Is the failure rate for enterprise AI POCs super high because they’re “science projects” solving a problem that’s already been solved?Guest Bret Greenstein, Chief AI Officer of West Monroe, dropping all sorts of knowledge in AI with Alec podcast E22:“We already know that AI can do this work. Proving value is actually sort of a cop-out.”“It's much more phases than POCs.”“A lot of POCs were done by technical people who proved it could be done with no line of sight to the implementation and adoption.”“And the people who actually know how work is done are not sitting in an IT center.” “They're near where the work is.”“So the best projects, and I've seen hundreds of them, are when the business is involved with technology and the leadership understands what they're trying to accomplish.” “Otherwise it's just a science project.”“And there's a lot of science projects or people pushing for a license for a thing that's going to magically change your life. It's like people selling a pill to make you live longer.”If that didn't convince you to watch the episode, here are 5 more:1️⃣ Humans in the loop or on top of the loop. They’re your decision maker. Always have been and always will be.2️⃣ Operationalizing AI-first ambitions? “You can't have leaders who don't understand it. Now that being said, I don't think they have to be data scientists, but they do have to understand the nature of it.”3️⃣ “If you can’t imagine it, you can’t make it.”4️⃣ Evolution from Prompt engineering to Context engineering because “knowing which data matters is really useful” and “providing more data that’s relevant” is critical.5️⃣ In the enterprise AI era, buy vs build vs partner? “Look for no regret moves” as the “world is literally forming around us” aka make investments that become more valuable not less valuable as things change.Chapters + Timestamps:00:00 - Introduction00:26 - 100K Employees Doing 300K Work: The AI-Native Vision02:58 - Internal Audit Case Study: 50% Workload Reduction06:02 - From Fear to AI Whisperer: Building Soft AI Skills08:24 - Why Your Data Doesn't Need to Be Clean (Yet)13:04 - The Death of Application-Centric Architecture16:27 - Chat Interface vs Beautiful UI: What Users Actually Want19:01 - The No Regrets Move Strategy for AI Investment23:30 - Forward Deployed Engineers vs Business Transformation25:56 - The POC Trap: Why "Science Projects" Fail27:35 - Leaders Don't Need to Be Data Scientists (But...)30:02 - Electricity Metaphor: Reimagining Your Business30:50 - "If You Can't Imagine It, You Can't Make It"33:05 - Closing

One of the most beautiful ways to describe the essence of the collaboration between athletes + coaches?“We see movement as a language that actually needs to be taught.”Ryan Talbot, co-founder and CEO of VueMotion dropped that phenomenal quote on me during our AI with Alec interview.VueMotion is “a computer vision AI machine learning technology” company that works with professional and collegiate sports teams to introduce objective data into critical decision making processes. Enabling teams to capture and analyze data from the natural environment aka practice, games etc vs requiring athletes to go into labs / academic environments.Imagine how much better teams could perform by learning the language of movement, introducing objective data and feeding both into the genius machine to augment human intelligence, judgement and ultimately performance?Especially given that they are using data / tech infrastructure that’s 20+ years old to make critical sports decisions today.In the AI era of business, this is the equivalent of using tools comparable to a rock to hammer nails vs a nail gun…I’m +10000 here for this 👇“And we're really focused on our goal to be able to analyze a million plus people a day using computer vision, machine learning, to be able to change lives.”Chapters / Timestamps:0:00 - Introduction1:29 - What is VueMotion3:32 - Sports Performance 2.05:50 - The Founding Story9:22 - Industry Problems + Solutions 13:00 - LLMs + the Future 15:40 - Growth Strategy + Scale 18:02 - Real-World Impact21:29 - Closing Thoughts

Kyle Wallack’s art is phenomenal, but the way this “refined graffiti artist” has embraced AI? That’s an inspiration for artists everywhere.

This AI with Alec interview is for you if you’re into…

Street Art.

AI enabling artists to spend more time creating art vs everything else.

Intersection of culture, fashion, sports, creativity etc.

Humans + Machines not Humans vs. Machines.

6 of my favorite parts of our conversation?

1️⃣ Intersection of high fashion + sports:

“Michael Jordan shooting a Louis Vuitton basketball.”

2️⃣ AI enables Kyle to overcome dyslexia:

“I always felt like I can’t portray what I’m thinking in my mind because it’s just hard for me to put it together in a way that makes sense.”

3️⃣ Kyle isn’t running away, he’s running towards AI:

“I look at AI as just another tool we’ve created to help us.”

4️⃣ Clearly Coinbase recognizes talent:

Kyle collaborated with Coinbase on “a robot on the back of a jacket for the NBA All-Star game, and Franco Finn wore that.”

5️⃣ AI is good for artists:

“I can speak on 10 different reasons why AI is good for artists…It’s going to make you push yourself, if you use it the right way…I’m trying to become better versed in art, just out of respect for people who did it before me and out of the need to continuously get better.”

6️⃣ Kyle’s response to fears about AI creating art was absolutely lights out:

“The biggest fear that I've heard is, my friend who's in medicine said this to me. He's like, ‘What are you going to do? They physically have AI painting oil paintings?’ And I was just like, yeah, but that's not my painting…there's still that human factor that comes with making physical art.”

👇 Most importantly, link to where you can find and buy Kyle’s gorgeous art: https://kylewallack.com/

Chapters and timestamps:0:00 - Introduction & Discovery1:27 - The Cancer Journey & Finding Art3:19 - Robot Art Origins & Evolution5:27 - Sports Influence & Legacy Mindset7:53 - Culture Meets Sports: The Jordan & Ali Pieces12:14 - AI as Accessibility Tool for Dyslexia16:31 - AI as Creative Secretary & Manager19:03 - 10 Reasons Why AI is Good for Artists25:48 - Addressing AI Art Fears27:28 - Upcoming Projects & Contact Info

Enterprise AI agents help us “focus on making the beer taste better” aka the 20%+ of efforts that drive 80%+ of the results.

Kurt Muehmel, Head of AI Strategy at Dataiku, returned to AI with Alec for another AI deep dive. 

This time, the focus was enterprise AI agents and why they represent a fundamental shift in how work gets done.

The thought provoking conversation moved through a number of critical success drivers for enterprise AI agents including the following:  

1: Three-part definition of AI agents that cuts through agent washing hype

2: Why technology capabilities are racing ahead of enterprise adoption

3: The “AWS moment” for AI…are we approaching standardization of the abstraction layer between intelligent systems and human collaboration?

4: Implications of AI agents helping us “traverse the boundaries of traditional software”

5: Building the “hybrid human agent workforce” including the tooling and mindset shifts required

Link to "The LLM Mesh: An Architecture for Building Agentic Applications in the Enterprise":

https://pages.dataiku.com/oreilly-tech-guide-llm-mesh

00:00 - Introduction & Welcome

00:29 - Enterprise AI Agents Fundamentals

01:28 - Three-Part Definition: LLM, Autonomy, Integration

05:48 - Breaking Software Silos & Application Boundaries

09:28 - Focus on Making the Beer Taste Better

12:22 - Enterprise Use Cases: From Support to Complex Judgment

18:38 - Single Agent vs Multi-Agent Architectures

21:42 - Digital Labor & Cultural Adoption Challenges

28:19 - The Hybrid Human-Agent Workforce

30:18 - Future Model Architectures & The AWS Moment

36:08 - Services-Oriented Architecture for AI

37:00 - Wrap-up & Kurt's O'Reilly Book on LLM Mesh

Is AI thinking or just imitating our thinking?

A new Apple research paper, "The Illusion of Thinking," suggests these reasoning models are doing the latter. Sophisticated mimicry at a previously unimaginable scale.

In an AI with Alec fireside chat, I spoke to Henry Saba to gather his perspective on the paper and get his take on what this really means for businesses and builders in the near-term.

As a leader in the space of building data infrastructure, automation and intelligent systems for Citadel and Lazard before starting his own company, Specialized Data Company, Henry’s perspective was insightful and clear.

Exactly the kind that is the most helpful when trying to make sense of the exponential change all around us:

1️⃣ Biggest takeaway from the paper is “their inability to solve generalizable problems”

2️⃣ Think of LLMs as pattern-matching machines (aka massive idioms / Mad Libs → s / o Jack Clark and Rick Rubin) vs LRMs as systems that appear to think through problems step-by-step

3️⃣ Put “AI at the core of the systems that you build” but be sure you’re asking it to solve problems it has seen before

4️⃣ Near-term, “tightly scoped” and “very refined, very purpose-built agents for specific use cases” is where a ton of value will be generated

5️⃣ Gap between AI marketing hype and product / capabilities is getting really wide, critical to know the difference

Research link: https://ml-site.cdn-apple.com/papers/the-illusion-of-thinking.pdf

If I told you the average employee toggles 1,200 times per day between applications, would you believe me?In my latest AI with Alec podcast, Chuck Doerr, co-Founder + CIO of HERE, not only confirmed this staggering reality but laid out a number of practical ways to operationalize enterprise AI aka translate the AI-first memo movement into results.An origin story that has resulted in deep and wide adoption across the Financial Services industry, Chuck’s perspective on enterprise AI hype vs results, especially within Financial Services is nuanced and refreshing to hear.1️⃣ Harvard research shows the “Toggle Tax” includes knowledge workers toggling 1,200 times per day, context switching that is a hidden productivity killer HERE is addressing via “Super Search”2️⃣ Helping enterprises convert an AI-first memo into a mindset shift and ultimately an environment where the work can start to get done within “this framework of trust around using AI” in a way that embraces “observability, explainability and verification”3️⃣ Enabling enterprise AI to interpret the digital exhaust in real-time, unlocking next best action, scale the problem solving approach of top performers, introduce analytics that can benefit all sorts of cross-functional initiatives is stepping into an enterprise AI operating system…a territory I’m all aboutWhile HERE has deep roots in Financial Services, their expansion into Call Centers and other environments where decisions must be made “with high fidelity, low error rates, and rapidly” through a “single pane of glass” makes a ton of sense.Check out the interview to learn more.

After meeting the client in real life, the first job for an Athena Intelligence client-facing leader is to “build a carbon copy of themselves in the product for the customer”...If that doesn’t capture your undivided attention, go to the 19 min and 12 second mark and listen to Brendon unpack this concept in 175 seconds. For laughs, watch my reaction...AI with Alec Episode #16 is with Brendon Geils, founder + CEO of Athena Intelligence.While everyone is now talking about enterprise AI Agents, my man has been doing it in highly regulated environments because “we want our brand to reflect the security and sensitivity of some of the most important institutions” (i.e.) Financial Services. “We're building an enterprise analyst, or as we like to say, an AI employee for regulated industries.”Gnarly.Top 3 takeaways:1: Athena is a horizontal play focused on enabling clients to solve the hardest problems, the ones they don’t see coming but need to read and react to, fast“I think we shine best when there's some brand new thing that hits someone's desk and there's not an off the shelf solution for it.”2: Brendon believes it’s “better to be small in this business” because you can “scale in other ways by building agents to do the job that maybe historically you would have to hire people for”“Because for this thing to work well, it's gotta live in less people's heads. The more heads you live in, the more it becomes a Google PM’d project, a Microsoft PM’d project. So I think for us, we have to stay small as long as we can.”3: Technology, no matter how powerful, is only one leg of the stool.“What I found was most of the time the technology wasn't the rate limiter for a successful pilot, either at Scale or at Palantir. It was your ability to introduce change management to an organization.”

What if the data opportunity on the business side of Sports is bigger than what was profiled in Moneyball on the team / athlete side?

I had a fascinating conversation with Josh Walker, co-founder and CEO of Sports Innovation Lab during an AI with Alec podcast episode who revealed the answer to that question and more.

1️⃣ Data Fragmentation + Silos: Teams sell rights to business partners handling concessions, merchandise, media and ticketing, creating massive blind spots and fragmented data that prevents unified fan understanding.

2️⃣ AI Enabled Hyper-Personalization is Coming: We’re heading to a world where “this is literally about you having a different SportsCenter than I have.”

3️⃣ Derivative Data Advantage: SIL connects dots to enable leagues, teams, sports etc to see things about their fans that they didn’t already know.

4️⃣ Advanced Data Infrastructure + Architecture: Anything knowledge graph, ontology etc captures my undivided attention so Josh and the team had me at “modeling that provides the most comprehensive fan intelligence, accurate targeting, 1P+ enrichment, and personalization at scale.”

Hope you enjoy the episode. I know I did!

What’s the first thing you think when you hear about a start-up launching data centers into outer space?For me, it went like this: “Wait, what? Why outer space…ohhhh wow, the AI energy issue. That’s a massive idea…”Watch this AI with Alec episode with Philip Johnston, co-founder + CEO of Starcloud who is doing exactly that and has “the best space engineers in the world working on it.”1: A bunch of “firsts” coming up“By the end of August we'll have an announcement out about the demonstrator satellite that's going up. So that will have the first terrestrial data center grade GPUs in space. So about a hundred times more powerful GPUs than are currently in orbit.”“We'll be the first to train an AI model in space, the first to run high-power inference, first to do fine tuning of a model in space. Lots of firsts are coming up on that mission.”2: Yes it’s hard but “nothing is against the laws of physics.”“But like we're heading to a world where the launch cost is insignificant compared to everything else. And I think people just have a hard time getting their heads around that. mean, yes, space is hard. Yes, we have to figure out the cooling challenge of building an enormous radiator. Yes, we have to figure out radiation and other things. But all of that, nothing is against the laws of physics. You just have to be really dedicated and figure it out.”3: Swim like a dolphin on the moon? I’m +10000 in.“If you're swimming on the moon, you could basically swim like a dolphin so you can propel yourself out of the water. And like you could actually, you could do the reverse of a dive. Like you could propel yourself so far out of the water you could land on a diving board. Anyway, yeah, you could basically swim like a dolphin on the edge of a crater with earth rising in the background, 24 seven sunlight…”Speaking of dolphins, I might have to finally start watching White Lotus…

The enterprise AI playbook doesn’t exist. We’re all just trying to figure it out as we go. But thanks to AI leaders like Will Croushorn of Wendy’s, we get to hear insights from those doing the work.Will joined me again on an AI with Alec podcast and just like last time, he delivers with a refreshing amount of clarity. The kind that comes from hands in the dirt, enterprise AI experience.1: “There's no playbook for this. I think that every company, every organization, every, every employee, every person is trying to figure out how do I operate in this new world?”2: "If it can be automated, it probably will be…the things that are the most human, those are the things that get highlighted and everything else gets automated."3: “The question for me is how do we get from A to B and can we do it with these tools in a way that's never been done again that serves the customer in a way that we couldn't do otherwise.”If you’re into AI and aren’t following Will, today is most definitely the day to fix that.Hope you check it out and enjoy it. I know I did.Chapters & Timestamps0:00 - Introduction to Will Croushorn and his work at Wendy's2:00 - "No playbook exists" - Navigating the new AI landscape5:27 - The shift from selecting products to being chosen by AI agents8:59 - Human creativity and remarkable experiences in an AI world12:20 - Context-aware retail: Merging physical and digital experiences15:43 - Building trust and user opt-in for data sharing19:39 - The future of personalized marketing and advertising21:18 - Transforming everyday experiences like drive-throughs24:51 - Starting with consumer needs before technology28:16 - Making sense of data and asking the right questions31:35 - Focusing on human connection in an AI-augmented world

In AI, there’s only one “Igor” and I am SO STOKED to share that he joined me for an AI with Alec interview.

Please find time to hear legendary AI pioneer Igor Jablokov, founder + CEO of Pryon, move through all sorts of fascinating topics.

Chapters:

00:00 - Introduction to Igor Jablokov

00:53 - Igor's Journey: From IBM to Siri and Alexa

02:24 - Understanding Pryon and Enterprise AI

04:47 - Working with JD Vance and Rise of the Rest

07:18 - The Three Acts Post-ChatGPT

09:04 - Critical Take on AI Company Practices

11:04 - Lessons from Social Media

12:27 - Thoughts on DeepSeek

15:08 - AI and the Arts: The Opera Analogy

19:07 - Future of Work and AI (2030 Vision)

22:53 - Enterprise AI Adoption

25:35 - The Evolution of AI: From "Punk Rock Days" to Mainstream

27:48 - Closing Thoughts: AI's Role in Solving Global Challenges

Key Highlights:

- Igor reveals how Pryon was the original codename for what became Alexa

- Discussion of the three phases of enterprise response to ChatGPT

- Unique perspective on AI's relationship with creative arts, comparing AI to opera

- Insights on augmented intelligence vs. replacement of workers

- Vision for 10x productivity growth through AI augmentation

- Candid discussion about AI companies' practices and ethics

Guest: Igor Jablokov

- Founder and CEO of Pryon

- Former IBM AI team lead

- Pioneer behind technologies that became Siri and Alexa

- Early AI innovator, pioneer and absolute legend aka AI OG

Since ChatGPT launched 2 years ago, a lot of sophisticated AI investors, entrepreneurs etc have suggested it’s not an “if” but a “when” a solopreneur will build a unicorn.

Could Josh Mohrer, founder of Wave: AI Note Taker for iOS and Android, become that unicorn?

What Josh has built by bootstrapping is incredible and a massive inspiration.
Wave records 6,000+ hours per day for 22,000 subscribers…
Josh always knew the engineers were the real rockstars aka he always had “engineer envy” and for Josh, “this just to scratch my own itch”...

If you don't know about NotebookLM, now you know. It has ben described as "the next ChatGPT" and once you give it a shot, you'll experience the AI magic and know why.

During the episode, we cover:

1: Re: AI, we're past "why" and all in on "how," especially within the Enterprise.

2: This isn't a promise to change the way you work, it's a tangible reality available to anyone. Right. Now.

3: The discipline to build a great product is what it's all about. Reminds me of what the legendary Rick Rubin refers to as the "ruthless edit"...

The 12 minute conversation was fascinating as we covered a lot including the following questions: 1: Why do bankers need more than just a chatbot? 2: How can AI enhance the combination of public and private data sets to create a valuable, accessible and Wall Street specific Knowledge Graph that ultimately enhances the art of storytelling around asset valuations? 3: What are the four parts of the Rogo tech stack? 4: One thing in AI that Tumas believes that most others don’t? s / o Julia Lauer