Barrchives

What if deep learning isn’t the future of AI—but just part of it?In this episode, Ori Goshen, Co-founder and Co-CEO at AI21 Labs, shares why his team set out to build reliable, deterministic AI systems—long before ChatGPT made language models mainstream.We explore the launch of Wordtune, the development of Jamba, and the release of Maestro—AI21’s orchestration engine for enterprise agentic workflows. Ori opens up about what it takes to move beyond probabilistic systems, build trust with global enterprises, and balance research and product in one of the most competitive AI markets in the world.If you want a masterclass in enterprise AI, model training, architecture tradeoffs, and scaling innovation out of Israel—this is it.🔔 Subscribe for deep dives with the people shaping the future of AI.This episode is broken down into the following chapters:00:00 – Intro00:47 – Why AI21 started with “deep learning is necessary but not sufficient”02:34 – Building reliable AI systems from day one03:46 – The risk of neural-symbolic hybrids and early bets on NLP05:40 – Why Wordtune became the first product08:14 – From B2C success to a pivot back into enterprise09:43 – What AI21 learned from Wordtune for enterprise AI11:15 – Defining “product algo fit”12:27 – Training models before it was cool: Jurassic, Jamba, and beyond13:38 – How to hire model-training engineers with no playbook14:53 – Recruiting systems talent: what to look for16:29 – How to orient your models around real enterprise needs17:10 – Why Jamba was designed for long-context enterprise use cases19:52 – What’s special about the Mamba + Transformer hybrid architecture22:46 – Experimentation, ablations, and finding the right architecture25:27 – Bringing Jamba to market: what enterprises actually care about29:26 – The state of enterprise AI readiness in 2023 → 202531:41 – The biggest challenge: evaluation systems32:10 – What most teams get wrong about evals33:45 – Architecting reliable, non-deterministic systems34:53 – What is Maestro and why build it now?36:02 – Replacing “prompt and pray” with AI for AI systems38:43 – Building interpretable and explicit agentic systems41:09 – Balancing control and flexibility in orchestration43:36 – What enterprise AI might actually look like in 5 years47:03 – Why Israel is a global powerhouse for AI49:44 – How Ori has evolved as a leader under extreme volatility52:26 – Staying true to your mission through chaosSubscribe to the Barrchives newsletter: https://www.barrchives.com/Spotify: https://open.spotify.com/show/37O8Pb0LgqpqTXo2GZiPXfApple: https://podcasts.apple.com/us/podcast/barrchives/id1774292613Twitter: https://x.com/barrnanasLinkedIn: https://www.linkedin.com/in/barryaron/

What does it take to reimagine the browser—one of the most commoditized technologies in the world—for the enterprise?In this episode, Ofer Ben Noon, founder of Talon and now part of Palo Alto Networks, shares the wild journey from exploring digital health to building the world’s first enterprise-grade secure browser.We dig into:Why the browser became the new security perimeterHow Talon raised a $26M seed and scaled fastWhat it takes to compile Chromium daily (and why it’s so hard)Why Precision AI is essential to secure AI usage in the enterpriseAnd how generative AI, SaaS sprawl, and autonomous agents are reshaping enterprise risk in real timeIf you care about AI x cybersecurity, endpoint security, or enterprise infrastructure—this is a deep, real, and tactical look behind the curtain.This episode is broken down into the following chapters:00:00 – Intro01:05 – Why Ofer originally wanted to build in digital health02:15 – The pandemic shift to SaaS, hybrid work, and browser-first04:44 – Why Chromium was the perfect technical unlock05:27 – The insane complexity of compiling Chromium07:10 – What makes an enterprise browser different from a consumer browser09:36 – Browser isolation, web security, and file security10:50 – Why Talon needed a massive seed round from day one11:53 – What an MVP looked like for Talon14:08 – Early skepticism from CISOs and how Talon earned trust16:50 – Discovering new enterprise use cases over time17:11 – How AI and Precision AI power Talon’s security engine19:21 – Why Ofer chose to sell to Palo Alto Networks21:06 – Petabytes of data, 30B+ attacks blocked daily23:44 – The risks of LLMs and generative AI in the browser24:24 – What Talon sees when users interact with AI tools25:05 – The #1 risk: privacy and user error26:43 – Why AI use must be governed like any other SaaS27:22 – How Talon built secure enterprise access to ChatGPT28:05 – Mapping 1,000+ GenAI tools and classifying risk29:43 – Real-time blocking, DLP, and prompt visibility31:25 – Why user mistakes are accelerating in the age of agents32:04 – How autonomous AI agents amplify risk across the enterprise33:55 – The browser as the new control layer for users and AI36:57 – What AI is unlocking in cybersecurity orgs39:36 – Why data volume will determine which security companies win40:28 – Ofer’s leadership philosophy and staying grounded post-acquisition42:40 – Closing reflectionsSubscribe to the Barrchives newsletter: https://www.barrchives.com/Spotify: https://open.spotify.com/show/37O8Pb0LgqpqTXo2GZiPXfApple: https://podcasts.apple.com/us/podcast/barrchives/id1774292613Twitter: https://x.com/barrnanasLinkedIn: https://www.linkedin.com/in/barryaron/

Vanta helped create the automated security and compliance category—and now, they’re redefining it with AI.In this episode, Christina Cacioppo (CEO & Co-Founder) and Iccha Sethi (VP of Engineering) join Barr Yaron to go deep on how AI is transforming the way Vanta builds products, evaluates models, and helps companies earn and demonstrate trust.They cover:- Why compliance is the perfect playground for AI- How Vanta balances reliability, explainability, and scale- What it takes to build golden datasets in high-stakes domains- The real-world AI infrastructure behind Vanta AIIf you care about real AI product development—not just hype—this is a masterclass in doing it right.🔔 Subscribe for more deep dives with leading AI builders and thinkers.This episode is broken down into the following chapters:00:00 – Intro01:06 – Christina’s early entrepreneurial roots (Beanie Babies & all)02:51 – From venture to founder: why Christina started Vanta04:00 – What Vanta actually does05:32 – Iccha on why she joined as VP of Engineering07:09 – When Vanta started leaning into AI08:33 – AI’s growing role in Vanta’s product roadmap09:52 – How AI powers questionnaire automation12:25 – Using LLMs to map policy docs to cloud configs13:27 – Building trust: human-in-the-loop and explainability16:03 – Vanta’s evaluation system for AI features18:17 – How golden datasets are constructed (and maintained)20:59 – Feedback loops: online eval from user behavior22:43 – How model feedback informs product updates23:38 – What Vanta wants from foundation models (but isn’t getting yet)24:32 – Retrieval: how Vanta processes customer documents27:13 – The hardest technical challenges in AI integration29:41 – Internal adoption: how non-technical teams are using AI too31:52 – Vanta’s centralized AI team & how other teams plug in33:27 – Internal education: building AI intuition org-wide34:31 – From prototype to production: experimentation culture36:41 – Customer sentiment around AI in compliance workflows38:22 – Enterprise buyers & the AI “kill switch”39:06 – Personalized experiences as the future of trust40:21 – How enterprises are approaching AI risk assessments41:50 – What excites Iccha and Christina about the future of AI at VantaSubscribe to the Barrchives newsletter: https://www.barrchives.comSpotify: https://open.spotify.com/show/37O8Pb0LgqpqTXo2GZiPXfApple: https://podcasts.apple.com/us/podcast/barrchives/id1774292613Twitter: https://x.com/barrnanasLinkedIn: https://www.linkedin.com/in/barryaron/

Karan Goel, Co-founder and CEO of Cartesia, dives into the future of voice AI and the groundbreaking use of state space models (SSMs) for audio applications. He details his transition from academia at CMU and Stanford to entrepreneurship, emphasizing the innovative efficiency of SSMs over traditional models. Karan also reveals how Cartesia is developing Sonic, an ultra-low latency text-to-speech model, and elaborates on the importance of rapid execution in voice AI, all while navigating the startup landscape.

Kashish Gupta, Co-founder and Co-CEO of Hightouch, discusses the revolutionary impact of AI on marketing. He delves into the rise of Composable CDPs, illustrating how they empower marketers with data democratization. Gupta reveals how reinforcement learning agents optimize campaigns through tailored reward systems and highlights the technical challenges faced in building these AI models. He emphasizes the balance between AI efficiency and the creative input needed from marketers to truly engage customers in an increasingly automated landscape.

This episode consists of the following chapters: 00:00 - Introduction to Jesse Zhang and Decagon02:33 - Why customer support emerged as a clear use case for AI05:00 - The importance of discovery and understanding customer value08:20 - The Decagon product architecture: core AI agent, routing, and human assistance11:01 - How enterprise logic is integrated into the AI agent15:45 - Shared frameworks across different customers and industries17:12 - How AI agents are changing organizational planning19:59 - Automatically identifying knowledge gaps to improve resolution rates22:57 - Handling routing across different modalities (text and voice)26:09 - The continued importance of humans in customer support30:17 - The evolving role of human agents: supervising, QA, and logic building36:57 - Value-based pricing tied to the work AI performs39:17 - How sophisticated buyers evaluate AI customer support solutionsSubscribe to the Barrchives newsletter: www.barrchives.comSpotify: https://open.spotify.com/show/37O8Pb0LgqpqTXo2GZiPXfApple: https://podcasts.apple.com/us/podcast/barrchives/id1774292613Twitter: https://x.com/barrnanasLinkedIn: https://www.linkedin.com/in/barryaron/

Regal co-founders Alex Levin and Rebecca Greene see a future where AI doesn't just assist human agents in contact centers - it replaces them entirely at the front lines. In this episode of Barrchives, we discussed how they're building voice AI technology, the unique challenges of working with audio models, and what the future of customer service looks like.

Most of what you hear about AI right now is in text, but Mikey Shulman (co-founder and CEO of Suno) would tell you that audio is a much more interesting medium to work with. How do you use AI to generate music? What makes audio data uniquely difficult to parse? And how do you build audio models that cater to unique, subjective human preferences on music? Suno is building a future where anyone can make great music. In this episode of Barrchives, I sat down with Mikey (who, like his co-founder, is a musician) to talk about how they do what they do, from why they chose a transformer-based architecture to how they test new models when outputs are so subjective.

AI and data teams in a sense, kind of, do the same thing: make decisions based on data. So how do you build AI that *helps* data teams do their best work? Hex was one of the first companies in their space to embrace language models and build code generation features into their data workspace. In this episode of Barrchives, I went deep with Hex’s co-founder and CEO, Barry McCardel, about Hex’s journey towards becoming an AI company.

Erik Bernhardsson, CEO of Modal and former leader at Spotify, joins the conversation to discuss his entrepreneurial journey and the challenges of building infrastructure for AI teams. He shares insights into enhancing developer experiences and the importance of user-centric design. Erik dives into the impact of generative AI on customer attraction and the complexities of managing resources like GPUs. The discussion also touches on the need for low-latency solutions in AI applications and how well-positioned companies can navigate market growth.