Building With AI

In this episode of Building With AI, host Hamza Choudery sits down with Scott White, a Product leader at Anthropic. Scott brings a wealth of experience from his time at companies like Dropbox, Wealthfront, Instacart, and Airtable. At Airtable, Scott first gained hands-on experience integrating LLMs into software.

At Anthropic, Scott contributed to the development of Claude. During the conversation, he takes us behind the scenes, sharing the principles and frameworks used by one of the most innovative AI companies in the world.

During the episode, he discusses:

• How the Product org is structured at Anthropic,
• His framework for de-risking AI products,
• Anthropic's AI development process,
• Bridging the gap between Research and Product,
• And much more!

—

[00:00] Up next:

[00:52] Episode introduction

[02:04] Autoblocks: AI Optimization Platform

[02:30] Scott's journey

[07:53] The product team at Anthropic

[11:31] How AI changes product development

[17:10] Anthropic’s AI product development process

[22:55] The key to AI product innovation

[26:48] Claude in 10 years

[34:12] Bridging the gap between Research and Product

[38:13] Trust and safety in AI

[44:14] Scott's hottest AI take

[45:57] Scott's advice for product builders in the LLM space

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Scott White:

• Linkedln: https://www.linkedin.com/in/scottiewhite/

Learn more about Anthropic:

• Website: https://www.anthropic.com
• LinkedIn: https://www.linkedin.com/company/anthropicresearch/
• X: https://x.com/anthropicai

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://twitter.com/AutoblocksAI

In this episode, host Hamza Choudery is joined by Sridhar Ramaswamy, the CEO of Snowflake. Before taking the helm at Snowflake, Sridhar scaled Google's ads business from ~$1B to $115 billion and founded one of the first AI search companies, Neeva (which was acquired by Snowflake in 2023).

During the episode, he discusses:

• The first principles for AI product building,
• The importance of product velocity,
• The future of the foundation model layer,
• Creating data flywheels with AI,
• And much more!

—

[00:00] Up next:

[00:29] Episode introduction

[01:28] Autoblocks: AI Optimization Platform

[01:52] Sridhar's journey

[08:47] The birth of Neeva and the GPT moment

[13:32] Lesson from Google for startups

[16:19] The Snowflake acquisition

[21:48] Snowflake's AI product strategy

[26:35] Snowflake's internal and external AI applications

[28:39] Snowflake Cortex

[30:37] The future of the foundation model layer

[34:52] The AI data flywheel

[38:09] Building reliable LLMs

[41:52] Lightning round

[43:08] Sridhar's hottest AI take

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Sridhar Ramaswamy:

• Linkedln: https://www.linkedin.com/in/sridhar-ramaswamy/
• X: https://x.com/RamaswmySridhar

Learn more about Snowflake:

• Website: https://www.snowflake.com/en/
• LinkedIn: https://www.linkedin.com/company/snowflake-computing/
• X: https://x.com/snowflakedb

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://twitter.com/AutoblocksAI

In this episode of Building With AI, host Haroon Choudery sits with Tomasz Tunguz, Founder and General Partner at Theory Ventures. Tomasz brings a valuable 10,000-foot view of the AI space, drawing from his 16 years of experience as a VC and previous work building machine learning systems at Google.

During the episode, Tomasz discusses:

• The modern data stack for LLMs,
• The significance of human evaluations,
• How to Identify high-value use cases for businesses,
• Midjourney's successful distribution strategy,
• And much more!

—

[00:00] Up next: 

[00:57] Episode introduction 

[01:59] Autoblocks: AI Optimization Platform 

[02:24] Tomasz's introduction 

[03:47] The two waves of AI 

[06:09] The modern data stack for LLMs 

[11:27] The golden set of evaluations 

[15:39] Why you need human evaluations 

[17:27] Don't write off LLMs 

[21:04] Identifying high-value AI use cases for your business 

[25:30] The top industries LLMs will disrupt 

[26:52] Advice for incumbents to win the AI race 

[33:05] Advice for AI startups to disrupt industries 

[34:31] The innovation equation 

[37:16] Midjourney's successful distribution strategy 

[39:15] Will AI move beyond non-determinism? 

[43:21] The future relationship between AI and professionals 

[48:05] How to keep up with AI developments

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Haroon Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Tomasz Tunguz:

• Linkedln: https://www.linkedin.com/in/tomasztunguz/
• X: https://x.com/ttunguz

Learn more about Theory Ventures:

• Website: https://theory.ventures
• LinkedIn: https://www.linkedin.com/company/theory-ventures/
• X: https://x.com/theoryvc

Where to find Haroon Choudery:

• Linkedln: https://www.linkedin.com/in/haroonchoudery/
• X: https://x.com/haroonchoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://twitter.com/AutoblocksAI

In this episode of Building With AI, host Hamza Choudery sits with Jayodita Sanghvi, Senior Director of Data Science at Included Health. Jayodita completed a PhD in Bioengineering at Stanford before jumping in industry with the goal of raising the standard of healthcare for everyone.

In her current role leading data science at Included Health, her team uses AI to help simplify healthcare for users by providing services like clinical care and specialist recommendations.

During the episode, Jayodita discusses:

• How LLMs are going to improve healthcare delivery,
• Why Included Health relies primarily on prompt engineering,
• How to get the most out of early adopters by using them for feedback loops,
• Regulation and bias considerations for AI in healthcare,
• And much more!

—

[00:00] Up next:

[00:57] Episode introduction

[02:04] Autoblocks: AI Optimization Platform

[02:29] Jayodita’s introduction

[05:25] A look into Included Health

[08:53] Jayodita on LLMs in healthcare

[15:36] The GenAI dream team

[17:17] API vs. OpenAI: What does Included Health use?

[18:16] Why Included Health relies primarily on prompt engineering

[20:40] How to get the most out of early adopters

[22:52] Regulation and bias in AI for healthcare

[26:55] Startups and incumbents: Will both win?

[30:04] The future of ML & LLMs in healthcare

[33:43] Jayodita's hottest AI take

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Jayodita Sanghvi:

• Linkedln: https://www.linkedin.com/in/jayodita-sanghvi-34b1057/

Learn more about Included Health:

• Website: https://includedhealth.com/resources/
• LinkedIn: https://www.linkedin.com/company/includedhealth/
• X: https://x.com/IncludedHealth

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://x.com/AutoblocksAI

In this episode of Building With AI, host Haroon Choudery sits down with Rahul Sonwalkar, the founder and CEO of Julius.ai, an AI data analyst that helps users analyze datasets, create visualizations, and gain insights from structured data. They discuss Rahul's entrepreneurial journey, including his time at big tech companies like Facebook and Uber, the infamous "Ligma Johnson" prank, and the launch of his own venture.

During the conversation, they discuss:

• The inspiration for creating Julius before the "ChatGPT moment,"
• Julius's rapid growth to 500,000+ users,
• How to think about competition in the AI space,
• The importance of testing and user feedback,
• And much more!

—

[00:00] Up next

[00:54] Episode Introduction

[02:00] Autoblocks: AI Optimization Platform

[02:39] Rahul's Background & Introduction to Julius.ai

[04:29] Product Iterations before the "ChatGPT moment”

[05:58] Importance of User Testing & Feedback

[14:28] User Feedback Loop at Julius.ai

[17:58] How to think about competition in the AI space

[24:34] Specialization vs Super Apps

[28:42] Comparing Code Interpreter and Julius.ai

[33:57] The infamous "Ligma Johnson" prank

[35:59] Pitching Julius.ai’s vision to Jensen Huang

[37:28] Engineering & Infrastructure Challenges

[41:22] Product Design Challenges

[42:01] Future Milestones for Julius.ai

[43:11] Advice for Aspiring AI Founders

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Haroon Choudery (CEO and Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Rahul Sonwalkar:

• Linkedln: https://www.linkedin.com/in/rahulsonwalkar23/
• X: https://x.com/0interestrates

Learn more about Julius.ai

• Website: https://julius.ai
• LinkedIn: https://www.linkedin.com/company/julius-ai/
• X: https://x.com/juliusai_

Where to find Haroon Choudery:

• Linkedln: https://www.linkedin.com/in/haroonchoudery/
• X: https://x.com/haroonchoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://x.com/AutoblocksAI

In this episode of Building With AI, host Hamza Choudery sits down with Quentin Clark, Managing Director at General Catalyst. Drawing from his extensive background, including senior engineering and product roles at Microsoft and serving as CTO at SAP and Dropbox, Quentin offers a privileged perspective on the latest wave of AI.

He provides an insightful breakdown of the modern AI stack—covering hardware, foundation models, platforms, and applications—and shares his bets on where value will accrue.

During the episode, they discuss:

• Reimagining AI-native workflows,
• The “ChatGPT moment,"
• Challenges with enterprise AI adoption,
• Competition between incumbents and startups,
• Advice for product teams building AI-powered applications,
• And much more!

—

[00:00] Up next: 

[00:52] Episode introduction 

[01:51] Autoblocks: AI Optimization Platform 

[02:17] Quentin's introduction

[07:41] The “ChatGPT moment" 

[25:48] The modern AI stack 

[35:57] The future of AI in software engineering 

[41:43] Competition between incumbents and startups 

[46:57] Navigating non-determinism in AI

[49:37] The foundation model layer: API vs. open-source

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Quentin Clark:

• Linkedln: https://www.linkedin.com/in/quentin-clark/
• X: https://x.com/quentinclark

Learn more about General Catalyst:

• Website: http://generalcatalyst.com
• LinkedIn: https://www.linkedin.com/company/general-catalyst/
• X: https://x.com/generalcatalyst

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://x.com/AutoblocksAI

Akilesh grew up watching his dad, an oncologist, slog over tedious documentation tasks. Inspired by his father and their shared vision of a Jarvis-like assistant for doctors, he dove into the intersection of healthcare and AI.

During the conversation, they discuss:

• His experience training custom NLP,
• How DeepScribe responded to the “ChatGPT moment,”
• The importance of AI evaluation and testing,
• The future of DeepScribe and Ambient AI,
• And much more!

—

[00:00] Up next: [00:56] Episode introduction

[02:04] Autoblocks: AI Optimization Platform

[02:29] Akilesh’s introduction

[05:42] How LLMs Revolutionized NLP Development

[08:57] How does DeepScribe work?

[13:32] How DeepScribe responded to the ChatGPT moment

[20:18] How to choose the right LLM evaluation technique

[22:01] Building the ultimate AI product team

[24:32] DeepScribe's approach to evaluating at scale

[27:40] Challenges of building with non-deterministic systems

[29:01] DeepScribe's go-to-market strategy

[32:25] The future of DeepScribe and ambient AI

[36:40] Regulatory considerations for health tech companies

[38:48] DeepScribe's product roadmap

[40:42]  DeepScribe's feedback loop

[42:29] Opportunities in health tech

[44:14] Akilesh's hottest AI take

[49:23] Open source vs. foundation models

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Akilesh Bapu

• Linkedln: https://www.linkedin.com/in/akileshbapu/
• X: https://x.com/akileshbapu

Learn more about DeepScribe

• Website: https://www.deepscribe.ai
• LinkedIn: https://www.linkedin.com/company/deepscribe/
• X: https://x.com/DeepScribeAI

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://x.com/AutoblocksAI

In this episode of Building With AI, Haroon Choudery is joined by Dan Dayon, the founder of Readwise - a platform that enhances memory retention from reading by providing an all-in-one solution for managing digital content. Last year, Readwise ventured into the world of AI by launching Ghostreader, their AI reading assistant.

During the episode, Dan talks about:

• UI/UX challenges he faced when designing AI products,
• The feature that unexpectedly became a fan favorite,
• Readwise's plans for the future,
• Automating "bullshit jobs,"
• And much more!

—

[00:00] Up next: 

[00:58] Episode introduction 

[02:09] Autoblocks: AI Optimization Platform 

[02:34] Dan’s introduction 

[04:44] Readwise's journey 

[15:56] Building traditional features vs. AI features 

[23:25] Why Readwise won't build a chat feature 

[27:46] The fan favorite feature 

[31:18] Readwise's plans for the future 

[40:11] Dan's "BS jobs" advice

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Haroon Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Dan Dayon:

• Linkedln: https://www.linkedin.com/in/danielfdoyon/
• X: https://twitter.com/deadly_onion

Learn more about Readwise:

• Website: https://readwise.io
• LinkedIn: https://www.linkedin.com/company/readwise/
• X: https://twitter.com/readwise

Where to find Haroon Choudery:

• Linkedln: https://www.linkedin.com/in/haroonchoudery/
• X: https://x.com/haroonchoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• LinkedIn: https://www.linkedin.com/company/autoblocks/
• X: https://twitter.com/AutoblocksAI

In this episode of Building With AI, host Haroon Choudery is joined by Naveen Rao, VP of GenAI at Databricks. Naveen is a serial entrepreneur who established the first AI chip company, which he sold to Intel in 2016. He then founded MosaicML in 2020, which was acquired by Databricks for $1.3 billion last year.

During the episode, he shares:

• MosaicML’s jobs to be done,
• Why MosaicML teamed up with Databricks,
• The economics of training AI models,
• Advice for teams building with AI,
• And much more!

—

[00:00] Up next: 

[00:51] Episode introduction 

[01:40] Autoblocks: AI Optimization Platform 

[02:06] Naveen's Introduction 

[05:17] Naveen's experience scaling AI products 

[06:52] Why we need to scale AI 

[11:40] MosaicML's jobs to be done 

[14:09] Why did MosaicML team up with Databricks? 

[18:24] LLMs 101 

[24:47] The economics of training AI models 

[28:11] Training a new model vs. adapting a general model 

[30:51] The shelf life of AI models 

[34:33] A look into DSPy 

[41:11] The challenges of DSPy 

[43:59] Techniques and architectures that excite Naveen 

[47:58] Advice for teams building with AI

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Naveen Rao:

• Linkedln: https://www.linkedin.com/in/naveen-rao-bba5b01/
• X: https://twitter.com/naveengrao

Learn more about Databricks:

• Website: https://www.databricks.com
• LinkedIn: https://www.linkedin.com/company/databricks/
• X: https://twitter.com/Databricks

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• LinkedIn: https://www.linkedin.com/company/autoblocks/
• X: https://twitter.com/AutoblocksAI

In this episode of Building With AI, host Hamza Choudery is joined by Vinod Khosla, the Founder of Khosla Ventures and a renowned technologist. Vinod has played a key role in Silicon Valley's history and is currently at the forefront of the latest wave of AI adoption as the first investor in OpenAI.

During the conversation, they discuss:

• Vinod's philosophy on risk,
• Advice for entrepreneurs building with Al,
• Open-source vs. API models,
• The Al race between the US and China,
• And much more!

—

[00:00] Up next: 

[00:44] Episode introduction

[01:45] Autoblocks: AI Optimization Platform

[02:11] Vinod's introduction

[04:50] Building a techno-optimist worldview

[07:45] How do you become a visionary?

[12:44] Vinod’s manifesto

[16:36] The power of risk

[19:27] AI in the Khosla household

[22:54] AI’s impact on industries

[24:43] Insights into a successful product team

[26:21] Current opportunities in AI

[29:10] The AI race between the US and China

[33:00] Predictions on the AI race

[36:55] The future of vertical specific models

[38:34] Balancing dependencies and innovation

[40:00] AI as a reasoning engine

[44:08] The role of data in AI’s evolution

[47:02] Vinod’s hottest AI takes

[49:43] Vinod’s biggest motivation to get up

—

Brought to you by Autoblocks (https://www.autoblocks.ai/) — how product teams deliver better AI experiences.

—

Building with AI is a weekly podcast featuring conversations with top AI product builders and leaders. Through thoughtful dialogue, host Hamza Choudery (Co-founder of Autoblocks) uncovers tactical insights that help listeners build & scale world-class AI products.

—

Subscribe to Building With AI:

• Website: https://www.autoblocks.ai/bwai
• X: https://twitter.com/buildingwith_ai

Where to find Vinod Khosla:

• X: https://twitter.com/vkhosla

Learn more about Khosla Ventures:

• Website: https://www.khoslaventures.com
• LinkedIn: https://www.linkedin.com/company/khosla-ventures/
• X: https://twitter.com/khoslaventures

Where to find Hamza Choudery:

• Linkedln: https://www.linkedin.com/in/hamzachoudery
• X: https://twitter.com/hamzachoudery

Learn more about Autoblocks:

• Website: https://www.autoblocks.ai/
• X: https://twitter.com/AutoblocksAI