more, more easily understood and better engaged with, right. I know like how I set up my cleft, right? I use too many adverbs and too many complex clauses in my speaking and in my. Writing, and so I am just the bullets, simple sentences, right? Avoid, adverbs and complex clauses. That's really, really helpful for me to be able to give something that is easily read and understood by collaborators.

I think, just to bridge the gap back to Johnny and Justin who created cleft in the first place. Part of what they were wanting to create was something that is easily used by the neurodiverse, right? Johnny and Justin are ADHD, Johnny is autistic and cleft really helps them to produce writing and notes that are more easily understood by normative brain [00:16:00] function.

And there's this idea of executive function. That is a part of that is like lacking for some neurodiverse individuals. And that's one of the things that's really awesome to use, cleft to do is, okay, I'm gonna keep rambling and I might have more thoughts to say, but it's gonna smush it down into something that's easily, you know, saved and used for myself later or for somebody else.

Andy Abramson: I don't wanna get into the spectrum 'cause that was the arena I grew up working in, but the. Idea of neurodiversity and people who are slightly autistic and on the spectrum these tools help them become far more functional because they're already highly functional. They're functioning on a slightly different wavelength.

It's the difference between AM and FM and short wave. These folks are in short wave, which is a very unique subset of the radio magnetic spectrum. Well, to put this into brainwaves, their brainwave. Functions differently. It doesn't mean they're any less human in you or I, it just means that they process information, hence thought processing.

But [00:17:00] I'll give you a great example of the power of something like cleft in a meeting yesterday introductory meeting. I turned on cleft through my cell phone, my iPhone on the table, and transcribed it Now. That's the other big thing. This is transcription, not recording. Which transcription is fully legal.

You don't have to tell anybody you're transcribing ethically, maybe. I don't know. I don't want to get into the yes or no on that one because there's no difference on, in my mind of that phone transcribing versus me taking out a patent pencil. Talk about neurodiverse. I've been getting failures in handwriting since I was in second grade.

I've been typing since I'm 14. I have a hard time signing a check. Having it look, having my signature look like the signature card from 25 years ago or 35 years ago. Sorry, it doesn't work. You know what I said to the banker? You know what? All that money you cash to those checks that don't match up the signature card, can I get it back?

Anyway the reality is, here's how [00:18:00] cleft was put into action meeting at 10 30 in the morning till 1130 yesterday, two sessions. First session was pretty much of the half was, hi, this is who we are. This is who you are. It was, you know what measuring contest. Then the second half hour and thank god cleft does 30 minutes, 'cause the meeting was exactly 30 minutes in the second half and I ended it at 29 50.

I took the cleft output and then put it into a little routine I built called making sense of too much information. It structures the information the way I need it. I then took that output, threw it into a gamma, which is a online presentation making tool. That's fantastic. Talking about thought processing at a presentation.

Cleaned it up in about five minutes, and we used that in our two 30 meeting because I said to my colleague and associate in this other business venture, we're doing that. We need to tighten up how we [00:19:00] present and here it is now. In the old days, that presentation would've had to go to a presentation development team copywriter, go through three rounds of review, and it would've taken about a week literally to have a team come up with the words, the structure in which, by the way, between cleft and my little routine on called making sense of too much information, it was done in seconds, not weeks.

It was done in less than 10 minutes on my end, between the time I got back to my office at about 1230, and by the time we did our next meeting at two 30. Now, I want you to think about what that means to the people who are listening and watching what we're saying today. The ability to rapidly and quickly innovate, adapt, and communicate has changed dramatically.

If you know how to use the tools, and it doesn't matter [00:20:00] which one you use, and this is what gets into personal choice, whether you like a Glock or a sig sour, if you're into guns, whether you like an Irish whiskey or a scotch whiskey or a bourbon, if you're into whiskeys the end of the day, they're whiskeys.

Oh no, they're STA words. Oh, there's a scotch, my friend, Irish whiskey. So what I'm saying is each of these tools that are out there. You have to decide which one is best for you. And I would go one step further. You need to know about a few of them because they each approach things differently. And, but, so yesterday I'm sitting there, it's two minutes before the meeting starts and I'm sitting, do I use, just press record and then I throw it up to my and d AI engine that I've also built.

It does a lot of things like cleft, which was just highly personalized. And it's a it's a, an adaptation of something named that a guy named Thomas J. Frank, who's a Notion Wizard had created, but I've adapted it and it uses tools like Pipe Dream and Whisper [00:21:00] AI and other things. Or do I use cleft or do I use Hedy?

Well, Hedy is really designed for me to interact with the conversation while I'm having the conversation. Do I use granola, which gives me a great summarization and they're all on my phone. I mean, I can, do I use Sono? Do I try this new one? And name's Speller? And I was like, you know what? I think cleft is the right one for this situation.

So by doing that. It saved me time. It gave me actionable insights. It gave me proof points to share with my colleague and say, you know what? We really need to handle it this way. And what did that do? Well, it short circuited the next call in such a way positively that we got to the point of the discussion sooner.

So instead of spending. 30, 35 minutes telling everybody who we are, what we are, why we're there. [00:22:00] That happened in the first five, and then all the questions that were raised, we over we preceded, we poisoned the wells. So because you start doing a bunch of these meetings and this is a pitch meeting, you pretty much can guess what people are going to ask.

And by taking the information and processing the thoughts that came up in the meeting, we were able to get down to the nitty gritty a lot faster and do it all in 30 to 40 minutes. I. And now what I'll do is I'll take the information from that second call. I'll throw it back into cleft. I'll get an updated version of it, and then I'll go back to Gamma after I put it into making sense of too much information.

And I'll go to Gamma and I'll clean up my presentation and it'll be tighter and tighter and tighter. What if we done, we're doing revs, we're doing revisions. The same way an artist does illustrations. He comes up with one design. It comes up with a little tighter [00:23:00] design, goes from loose comp to tight tight comp, whether it's an ad, a billboard, a painting.

You think about it. Monet didn't do the water lilies one time. Monet did many of the water lilies. Eventually he got what is known as the water lilies that's worth millions and millions of dollars. But there are other versions of it. From the, that day in Chna there is the JNet painting as well.

These are iterations, cleft lets us iterate.

Tyler Sellhorn: Yeah. I think the, like listening to you describe how you use cleft and where it compresses the time between a thought and a first draft. Right. I think that's been a theme of all of our conversations is, you know, with the people that are using cleft they see that time to a first version or a version that can be used even just inside of a presentation or in a conversation is so much faster.

And [00:24:00] that, that, that version. Is able to be understood by others in a way that's much faster because you already have the context, you already have the ideas, right. But getting it into a version that is able to be, understood and used by others is so crucial. Okay. I wanna invite you to conclude here, Andy, with a.

You, you're talking about all these different tools that you're experimenting with. What do you imagine is the next version of these kinds of thought processing tools? What is the next version that you feel like is just out beyond our reach that we're building towards? I

Andy Abramson: think we're building towards a more automated but more. Mindful and deeply understood creative process of how we assemble interpolate, apply and act upon the thoughts, [00:25:00] words, and ideas that we and others have. And we are heading towards a period of time where our ability to, our ability to emote through words eventually moves into pictures.

So I see the day where not only are we describing things in our thought processing application. We've set a style for imagery, and that imagery then converts to video, and that video then converts to panoramic 3D holographic projected content, and, but it all starts with a word or it all starts with a thought.

And then of course we'll have the neural implant. That means we don't have to even verbalize or type. It'll just read our thoughts and create. But that's, multiple [00:26:00] generations down the road before the Star Trek Holodeck exists, but we are entering an era of where wonderment becomes reality and tools like Cleft and Hei.

And granola and Dialpad and many others that we're using that can process our thoughts, deliver us the results we need. And if we don't like it, we say, sorry, this is not what I want. Do it over again. And that's really where we're at. We are, we have that. We have the power and the power rests with us, not with the ai.

The AI takes our instructions. It'll learn. It'll become as smart or smarter than us. Heck, what's the first rule of hiring people? Always hire somebody smarter than you. Well always go out and find an AI tool that's better than you, and that's why I.

My pleasure, Tyler.

Tyler Sellhorn: [00:27:00] Speak easy is brought to you by cleft. With our cross platform applications, you can simply speak your mind and cleft will collect your thoughts. Cleft enables you to process verbally wherever you are and send your best bits later. Check us out@cleftnotes.com. That's C-L-E-F-T-N-O-T-E-S. Dot [00:28:00] com.