Dig In: How The Heck Do I Use Claude? (And Other Useful AI Tools) with Grace Uvezian, VP of Marketing at Appetronix

Summary

---

My Food Job Rocks! – Episode Summary

Dig In: How The Heck Do I Use Claude? (And Other Useful AI Tools)

Guest: Grace Uvezian, VP of Marketing at Appetronix
Host: Adam Yee
Date: April 30, 2026

---

Episode Overview

This episode is a thorough, hands-on exploration of using AI tools—especially the large language model Claude—for marketing, sales, and productivity in the food industry. Adam Yee, food scientist and host, interviews Grace Uvezian, a seasoned food tech marketer and AI enthusiast, to dig into practical strategies, philosophies, and real-world hacks for employing Claude and other AI automation to supercharge marketing and business workflows.

Grace details her journey from early marketing automation to building complex, custom prompts and agent workflows, offering actionable advice for anyone (not just food scientists!) who wants to harness AI. The episode provides a rare, practical walkthrough of tool selection, prompt engineering, upskilling, and fostering an AI-positive company culture.

---

Key Discussion Points and Insights

1. Grace’s Background in Food Tech & Marketing ([05:14]–[08:12])

• Career trajectory: Bucknell grad → Startup world (City Maps, Spice) → Consulting (Mars Wrigley) → B2B ingredient tech (Perfect Day, Voyage Foods) → VP at Appetronix.
• Current role: Appetronix builds autonomous robotic kitchens for hard-to-serve venues (airports, stadiums, etc.)—already launched in Columbus International Airport.

2. Types of Marketing & Go-to-Market Strategy ([09:03]–[10:40])

• Growth marketing: “What are we putting out there to grow our business or our product?”
• Go-to-market: Pre-launch steps to create awareness—especially nuanced in food tech vs. AI startups.
• Quote ([09:37]):
  “Growth marketing is what are we putting out there to grow our business or our product and go to market is what's actually the steps we're taking before we launch to make sure people know about us.” – Grace

3. Early AI Automation & Zapier Hacks ([13:38]–[16:57])

• First AI/automation: Mastered Zapier for marketing automation (alerting, PR outreach, automatic pitch drafting), generating significant earned media impact.
• Philosophy: Connect different tools to free up time for creative and strategic work.

4. AI Tool Stack: Comparing Zapier, Claude, Perplexity, ChatGPT, Gemini ([16:57]–[23:31])

• Zapier: Great for linking tools/platforms, but “expensive by the zap.”

• Claude: “Zapier on steroids”—flexible, connected deeply with user data, better for marketing and content.

• Perplexity: Superior for deep research and supplier discovery; more reliable citing and holistic answers.

• LLM evaluation: Use same prompts cross-platform to test performance and keep focus on what “feels right.”

  Quote ([18:28]):
  “I started experimenting with ChatGPT years ago...I just didn't like that I kept having to prompt it so many times. That is what initially inspired me to try Claude and as soon as I connected it to my ecosystem of data I was amazed at what it was producing.” – Grace

5. Advanced Prompting and Building AI Agents ([23:31]–[32:52])

• Voice-based prompting: Using Whisper Flow (voice-to-text tool) leads to richer, more informative prompts than typing.
  • “My first tip is don’t type it. Use whisper flow or…talk to your phone.” ([30:09]) – Grace
• Prompt engineering strategy:
  • Start with a high-level brief
  • List all features/requirements by category
  • End with a “memory” layer so AI learns and repeats your preferences.
• Example use cases: Inbox automation, daily TikTok/LinkedIn digests, presentation content generation, file organization.

6. Security, Comfort, and Responsible Data Use ([25:04]–[26:43])

• Best practice: Only grant AI access to selected local folders, never upload sensitive customer data, treat it similar to giving Google or cloud services access.

• Overcoming security hesitancy: Comfort grows with responsible experimentation and control.

  Quote ([25:04]):
  "I am more comfortable actually when Claude is running locally... No matter what though for my job, I never upload customer data to any of this. So I'm just like methodical about that." – Grace

7. Mastering Daily Workflows with AI ([34:17]–[36:58])

• Use of Cowork/co-working features in Claude: Custom MD (Markdown) files act as “pre-prompts” containing goals, instructions, brand guidelines, etc.
• Automated content curation: Claude agents deliver top 10 TikToks or LinkedIn posts daily, including one-line summaries and content ideas.
• Content repurposing and distribution made easy: Custom mini-apps automate posting across platforms.

8. Sharing, Learning, and AI Culture ([37:31]–[42:27])

• Learning by doing: Use Reddit, Substack, and YouTube podcasts; solve daily pain points with AI.

• Building AI skills in organizations:

  • Don’t mandate AI adoption—start with your own workflows first.
  • Share working hacks organically with coworkers.
  • “It’s less scary than rolling out an AI company strategy.”

• Overcoming analysis paralysis: Share more, even if you feel “not expert enough”—it helps everyone accelerate.

  Quote ([39:16]):
  “The best way to adopt AI into your organization is to start with your own workflows… then tell coworkers, ‘oh, I built this cool thing. Do you want it?’” – Grace

9. What AI Can—and Can’t—Do ([42:44]–[45:31])

• AI cannot: Build relationships, do true sales, or read/react to nuanced human behavior.
• AI can: Automate content, data analysis, create presentations, build web apps, connect tools and workflows—even for non-technical users.
• Great analogy ([43:28]):
  “I always describe it … as an inexperienced intern…if you give it the wrong instructions, it’s kind of on your fault.” – Adam
• Prompting is critical: Bad output = usually bad prompt/lack of context.

10. Upskilling Resources and Influencer Recommendations ([48:11]–[49:43])

• Podcasts:
  • Greg Eisenberg’s Startup Ideas Podcast
  • Y Combinator’s AI episodes (on YouTube)
• Content creators:
  • Serena Romanov on TikTok/LinkedIn (AI research, approachable content)
• Grace’s Substack and LinkedIn: Frequent posting of template prompts, workflow breakdowns.

11. Societal Impact and Innovation Trends ([49:43]–[51:36])

• AI & GLP1s: Both are disruptive; innovation in food and health will transform daily routines and diets.
• Sharing is crucial to accelerate industry progress & collective intelligence.

---

Notable Quotes & Memorable Moments

• On tool differences and selection ([18:28]):
  “Claude is really my hub for anything marketing and sales that I do.” —Grace
• On security ([25:04]):
  “I am more comfortable actually when Claude is running locally... No matter what though for my job, I never upload customer data to any of this.” —Grace
• On building prompts ([30:09]):
  “My first tip is don’t type it. Use Whisper Flow or…talk to your phone.” —Grace
• Human-in-the-loop, not robot intern ([43:28]):
  “An inexperienced intern … if you give it the wrong instructions, it’s kind of on your fault.” —Adam
• Promoting a sharing culture ([39:16]):
  “The best way to adopt AI into your organization is to start with your own workflows…then tell coworkers, ‘oh, I built this cool thing. Do you want it?’” —Grace

---

Timestamps for Key Segments

• Grace’s career and intro: [05:14]–[08:12]
• Defining growth marketing/go-to-market: [09:03]–[10:40]
• Appetronix and automation examples: [11:01]–[16:57]
• Comparing AI tools & when to use them: [16:57]–[23:31]
• Prompting strategies (voice, structure, MD files): [23:31]–[34:17]
• Security & AI comfort zone: [25:04]–[26:43]
• Content curation automation examples: [34:17]–[36:58]
• Sharing & training culture: [37:31]–[42:27]
• What AI does well (and not): [42:44]–[45:31]
• AI upskilling resources: [48:11]–[49:43]
• Societal and food innovation trends: [49:43]–[51:36]
• Where to find Grace online: [51:45]

---

Practical Takeaways

• Test multiple LLMs with your real workflows—find what feels natural and fits your voice/needs best.
• Begin prompt engineering with clear objectives and end with examples/data/content for the model to learn your preferences (“memory layer”).
• Voice tools like Whisper Flow can unlock much richer prompts—don’t limit yourself to just typing.
• Security: Only give tools access to work data/folders you’re comfortable with—never upload sensitive info.
• Share your successful hacks freely within your organization and online; this creates a culture of adoption by example.
• AI isn’t magic—it’s a super-helper for tedious or high-volume tasks, not a replacement for human relationships or nuanced creativity.

---

Resources & Where to Find More

• Follow Grace:
  • LinkedIn: Grace Uvezian
  • Substack: "graceu" ([link in show notes])
• Podcasts/Influencers Mentioned:
  • Greg Eisenberg’s Startup Ideas Podcast
  • Y Combinator AI panels (YouTube)
  • Serena Romanov on TikTok
• Tools:
  • Claude
  • Zapier
  • Whisper Flow
  • Perplexity
  • Gamma, Canva (for presentation/visuals)

---

Bottom Line:
This episode offers a rare, inside look at how top food tech marketers are weaving AI like Claude into daily practice. If you want to move from AI curiosity to hands-on mastery—or lead your team to higher productivity—this conversation is full of actionable gems and reassuring perspective. Share your learnings, prompt boldly, and automate what you hate!

---

Transcript

[Grace Uzevian]

Foreign.

[Adam Yee]

Welcome to the My Food Job Rocks podcast where we interview experts in the food industry and learn about the career paths while gleaning some interesting insights that might help you in your career. I'm your host and food scientist Adam Yee and today we're digging into large language models, specifically with a little known program called Claude. And I got my friend Grace Uzevian, VP of Marketing at Appetronics, to teach me a thing or two of about this AI tool. Grace has spent nearly a decade marketing food technologies most people thought were science fiction, selling to everything from early stage startups to Fortune hundreds. She was the second hire at Spice, a robotic restaurant acquired by Sweet Greens and then Wonder. She went on to focus on B2B marketing at Perfect Day and Voyage Foods, two companies reinventing when greens can be. Today she's VP of marketing at Appetronics, which designs autonomous kitchens for venues with captive audiences like hospital stadiums and airports. She also has consulted for brands such as Mars Wrigley and focuses much of her writing and speaking on how to actually use AI for sales and marketing. You can find her work on substack or LinkedIn. Her substack is substack.com@graceu.

[Podcast Host / Interviewer]

So this conversation with Grace was hands

[Adam Yee]

down one of the most useful AI deep dives I've recorded. And she walks me through exactly how she uses Claude. She also talks about mcps, which we'll get into that acronym later.

[Podcast Host / Interviewer]

You may or may not care about

[Adam Yee]

it, and also other cool kind of tools that she uses like Zapier and Whisper Flow. She is able to use technology to really kind of make these very structured ways of getting information. For example, she has designed a prompt that allows her, or maybe should I say agent that allows her to automate her email inbox and daily TikTok and LinkedIn digests. Have them kind of deliver to her and you'll learn a lot about that there. I also really like the discussion. I have a lot of hot takes on AI. For example, in my opinion, if you get a bad response from Claude, I believe that's on you. And I think that's a line a

[Podcast Host / Interviewer]

lot of food scientists have to really

[Adam Yee]

sit with when they dismiss the technology. Is that it just might be you asking the question is not as clear as you thought it was. So I use LLMs on the daily and I'm honestly I'm studying it in college. I go to meetups about large language models in AI. I go to conferences. For example, a month ago I went to one in Durham, North Carolina And I chat with a lot of people using this technology. Grace posts about large language models constantly and the stuff she posts, I was really impressed and I decided to reach out to her and ask a bunch of questions. And I'm sure you've heard of anthropiclaud. If you haven't, then you're probably just out of the AI bubble. Um, but it recently overtook OpenAI in revenue. I use all three of these. I use Claude, I use gemini, I use OpenAI, which is ChatGPT.

[Podcast Host / Interviewer]

And generally all three are kind of

[Adam Yee]

down top dogs it seems like. And everything else is a little bit below. Funny enough, I get like two of them for free and then I think

[Podcast Host / Interviewer]

I just pay for ChatGPT.

[Adam Yee]

But there's definitely some very useful things that I've come to like about each one of these platforms. So I love this technology. In my opinion. It has saved me a ton of time in my business. It has saved me a ton of time kind of studying and it's also giving me a really awesome way of just being more curious. And when I'm curious I can ask questions and get a different perspective. Is it better than Google? Yes. Is it as.

[Podcast Host / Interviewer]

Is it better than an expert?

[Adam Yee]

Probably not. But I would say it's like maybe 60, 80% as good as talking to an expert in your field. You get an opinion instantly and it allows you to kind of just rethink of the problem. Does it get it wrong? All the time. But I think it also gives you an understanding of the check that you need to have as an expert in your field. So in my opinion this makes you more of an expert in your field if you ask it the right questions.

[Podcast Host / Interviewer]

I'm glad to be talking to a

[Adam Yee]

lot of experts on large language models. And by the way, there are a lot of and not enough at the same time.

[Podcast Host / Interviewer]

Surprisingly.

[Adam Yee]

I think the most common thing is that people just don't share what they learn about this technology very often. So you can kind of be kind of self isolating and kind of just not think you're good enough. But in my, and I think I,

[Podcast Host / Interviewer]

I try to encourage people who are

[Adam Yee]

good at large language models and are good at talking about them to talk about them more. And that's why I want to bring Grace on, is to really do that because she does a great job talking about how she finds this technology useful. We hope this discussion encourages you to explore large language models in your work. Everyone is still a novice using this. Enjoy this episode with Grace. I'm sure you'll learn A lot.

[Podcast Host / Interviewer]

Great. Awesome. So, Grace, we've communicated a lot, and we've also been in a ton of different areas at the same time. I think when I recall, you live in Boston, the Bay Area, and so have I. And where are you now?

[Grace Uzevian]

I'm in New York City now.

[Podcast Host / Interviewer]

You're in New York. Oh, man. The top. Top three. You've lived in all the major cities. What is your favorite city? At first, tell me how many places you've lived, and then, like, what. What is kind of your favorite city?

[Grace Uzevian]

I've lived in only four cities, so just during. I went to Bucknell University, so I was in Lewisburg, Pennsylvania, during those years. But then, yeah, Boston. I'm from Manhattan originally, so grew up there and then returned now and then San Francisco.

[Podcast Host / Interviewer]

That's great. Okay. And which one was your favorite? It's a hard question. I know. I get that a lot, too.

[Grace Uzevian]

Yeah. I love New York City because my lifelong friends, family, co workers are here, and that's been really nice. Or like lifelong coworkers. I think I just had some of the best years of my life in San Francisco because I was able to connect so deeply with nature, which, as someone growing up in Manhattan, was very new for me.

[Podcast Host / Interviewer]

That's interesting that you mentioned nature in San Francisco, because you're right. California has tons of nature, but it also has tons of tech.

[Grace Uzevian]

Yeah. I found the. Even though there's so much tech, it's like people were so willing to just go on a walk. Like, I feel like I would meet people and say, let's just go on a walk in the Presidio or in Golden Gate park, which I. I miss that. I wish I could have that in New York City more.

[Adam Yee]

I agree.

[Podcast Host / Interviewer]

I think Presidio. Well, you have. You have the. What. What's the. What's the big park there in New York City?

[Grace Uzevian]

Central Park.

[Podcast Host / Interviewer]

Central. Yeah. I don't know. I do. I do agree, like, the. The little pockets of nature in San Francisco are very underrated, I feel, compared to what they always talk about, which is, like, down near Market street, right? Yeah, yeah. But I enjoyed my time in San Francisco, too, but I'm also surrounded by trees in. In Raleigh, North Carolina. So anyways, Grace, you have a, you know, a prolific. So, Grace, you've. You've had a prolific career in kind of marketing and food. I'd love to get just kind of a real quick rundown of, you know, what did you study, and then kind of your. Your journey until the current company you're in Appetronics.

[Grace Uzevian]

I studied economics and political science at Bucknell. I also. That was where I first got my first exposure to a completely flat startup and I loved it. So my first internship was with a company called City Maps that was acquired by TripAdvisor and I did all marketing and kind of go to market strategy for them from once I kind of entered the actual workforce. I was the second hire at Spice, which was acquired by cgreen and now Wonder. I launched them and then I got the bug to move out to California where I started initially consulting most notably for Mars Ripley, completely on brand strategy and a lot of their paid media efforts. And then I moved to two B2B ingredient companies in between where I focused on account based marketing for them. And now I'm VP marketing at Apetronics, which we're making autonomous kitchen robots. So very similar to my earlier food tech roots.

[Podcast Host / Interviewer]

Yeah, fun fact, I interviewed the founder of Appetronics or the CEO Nippon. Is that how you pronounce his name? I don't want to. That episode scored really well for some reason.

[Grace Uzevian]

Really?

[Podcast Host / Interviewer]

It was super like compared to our other episodes. I don't know if like the, the metrics were weird, but like it was like always the like tenfold most downloaded episode in that specific like niche podcast where it's just me and a friend of mine in tech ref on news. I just found that like really interesting. And the guy who like pitched, pitched Nippon to us, he didn't mention my food job rock sale. He mentioned this other podcast which I found really weird, the less popular one. So you know, you guys are doing some really exciting things in that area and you guys are building out a ton, I believe in Ohio. Is that right?

[Grace Uzevian]

That's correct. Yeah. Our main customer is Donato's Pizza, based in Columbus.

[Podcast Host / Interviewer]

Very cool. I love to understand what kind of marketing you specialize in. Or what do you like, I guess. Or what did you. What did you seem to fall into as you've kind of progressed through your career?

[Grace Uzevian]

I love growth marketing. So whether it's through organic content strategy or paid, and then I've also just fallen into a lot of go to market strategy which they're both so heavily connected.

[Podcast Host / Interviewer]

So a lot of people here are like food scientists and don't really understand marketing. I would say I'd love to have a very simple breakdown of the difference between growth marketing and go to market marketing.

[Grace Uzevian]

Growth marketing is what are we putting out there to grow our business or our product and go to market is what's actually the steps we're Taking before we launch to make sure people know about us.

[Podcast Host / Interviewer]

And you would say like go to market is more like be pre product market fit. And then like growth marketing is like when you find product market fit or am I or does it not matter?

[Grace Uzevian]

It's different. So I feel like with AI startups it's completely different than food tech companies. I think you need your, you can, you need a product market fit before you launch an AI company. Just it's like such a different space. And I think software generally for a food tech company, I think there's enough space where it's like depending on how we launched is like we are partnered with a legacy brand. So we have a lot of the product marketing and menu learnings from them and we're just a new innovative way to go to market and for them to grow. So it depends on who like who you're partnering with and how you're launching for food tech.

[Podcast Host / Interviewer]

Okay. And since you've done well, a lot of your work has been on like food tech, right?

[Grace Uzevian]

Yeah, completely.

[Podcast Host / Interviewer]

Very cool. So let's hear a little bit about Appetronics before you get into like the, the cool stuff you're doing with AI and just want to hear like kind of what the company's about and what do you do for that company.

[Grace Uzevian]

Yeah, we are building fully autonomous robotic kitchens. We're specifically targeting airports, universities, stadiums, so typically non commercial food service operations where it's hard to get recurring labor and where most of the other options close after like 7 or 8pm So I travel a lot for work and like every airport I'm in, everything is closed by 8pm we are an option that gives you fresh made fresh food after hours.

[Podcast Host / Interviewer]

So you're saying like a pizza chain would be open like at midnight in the airport. Is that generally how it works?

[Grace Uzevian]

Or our location would be open 24 7. Another pizza chain would close around 7 or 8 typically.

[Podcast Host / Interviewer]

Hmm. Is that like available right now or, or is it coming soon? What?

[Grace Uzevian]

No, no, it's available. We are already in market in Columbus International Airport. We launched last June.

[Podcast Host / Interviewer]

Could you walk me through? I guess like this is totally new for me. I didn't know you guys like have a pizza chain already up in Ohio. Like how do you, how does that work? Like, like, could you walk me through? I'm out of my flight, it's like midnight, which I've always pretty hungry at then I'd love to understand this.

[Grace Uzevian]

Yeah, so you will. In terminal B, in Columbus Airport, you would walk up to our robotic kitchen. There's a QR code for you to scan. Basically you just order on your phone. We have cheese pizza and pepperoni pizza. And then from there you would watch the robotic arms essentially pick up the dough, add sauce, add cheese, add pepperoni, put it into the oven. Then a different robotic arm would grab it from the oven after it's done cooking, cut it, and then put it into a locker for you to pick up.

[Podcast Host / Interviewer]

Wow. Okay, that's. Yeah. You know, you mentioned lockers, and I've seen it in the Raleigh Durham Airport. There is a locker based, like, system, and I assume that's like kind of an evolution of that, but that's really cool. And I guess you could say, like, there's a lot of coffee. Robot arm companies that have kind of done this in the airport too, right?

[Grace Uzevian]

Yeah, I know. Cafe X is still around in sfo. What I always find. And I have tried their coffee. I thought it tasted great. What I find is there's still a line because it's just one. They're just making one coffee at a time. So, you know, it's not that different in terms of, like, timing. You get to get your coffee where it doesn't save you a ton of time.

[Podcast Host / Interviewer]

Hmm. Yeah, I can see that.

[Adam Yee]

No.

[Podcast Host / Interviewer]

Okay, so next time I go to Columbus Airport, I'll. I'll look for it. Do you know what word is exactly?

[Grace Uzevian]

Terminal B. Terminal B.

[Podcast Host / Interviewer]

Got it. All right. Anyway, so Grace, you've. You've been writing a lot on LinkedIn about essentially using AI tools.

[Adam Yee]

Tools.

[Podcast Host / Interviewer]

And I think that's, that's really cool. I've been exploring them myself. You know, it's. It's hard to understand who's ahead, who's behind. It's hard to really understand how we use these tools. There's skeptics, there's. There's fanatics. I'd love to hear your discovery, specifically AI and, and how. And we're going to go through this journey together about, like, how you have used it and, and how you've kind of like improved and learned more about this over time.

[Grace Uzevian]

Yeah, I'll lump in kind of AI and marketing automation tools where I was actually first exposed to a marketing automation tool, zapier, back in 2017. 2018. And that was I. Once I figured out I could link up different systems and have it do a lot of the busy grunt work for me, I became obsessed with it. So I have had a little bit of an early advantage there.

[Podcast Host / Interviewer]

When Was that? Like three years ago.

[Grace Uzevian]

27 to 2018. Ish.

[Adam Yee]

Wow.

[Podcast Host / Interviewer]

Okay. Okay.

[Grace Uzevian]

My Main kind of like go to market focus. When I was at Spice, because I had a really small marketing budget, was to get us in as much earned media as possible. So pitching every single journalist. And what I found I could do with Zapier was I could basically set up email alerts for anytime a relevant journalist was covering. Food in Boston, robotics in the globe, automation in the globe. And so I set that up and then I had it add to a spreadsheet and then draft actually a pitch for me that I would then personalize based on the journalist style. So I, I generated like almost $5 million of PR value. I had many videos go like mega, mega viral, like 25 million views and over. And a lot of that was with this strategy because I was just able to get so much coverage on us because I was using Zapier. So that was like my first time. And then I kind of kept bringing that to every company I went to. Whereas now I'm just like in deep on all of the AI tools because it really resonates with how my brain works, like connecting the different systems together.

[Podcast Host / Interviewer]

Describe that a little bit more. So Zapier, from what I understand, I don't know a lot about, I've never used it, but it is something that connects tools together. Right. And talks to them, or it connects

[Grace Uzevian]

tools together, talks to them. It basically helps you move data from one place to another in systems that aren't usually connected.

[Podcast Host / Interviewer]

Can you give me like a example that's pretty straightforward.

[Grace Uzevian]

Yeah. For the PR example, I basically connected like web alerts, so an RSS feed with my Gmail account, with Google Sheets and then with a Google Doc that generated the pitch.

[Podcast Host / Interviewer]

Oh, that's interesting. So that is actually almost like an agent, right?

[Grace Uzevian]

Kind of, yeah.

[Podcast Host / Interviewer]

All right. That's definitely something we're going to dive into as well, which is. I think I was, yeah, I was in this conference in Durham. All they could. They cannot shut up about agents. Every startup I talk to that's in Y Combinator is doing something with agents. Agents are everywhere. Well, we'll dive into that for sure. So you, you were able to master Zapier and I guess they, they, they are quite kind of the pioneers on this. How has like this type of system has evolved as you've gone through your company, the companies you've worked for?

[Grace Uzevian]

I mean, once I so to jump through, basically, like once I learned about Claude and the idea of mcps, I found that to be a really interesting way to tie together like lots of different tools for what I was Using all within the Claude code terminal. So that was then, that was basically in my brain like a version of Zapier on steroids because Zapier, it charges you by the zap, so by every single test. So it's really expensive.

[Podcast Host / Interviewer]

Oh, interesting.

[Grace Uzevian]

While I was using the cheaper version N8N but now I find I can do everything I need in Claude.

[Podcast Host / Interviewer]

Let's talk about Claude. Obviously there's a lot of talk about Claude. I think people are still trying to figure out ChatGPT but I would say the majority of things I've. I've been seeing, at least my feed is about Claude, which I think has recently gotten a lot of publicity and a lot of credibility through essentially a lot of certain media events that have happened in the past, you know, super bowl ads to the government saying it sucks or whatever. So. And I think a lot of people probably in this audience don't really know about Claude or probably are curious about Claude. I'd love for you to really talk about Claude and, and why you believe it's better than ChatGPT and I'm sure other people do too or Gemini or, or whatever. Why is this the preferred system or, or LLM that people want to use now?

[Grace Uzevian]

Yeah, I started experimenting with Chat GPT years ago and I just found its writing style was so far from mine even when it. All of the data I could about myself, even when I connected it to my email and to Google Drive and I just didn't like that I kept having to prompt it so many times. That is what initially inspired me to try Claude and as soon as I connected it to my ecosystem of data I was amazed at what it was producing. The one thing that I still do use Perplexity for is for true research. So like research specific pricing or for our, like if I'm working with an R D team, I will put that into Perplexity still. But Claude is really my hub for anything marketing and sales that I do.

[Adam Yee]

I'm.

[Podcast Host / Interviewer]

I've heard of Perplexity I've heard especially in New York. Surprisingly a lot of people use Perplexity which I find interesting. But could you explain to audience why use that specific research? Why is that for what I understand better for research than other elements?

[Grace Uzevian]

Just when I've tested it. So at my last company I tested it to basically say like what are some interesting supplier options for a cocoa free chocolate for like a specific ingredient. The difference between what I found between Clawed and Perplexity, Perplexity was just much better. It gave a more holistic Answer. It gave me more options and some of the sources were just at the time, LLMs were, you know, imagining things a little more. And so I found like two more imaginary links in CLAUDE that I was like, I'm going to keep Perplexity for my research and then Claude for everything else.

[Podcast Host / Interviewer]

Yeah, I think we, we're seeing this kind of change in how we are getting links. I feel with a lot of these LLMs, right. Chat GBT, like when I, you know, when you search anything with Chat GPT, like months ago it was pretty bad. I, I forget how good it is now, but I think they're going to paid ads, so they might be different. I, I love to hear how you generally evaluate these tools because there's a lot of options. I, I, I have all, I have Claude, I have Chat GPT, I have Gemini, all from different, all different ways, but I have all the PRO versions of those.

[Adam Yee]

Not the Pro.

[Podcast Host / Interviewer]

What is it? The plus? I don't freaking know. There's all obviously these payment plans for each of these and obviously people don't have like that optionality compared to like, you know, someone who actually works with this. Right. I'd love for you to understand how you vet these tools and I'll also say how I do them as well. But you know, there's a lot of options. Right.

[Grace Uzevian]

Yeah. And I try not to get too bogged down in every single, like evaluating every single tool. I just love the CLAUDE interface and now I love the TO interface with claude. I find for me it like this is just a lot of fun building these things. So I want to keep it fun for myself. The way I really evaluate is I basically give the same prompt to anything. So like I said, I gave ChatGPT the same prompt as Claude. Claude ended up being my preferred, you know, LLM for like a bunch of different prompts. And then same thing between CLAUDE and Perplexity. I gave it the same prompt and said what actually feels more right for what I would do my own research on. And then I go with that. Yeah, and then the cost, you're right. Like I, I'm on the $100 a month Claude plan. It adds up I think just for like I'm able to get so much from it that I, I'm willing to invest in Claude.

[Podcast Host / Interviewer]

Yeah, yeah, I had the 21 and yeah, I recently hit my limit today for, for one example. Right. Yeah. And I would say I don't use it like I use it more than the average person, but probably not enough like, compared to like A professional. But you know what I've noticed, at least what I do is I don't like asking the same prompt, but I do ask a very similar problem depending on the projects. At least for me as a food scientist who's using it a little bit differently, I would say I found Gemini and Chat GPT relatively the same in terms of its responses, but Cloud has been really interesting. One thing that I'm trying to do with Claude specifically, which I just got, is understanding how to visualize data better. And I feel like Claude does a much better job at aggregating data and also visualizing a lot more. Yeah, the issue with like Gemini, if I give it the same prompt about like get me data on, I don't know, let's say beef prices, it would give me a very, you know, like three point graph and like doesn't look good at all. And it's like you can tell like someone made this up. Whether or not Claude is fabricating data or not, I, I can't tell. It does give you a more legitimate graph with a more legitimate like kind of rationale behind the data get. So I just find that really interesting in terms of like how these, how each of these LLMs are spitting out kind of the responses that you give them.

[Grace Uzevian]

Yeah, I find that too actually. That's it. It's a good point where a lot of the times if I have perplexity research, I'll still put it in Claude for the Visual Dashboard or whatever. I'm making sure there's great, you know, research or data I've gotten from Claude and now I want to make it presentation or website ready. I'm immediately going to Gamma and lovable. So I still have my different tools. Tools. Yeah,

[Podcast Host / Interviewer]

yeah, I've looked a little bit on Gamma, but I still stick to kind of a Canva and PowerPoint. Anyways, what I also noticed with Claude 2 is that at least it's on your desktop, right? Versus like the other tools which are kind of browser. And what I think can be scary for most people is you can tactically have it scan and search through your files. Right. And I think computer scientists love this, but I think people who are not like computer scientists are a little bit uneasy about it. I, I guess what can you tell people about that type of functionality? Right. Because I've been playing around with it. Like what I love it doing now is even though it's like invasive my computer, it organizes all the files that I wanted to organize. Right. Like I say like please organize all of my proposals for my consulting company and it will put them all in a file. I think that's very bare bones. Like that would take me an hour. This would be like five minutes. I'd love to understand kind of. One, do you do that type of stuff too? And two, how do you use it?

[Grace Uzevian]

Yeah, I got a comment on LinkedIn too the other day. Like if I'm concerned about some of the security between giving Claude access to all of my emails, my LinkedIn, like I have like I think 20 to 30 connections within Claude now. I am more comfortable actually when Claude is running locally. So if I, I pick the folder Claude is operating in within my desktop and then that's the only data it has for that specific like co working project or if I'm code. So I feel really safe there. No matter what though for my job, I never upload customer data to any of this. So I'm just like methodical about that. I don't want to compromise our customers data ever. But then for me personally when I set up all these, you know, all the connectors in Claude, I give all this data to Google anyways, so it feels like the same risk to me.

[Podcast Host / Interviewer]

Yeah, I guess. I mean, you know, your emails aren't like super confidential I feel.

[Adam Yee]

Right, like.

[Grace Uzevian]

Right, exactly.

[Podcast Host / Interviewer]

Yeah. And that's actually how I feel about a lot of, I don't want to say like concert of course is confidential. Right. But I do think that, well, food scientists, what I've learned is like are very cagey about anything they say on, on an lm. Right. Even though it's already scanned all your books, it already scanned, has scanned all your, all the recipes on like a cooking website and it just needs to be formatted a certain way. So I think what I'm trying to get out of this is like how do we, how are you more comfortable about essentially the large language models understanding you? Does that make sense?

[Grace Uzevian]

Yeah, just how am I more comfortable with them or using them?

[Podcast Host / Interviewer]

Is, is there something that you just feel that people should know about using these tools?

[Grace Uzevian]

I mean maybe I've gotten much more comfortable over time. Like I attended a McP hackathon like 8 months ago right when MCP had come out or maybe so what, what

[Podcast Host / Interviewer]

is just for the audience. What does MCP mean?

[Grace Uzevian]

Yeah, Moodle Context protocol. So it's kind of a way for any LLM to like know what tools it has available to it without ever changing the code base and without knowing the specifics of every single tool. So you can basically it just, it's like A menu of options for the LLM and then it helps you figure out what else it can do.

[Podcast Host / Interviewer]

Okay, got it.

[Grace Uzevian]

So, yeah, I think at that time I was like way more nervous about giving Claude specific custom MCP access for like, at the time, I think I connected it with like Vercel, 11 labs and maybe cursor, I can remember, but I think just over time I just did it one at a time. I connected it to Google first and then, you know, all the other things. And I just realized it was, even if I had slight security concerns, what I was getting from it was so much more worth it than anything else. So definitely now it's coming up on, you know, at least eight or nine months since that hackathon, but my comfort level has continued to grow exponentially.

[Podcast Host / Interviewer]

Yeah, you've been sharing more on LinkedIn about it. Right? So which I think is also a sign of you being more comfortable about it as well.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

Um, I'd love to understand how you're learning about this though, because I think you posted like this really cool organizer on your LinkedIn, I think yesterday or two days ago, whatever, and I asked you to share the prompt and you put it on your substack and I was looking at it like, it's very comprehensive. Like, you know, I don't, I don't think an average person would. I think it would be overwhelmed. Right. So. So, you know, and obviously that took years and like figuring out how to set up the prompt and all that. But how are you learning about like this type of, like the, the way of doing this?

[Grace Uzevian]

Yeah, I am very, very deep in the Reddit AI LLM, you know, subreddit community, so. And I'm just naturally a big Redditor, so that's definitely my starting point. I really love the Greg Eisenberg kind of like a lot of the content he produces, but I find the best thing is just to like, anytime in my day to day life where I'm like, oh, this is kind of annoying that I have to do this or I don't really want to do this task. I just start playing with Claude and then I figure out a way to make it easier. It's like my email inbox is always like, takes so much time. And I was like, I don't just need it organized, I need the drafts written for me. And so now that's what that's doing for me where I really like to never have a blank page anymore.

[Podcast Host / Interviewer]

Do you have a philosophy or a theory or I guess like a structure of how you Generally prompt, especially advanced level, as you. I think there are a lot of people who say, like, okay, you need the subject first, you need the context first, you need the task. Right. I guess. Do you have any tips or even resources that kind of help people understand how to write a. Essentially a really useful prompt?

[Grace Uzevian]

Yeah. My first tip is don't type it. Use whisper flow or write, like, talk to your phone and give it the prompt. Because I.

[Podcast Host / Interviewer]

What is whisper flow?

[Grace Uzevian]

It's basically like a tool to take your voice and put it into text.

[Podcast Host / Interviewer]

Okay.

[Grace Uzevian]

There's like, you know, the granolas of the world too, like true meeting notes. But whisper flow is. It always gets it exactly right. So more than anything, I'm not actually typing my prompts, I'm always writing them. And it's just like a stream of consciousness. Like for the email one, it was like, I want high priority up first. I want to make sure anything urgent is never missed. I want the checkboxes to be clickable. Like, it was just like a consciousness of what I wanted. And then from there I added the memory layer for drafting my email responses.

[Podcast Host / Interviewer]

So you have to talk to your phone about this. Why do you think that's more valuable than. Than typing it?

[Grace Uzevian]

I find you just. You just give it more information when you're talking about it. Like, it's just sometimes when I'm typing it, I'm like, I'm just looking at the Claude terminal, whereas I'm. When I'm on my phone, I'm saying it, but then I'm looking at my email and saying, oh, yeah, I want to add this. And I'm looking at a spreadsheet. I'm like, oh, yeah, I want to add this so I can actually look at different things while I'm prompting.

[Podcast Host / Interviewer]

Okay. And yeah, I'm getting very granular on this because, yeah, I love it. I don't do this, by the way. So once you get like, you know, the whisper flow, which, you know, transcripts your voice. Yeah. Then you edit it, I assume, right?

[Grace Uzevian]

Yeah, correct. I really edit it. The prompt initially. But then I say, show me the first visualization, like, as soon as possible that's in the first prompt. And then once I see it, I. The edits come much more naturally.

[Podcast Host / Interviewer]

Interesting. And what have you noticed to be kind of like the. The instructional structure. I don't know that's even a real word of how to input this here. I think what I'm trying to get is like, okay, there. There is a structure of how these elements read your feedback, right? Or your prompt. I'd love to understand how. What, what are best practices on doing that?

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

So after you edit it and all

[Grace Uzevian]

that best prompt is, I think basically high level. Give it a brief first. So like a high level of what you want. And then I go into every single feature I want and so, you know, features organized by category. So for the inbox one, it's easy to say, this is all the features I want in finance, this is all the features I want in marketing, communications. These are all the features I want in research. Like every single category I'll go through. And then the ending prompt is always kind of like a memory layer for me for that. So I end on memory. So it knows to kind of search through all of those categories to then draft my responses.

[Podcast Host / Interviewer]

When you say memory, why is that a big issue? Well, why do you need to do it? Why do you need to do it?

[Grace Uzevian]

I don't want Claude just guessing what to write. Like, I've used that email for 15, like 20 years. I know what it needs to say and Claude just needs to search through it to find my pattern. And so. Yeah, exactly.

[Podcast Host / Interviewer]

Got it. And there's also talks if you, if you're using Claude and you're using the cowork feature, which I don't know how many people here would do that, but they should because I found it really useful. But when I do, let's say the co working function, it produces an MD file. Right. I love. And at least in this conference, I've learned a lot. Or people just talk all about this Claude MD file. One, actually this is faithfully. Do you know what that is? And two, like, how do people. How have you seen people use it?

[Grace Uzevian]

How I use the MD file and how I've seen people use it is it really just gives Claude cowork, like all of the context it needs to do the task. Well, whether it's doing the task locally on your computer or in the cloud and using other connectors. At least that's my understanding of it. But it includes like goals and like what you're looking for it to do all in the one file. So it's always reading that initially and it, like that's what its instructions are based off of.

[Podcast Host / Interviewer]

Yeah. At least I see this as like an evolution of the custom GPTs that I think were kind of talked about like a couple months ago, which some people just might not even do that stage. But generally with Chad GPT and Gemini, Gemini's gems, they had an ability that you input something to kind of have it read it first. Right. Like almost like a. I don't know the proper word, but it's like a pre prompt.

[Grace Uzevian]

I feel like it's also somewhat similar with the mcps where it's like I have a lot of my Claude skills enabled. So it's like go to my like brand MD guidelines or I forget how I title it. It's like brand guidelines md. So anytime a specific co working prompt is relevant, I'll have it read there first. Because it's like everything about the company's brand guidelines so it's always delivered on or other things like that. I've used Cowork way more for research. Like I have a lot of research to me. Yeah. So every single morning I get like a top 10 TikToks of the day relevant to same thing with LinkedIn posts.

[Podcast Host / Interviewer]

Okay.

[Grace Uzevian]

And so I find it's like I love when Claude is prompting me to figure out how I apply it to my own content or to the company's content. And so every day I have like email digests coming in from TikTok and LinkedIn.

[Podcast Host / Interviewer]

And yeah, I was actually trying to set that up. I couldn't figure it out. So I might have to ask you after the call. No, no, this is great. Like, right. But I think this is like the important part. Like I don't think a lot of people know how to use this tool and it sounds like you do, I guess like with, let's say that for example the top 10 LinkedIn and the top 10 TikTok. I'm not going to say like say your whole prompt here, but I'd love to understand what you specifically asked it to look for like scrape or, or, or you know, did. Was it specifically like top 10 TikToks for.

[Grace Uzevian]

I ask it to pull the top 10 TikToks with high engagement, high impressions or high views in TikTok and deliver it to me in an email. I say link out to the specific TikTok, include a one sentence transcript of the main idea. And then also I have a third bullet point that it says how give me an idea for how I could apply this to my own content.

[Podcast Host / Interviewer]

And do you find it to be pretty accurate?

[Grace Uzevian]

I find it to be a helpful way for me to be prompted for what I should be writing and talking about. Like, especially now that I've been posting more on LinkedIn, I find it's like having the top 10 posts in food tech and in like deep tech is a really helpful way to also just get to know like the format and how interesting people are talking about what they do.

[Podcast Host / Interviewer]

That's. I need, I need to learn how to do that.

[Grace Uzevian]

Actually, you can share the prompt again,

[Podcast Host / Interviewer]

I don't know how I'm gonna share it to everyone else. But you know, I think that is something like, I think people just don't know about.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

Like, yeah, I think from what I understand, I'm in a lot, being a university, we, we do have like forums where we're trying to figure this out. And I will say, like, a lot of people don't really know the basics of this and I, I think just having people share more about how they're using it is very useful. And there is so much information out there and there's so much. I think people are just scared of it in general. Like, there's so much different. It's like analysis paralysis.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

And this, this kind of ties into the question of like, how do you, how do you design a culture that uses AI like in a company? But how do you get people who are kind of scared of using this to use it? I'd love to understand if you do this internally, you're trying to do it on LinkedIn. I see. But I'd love to hear some best practices.

[Grace Uzevian]

Yeah. I just gave a talk to Cisco about this because.

[Podcast Host / Interviewer]

Oh, you're so. Yeah, I'm excited you're getting there. You're an influencer in AI.

[Grace Uzevian]

The best way to adopt AI into your organization is to start with your own workflows. So I don't think it's ever a good idea to propose a company wide AI organization strategy. I think it's like day to day, what are the annoying low risk, very high volume tasks that we're all doing that we could automate. So for me it's like writing social captions. It's a lot of the email responses. Sometimes it's analyzing specific spreadsheets or like looking at the historic differences in our sales data, things like that, where it's like, I have to do this every week or every day, but I know I can just have AI do it from there. I tell my coworkers, oh, I built this cool thing. Do you want it? Do. I'm going to help you set up the prompt. And so it's, it's low risk like that where it's like, like with the email dashboard, it's the same idea where it's just like, I just wanted this for my own life and I'm happy to share it with everyone. So I don't, I think if you're just having the attitude like this is fun and I'm happy to pass it along if it's valuable. It's a less scary thing than like we are now having a company AI strategy for all social content.

[Podcast Host / Interviewer]

I do think that is how a lot of companies are trying to do this.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

Like they're getting a budget for AI, it looks good on a board deck and they're kind of like kind of forcing people to use it. And so people are trying to find out ways to like adopt it. Right. Or. But sometimes the KPIs don't work out because there are KPIs, even though they may not. Should not be maybe. So it sounds like for you though, it's one, make it yourself. Two, share it with others and be proactive in helping them build it. And which I think is really good advice. Do you have any like, I don't know, set meetings or set ways of like sharing what you build or, or do you. Or have you explored that or have anyone told you about that?

[Grace Uzevian]

No, but honestly, your comment yesterday of can you share the prompt with me? I was, I should be doing this externally too because I've just been doing it in, you know, docs and sharing it with people. But yeah, that's why I started the substack. It was, that was completely your recommendation. I haven't externally as much as I should.

[Podcast Host / Interviewer]

Yeah. Actually the biggest takeaway I got from this conference yesterday at Durham was, was share, like share constantly. And I think I'm probably going to be doing that too. I will say you, you are much more advanced than I am and I thought, I thought I kind of knew what's going on, but geez, the way you built that dashboard, which I'll link it to everyone, it is quite impressive what you built through that. No, I think we don't say that enough for people who are in this craft. Right. Because there is someone who's probably better than you, but there's a lot of people who are worse than you.

[Grace Uzevian]

It's funny.

[Podcast Host / Interviewer]

Yeah. I always have to say that because at least being in the Bay Area, like for the past like year, like everyone was like zooming through this.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

And even like during this conference, like someone who's like really into it had their AI do a presentation like right in front of everyone. Like it was reading apparently off the script. Like, not, not. I don't think it was off the script. What they claimed was like this AI was doing the presentation just through its like, context of his slide. Like, I don't, I don't know really how to explain it but like that's like super advanced. But some people there were just like oh yeah, I had to like you know, scan my website. So I just think like the, the knowledge gap is everywhere and we really don't know who's in action, who's not. Right.

[Grace Uzevian]

Yeah, it's true. I think I, I love the concept of just share more. I also find like a lot of the like I love following lovable on LinkedIn and a lot of the content they produce. Actually a lot of the content Anthropic produces too. Like they have one massive PDF that I've saved forever where it's like how anthropic teams use Claude and it's just like an interesting way to understand other way. Like much larger organizations are using their own tool. I was shocked when I read Anthropic's entire growth marketing department as one person. So I think even them sharing that was a really great way to open my mind into like oh, I could be sharing not just what I'm doing and the result, but also like all the behind thinking.

[Podcast Host / Interviewer]

Very cool, Very cool. So since you're pretty involved in this and you have, I would say a lot of expertise on this, what do you think AI can do versus what it can't do? What have you seen that works really well and versus what it doesn't work quite well at?

[Grace Uzevian]

I mean it cannot own your relationships with your customers and your fans and people you work with. It can't. I don't think it's that great at, you know, true sales. Like it can't read the room, it can't adjust to people, it can't kind of think on the fly, you know, like that is still a human. So I really lean in there. What it can do is almost everything else, like build presentations, content production, data analysis, building web apps. Like I've built a bunch of web apps already and I'm non technical and then kind of like what I was talking about earlier with Zapier, like connecting many different tools into one workflow really holistically.

[Podcast Host / Interviewer]

Yeah. I always describe it to people as like kind of a, an inexperienced intern. So yeah, an intern you can tell to do whatever you want, but if you give it the wrong instructions, it kind of is on your, your fault that you gave it the wrong instructions. Right? Like yeah, you know what I mean? Like I just feel like the intern who wants, who's willing to learn it, who can will do anything right to like you know, get a good, do good other job, will do It. But if you give them the wrong information or you're given the wrong just task, it'll do it. But it would be the wrong task. Right. And I've learned this a lot teaching interns about giving the wrong task. So I just feel like that's how I feel using the tools a lot is if I'm not understanding how to prompt it correctly or asking the right things. And I do think it's a really good intro. Like, to master this, you have to really be introspective of your own work and how you communicate your own work.

[Grace Uzevian]

Yeah. I find I've tried to train a couple of friends who are very like, anti any LLM on how they should be prompting it. And what I find is, like, if you get a really bad response, like the content, let's say it's like a social caption and it just generated a crummy one that is not in your brand's voice and you hate it. That's actually on you because you didn't give it any of the context.

[Podcast Host / Interviewer]

Exactly.

[Grace Uzevian]

And so I. I try to look at myself as like, the manager of the context to get what I need.

[Podcast Host / Interviewer]

Yeah. And I think that's good. Like, communication in general.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

Like, this is almost the same as like emailing an intern saying, like, do this. Like, right. Like, right, buy this bag of sugar. I don't know. Right. And if you don't give them any context about this bag of sugar.

[Grace Uzevian]

Exactly.

[Podcast Host / Interviewer]

They're just gonna buy something in the grocery store.

[Adam Yee]

Right.

[Podcast Host / Interviewer]

Or which you're not supposed to.

[Grace Uzevian]

Yeah. It's completely the same.

[Podcast Host / Interviewer]

Yeah. Yeah. So I think that's something I really want to take away from here is like, if you're scared of this or you found it to be, like, kind of not good. I think it might be there. There is a lot of training and education of how to communicate with these tools. Right. And I think that that becomes really important as you start to use these tools more efficiently.

[Grace Uzevian]

Yeah. Even for some of the presentations I've mentioned that I have it create now. I've given it amazing historic presentations I created on my own. So it's all uploaded in my project folder. So it knows what I like to sound like and how I like it to look and to feel. So it's never starting. I've set it up for success.

[Podcast Host / Interviewer]

I 100% agree. I'm starting to use Claude right now. And I've used this for Gemini and a few like my writing and whatever. But for Claude, I can, like, have it handle a lot more Data. But I've been giving my podcast, I've been giving my essays so you can, like, kind of know who I am, which I think is kind of neat.

[Grace Uzevian]

Yeah. I find also with, like, content distribution, I've been post, like, similarly, you have, like a whole media empire now.

[Podcast Host / Interviewer]

I don't. I just want to post a lot, that's all.

[Grace Uzevian]

But I find, like, sometimes I want to repurpose content between LinkedIn and versus TikTok and, you know, all of the platforms. So I've just built a couple different apps to do that for me because, like, that is a task that I just dread doing. I hate uploading it to TikTok and to YouTube shorts and to Facebook.

[Podcast Host / Interviewer]

That's true.

[Grace Uzevian]

It's just annoying. So I've.

[Podcast Host / Interviewer]

I have to talk to you a lot more because, like, I, I hate doing that too. And I think that's actually really important to understand too, is like, at least when I got out of the conference, also, is that you really have to understand what you don't like doing. And there is potential that that can be automated. Right. Yeah, I think that that's where a lot of the key insights can happen. So we're going to wrap up this interview. I heard you're in south by Southwest recently and you took a picture of you and I think Ed Ellison. Is that right? From the Prof. Tea. Yeah. I don't know. I just found that really funny. I listen to. Yeah, yeah.

[Grace Uzevian]

You said he. He answered your question too. Or Scott.

[Podcast Host / Interviewer]

Scott did. Scott did, actually. And funny enough, Ed Ellison interviewed my past co founder at Better Meat called Paul Shapiro.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

Which I thought was just really cool. I don't know. So with Scott, I asked. I asked Scott Galloway a question about alternative meat and saying, like, what can we do to, like, improve the alternative meat industry? Even if I said, like, you're in a nuclear winner, deal with it. Which is okay. But I don't know, I just found that, like, kind of cool that he answered it. So, you know, I think, yeah, obviously it's there. That podcast specifically I always recommend and talk about probably too much, but I would say, like, they're really proactive and, like, understand the impact of AI. Right. Do you have any other media or anything you went to in south by Southwest that you would recommend other people listen to about whether it's about the impact of AI or how to use AI better? Would love to hear that.

[Grace Uzevian]

Greg Eisenberg's podcast, I think it's called the Startup Ideas Podcast. That's a good.

[Podcast Host / Interviewer]

I saw his post I. I didn't know he had a podcast, though.

[Grace Uzevian]

Yeah. And then the Y Combinator podcast. I specifically listen to both of these podcasts on YouTube. I watch them and that helps a lot. Especially Y Combinator ones. Not every episode, but their AI ones are really interesting. They're long, but they're good.

[Podcast Host / Interviewer]

Yeah. I, I recommend a. A follower, an influencer. I saw her get her started on Tick Tock. Her name is Serena Romanov.

[Grace Uzevian]

Love her too.

[Podcast Host / Interviewer]

Yeah, she's super. Yeah. Also very soothing voice.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

But. But she's a really, really smart, really down to earth, like, content creator. She gave me, like, she posted a ton of good research papers that I found, like, really fascinating about the impacts of AI. But. Yeah. So obviously there's so many resources out there. Right. And we. You listed a lot of great ones. We're probably not going to go through all. And also I hope you become a good resource also.

[Grace Uzevian]

Thank you.

[Podcast Host / Interviewer]

If I just ask you to post things, I think you'll get this because a lot of stuff you're doing, I just think no one has any idea about, like, top 10 LinkedIn posts. Like, I have no idea how to do that. Right.

[Grace Uzevian]

Oh, wow. Thank you. This is sponsors to share a lot more because.

[Podcast Host / Interviewer]

Okay.

[Adam Yee]

Yeah, Good.

[Grace Uzevian]

I love that.

[Podcast Host / Interviewer]

Yeah. You know, you might be following the experts here, but everyone is just so new to this technology. I just think we just need to share more. I'm going to try to do my part, but I, I even, I feel very insecure about sharing what I, what I do in my workflow. Right. So, you know, I hope that gives you some solace.

[Grace Uzevian]

Yeah, totally. Even. Even at South By, I think, actually Ed Ellison, in their, like, podcast, the live version of the podcast, he was like, he counted the entire schedule and it was 89 talks on AI this year.

[Podcast Host / Interviewer]

Wow.

[Grace Uzevian]

Which he then said to Scott. And Scott was like, it's crazy because GLP1s are a much more transformative technology than AI will ever be. And I just found that such an interesting concept between the two of them where it's like, we are all in these. At least I'm in this bubble of, like, everyone is so deep in AI that taking a step out is also a good thing.

[Podcast Host / Interviewer]

No. 100%, actually, GLP1s is like my second pillar that I'm focusing a lot about as well. That's another podcast, but also it's hard to find experts on GLP1s because, like, everyone's so secret about it.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

So. So it's like super hard to find, like, an Expert interview on them.

[Grace Uzevian]

I'll send you some names.

[Podcast Host / Interviewer]

Oh, thank you. I was trying to get some big wigs, but they were like, oh, nah,

[Adam Yee]

we're not going to do it.

[Podcast Host / Interviewer]

And I was like, oh, come on.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

Anyways, no, I'd love to learn.

[Grace Uzevian]

He won Tangent there. Yeah.

[Podcast Host / Interviewer]

Yeah. I've talked a lot about it, and I think it is very important. I think both technologies are super important.

[Grace Uzevian]

Yeah.

[Podcast Host / Interviewer]

It's also, like, societally very important as well, which I find really interesting as well. But I think what I do find really interesting is, like, those are the most innovative technologies. Like, if people say we're not innovating, those two are, like, fundamentally going to change how we innovate, how we change our perception of not only, like, the world, but also our food.

[Grace Uzevian]

Yeah. Remember from one of your podcasts with Jenny Stoicovic? You both had. You were like, you know, we're not gonna. The foods we grew up eating are not the foods we're gonna die eating. And that feels more true than ever now. And the routines we grew up doing is going to be so different by the time we die, for sure.

[Podcast Host / Interviewer]

My final question is, where can we find you if we want to learn more about not only you, but also the cool prompts you're going to unleash into the world?

[Grace Uzevian]

LinkedIn is my primary. So, Grace Ubezian. It's U V E Z I A N. My substack is now raceU, and those are definitely my primary channels right now.

[Podcast Host / Interviewer]

Love it. Well, Grace, it was good to, I guess, digitally meet you in person. Yeah. But, you know, we've known each other for a long time on LinkedIn, so it just didn't feel. It just felt the same. Right.

[Grace Uzevian]

So totally. Yeah.

[Adam Yee]

Cool.

[Grace Uzevian]

This was lovely.

[Adam Yee]

This is the end of the show, and here's something to chew on.

[Podcast Host / Interviewer]

Instead of me blathering out for, like,

[Adam Yee]

five minutes about random stuff or something I find valuable in the call. I'm going to blather on.

[Podcast Host / Interviewer]

On a podcast I was on last

[Adam Yee]

week, which is the Wolfing Down Food Science Podcast, which is an NC State podcast made by. I guess now I can consider him my professor and my friend Gabriel Keith

[Podcast Host / Interviewer]

Harris, or Dr. Harris. He does this podcast with another friend

[Adam Yee]

of mine, Paige Luck, who is a lecturer. And, you know, I will say that I get credit for helping them make

[Podcast Host / Interviewer]

this podcast because they were inspired by me.

[Adam Yee]

But anyways, I've been a guest for

[Podcast Host / Interviewer]

this podcast about, like, three times right

[Adam Yee]

now, and I now go to NC State as a graduate student while also doing a bunch of other crap.

[Podcast Host / Interviewer]

Anyways, bye.

[Adam Yee]

Anyways, I, I wanted to share this excerpt from the latest podcast I did, which you can find on Wolfing down food science. What Dr. Harris and I did was we went to a movie. It was a documentary called the AI Doc how to Be an Apocalyptimist. And we watched it and it was super fun to hang out with my professor during the weekend. We then went to a Chinese restaurant and pontificated and talked about kind of the impacts of that movie. So the AI Doc is a, I think an independent documentary made by a bunch of like AI think tanks, plus a director of like Everything Everywhere all at once. And, and the narrative is that this guy is trying to get a lot of opinions on who or on is AI good or bad. And the answer is it's both. But I would say it's a really good interview.

[Podcast Host / Interviewer]

He got a lot of really powerful

[Adam Yee]

people in that documentary from people like the founder of OpenAI and the founder of Anthropic. The movie does a really good job of just, just showing both sides, but also really pushing on the fact that we do have the power as human beings to have a say on how this technology is being used. And there's a lot of pros and there's a lot of cons. As you can tell from this interview, I am pretty pro for it. I really like this technology. It has saved me a ton of time. It has saved me. It just has made me a more curious person and I learn a lot. If I did not use this technology, I would be stuck being a plant based meat scientist. But now because I've learned how to use the tool correctly, I can be any food scientist I want, which I think is amazing.

[Podcast Host / Interviewer]

Of course. Are there trials and errors? Yes.

[Adam Yee]

And do I make a lot of mistakes?

[Podcast Host / Interviewer]

Not as much as you think.

[Adam Yee]

But you know, you need an expert's mind to really understand like the outputs of these large language models. But I would say, like I have

[Podcast Host / Interviewer]

some credibility on these tools.

[Adam Yee]

I've done a lot of work with Stanford on these tools and State has given me a lot of resources to really dive into this. And I am like really investing my time and my money on this technology and so far I will say it's paid off. And generally my point of view is that any technology that is overall trending is important to ride the wave because you're going to learn how to innovate much more faster than a, than technology that may not be, may not be progressing as fast. So I do honestly believe plant based meat has stalled in terms of its innovation. But I learned a ton and I gained a lot of capital, knowledge and mindset through doing two. No, not two startups, three startups on alternative meat.

[Podcast Host / Interviewer]

I'm going to show you a quick excerpt on this podcast.

[Adam Yee]

It's at the end of the episode. Funny enough, the person I mentioned in

[Podcast Host / Interviewer]

this interview is Grace. I plan to interview a lot of

[Adam Yee]

experts of people using AI and because honestly it is something I really want to get into. I just some revelations is that I

[Podcast Host / Interviewer]

don't think a lot of people know how to use these tools and I

[Adam Yee]

do think that is something I want to teach people how to do or enjoy. This excerpt on the Wolfing Down Food Science podcast and if you want to listen to the full episode you can search on it.

[Podcast Host / Interviewer]

I'll put a link on the show

[Adam Yee]

notes but you can search on it anywhere you find podcasts. We have an access to a tool

[Podcast Host / Interviewer]

that is cheap now, it might not be cheap in the future.

[Adam Yee]

That and we have an ecosystem where people can say their opinions instantly. And I think that's really important to keep in mind like simple size. Like at least I've been diving into

[Podcast Host / Interviewer]

Claude for instance and a lot of. I have a lot of interviews coming up from my podcast My Food Shot Rocks, which is a plug about like someone who really figured out how to

[Adam Yee]

use Claude and her resources were just Twitter or X whatever Reddit and substack

[Podcast Host / Interviewer]

like you know there. I think that's, that's the really cool

[Adam Yee]

thing is that we, if you are curious and you want to be more of an expert, this tool can, can help you do that. And I think that's, I think now

[Podcast Host / Interviewer]

is the best time to do it

[Adam Yee]

whether or not you're a doomer or, or an optimist because, because it's, it's just, it's just not capital intensive to look into this. Yeah, yeah, it's not capital intensive. Curious or it's not even time intensive

[Podcast Host / Interviewer]

anymore because now you can just figure it out.

[Adam Yee]

Like that's what the positives I want to really say is that we are

[Podcast Host / Interviewer]

at a time where we can learn this and you can learn faster than

[Adam Yee]

anyone else in your field if you figure out this technology. And I do think like overall, overall it will help you. And I think if the secret really,

[Podcast Host / Interviewer]

and I think this is how people

[Adam Yee]

are going to get hired in the next couple of years is if you can communicate, you can make value out of this technology. And you know, is that going to last long?

[Podcast Host / Interviewer]

I don't know, but I do think in the next two or three years

[Adam Yee]

people are trying to figure that out and so really play with it, Become

[Podcast Host / Interviewer]

a kid, be a kid again and just explore what the world has to

[Adam Yee]

offer with this technology. I love it. Okay, well, let's end on that optimistic, positive note. Let's, let's play. Okay. Thank you so much, Adam.

[Podcast Host / Interviewer]

Really appreciate it. I'll be here many more times. Thank you for listening to the My

[Adam Yee]

Food Job Rocks podcast.

[Podcast Host / Interviewer]

If you like this podcast, give us

[Adam Yee]

a five star review on Apple Podcasts or Spotify. If you have any questions or suggestions on what we should talk about or who we should interview, let me [e-mail] on the start of CPG's R D radio, I interview my colleague Colleen Cottrell, who was a former colleague of mine at Motif Food Works. And if you're a food scientist, I

[Podcast Host / Interviewer]

think it's a really good one to listen to about just kind of what

[Adam Yee]

we talk about when it turns to ingredient functionality. I think this was a really good exploration of talking about ingredient functionality to an audience that really doesn't understand they just seen like all purpose flour and bread flour or they didn't see in a whey protein concentrate or a whey protein isolate.

[Podcast Host / Interviewer]

So you can check that out at the start of CPG podcast.

[Adam Yee]

It should be under R and D Radio.

[Podcast Host / Interviewer]

Thank you for the Spoon, Never for housing this podcast and we'll see you

[Adam Yee]

next week for another exciting interview with another food professional.

[Grace Uzevian]

If you enjoyed this episode, why not share it with a friend? And don't forget to give us a five star review. You'll find more content like this on the Spoon Tech Podcasts.