AI Tinkerers - "One-Shot"

Discover how AI is bridging the communication gap between humans and dogs, unlocking deeper insights into canine emotions, intentions, and even health. In this One-Shot interview, Praful Mathur, founder of Sarama, shares his groundbreaking work on a full-stack AI system designed to interpret dog vocalizations and body language. Praful, an experienced builder in AI, reveals how his innovative approach could transform our understanding of our furry companions.

What you’ll learn:

• The surprising history and potential of human-animal communication through AI.

• How to build custom AI models for complex, real-world data like dog vocalizations.

• Praful’s unique strategy for collecting and annotating large datasets from pet owners.

• The practical applications of AI in detecting subtle health issues in dogs.

• How AI tools can accelerate product development, from industrial design to strategic planning.

Key topics covered:

• The sophistication of dog understanding vs. current AI models.

• Leveraging dogs as ‘biological peripherals’ for detection.

• Training original AI models on novel datasets (SVMs, KNNs, transformers).

• Hardware and software architecture for real-time animal data collection.

• Using AI for rapid industrial design and company building tasks.

Join Joe from AI Tinkerers One-Shot as he takes a deep dive with Praful Mathur, an innovator pushing the boundaries of AI to create meaningful connections between humans and animals. This conversation explores the technical challenges and profound implications of building AI that truly understands our pets.

💡 Resources:

• Sarama - https://www.sarama.ai/

• Sarama on IG - https://instagram.com/withsarama

• Praful Mathur’s LinkedIn - linkedin.com/in/praful-mathur

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://one-shot.aitinkerers.org/

Social Media: @AITinkerers @PrafulMathur

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction

00:15 Joe’s Reflection on the Interview

01:32 Introducing Praful Mathur & Sarama

00:01:55 Understanding Dog Communication

00:03:39 Beyond Words: Dog Communication

00:04:37 Dogs as AI Peripherals

00:05:27 History of Animal Communication Tech

00:06:47 Sarama: What They’ve Built Today

00:07:47 Sarama Hardware & Data Collection

00:09:31 Sarama App & Cloud Processing

00:11:30 Understanding Dog Behavior & Emotion

00:12:57 Training Original AI Models for Dogs

00:16:32 Multimodal Data & Sensors

00:18:49 Confidence & Data Needs for Dog AI

00:20:19 ML Stack & Training Approaches

00:22:06 Live Demo: Dog Bark Analysis

00:28:20 Dog’s Vocabulary & Alerts

00:30:13 AI for Early Health Detection

00:32:13 AI in Meta Process & Design

00:35:04 Open Source Strategy & Data Collection

00:37:18 Communicating AI Insights to Humans

00:38:36 Agentic Coding Workflow

00:44:19 Model Comparison: Gemini vs. Claude

00:46:39 How Praful Found AI Tinkerers

00:47:48 Conclusion

What you’ll learn:

• How reinforcement learning can reduce AI agent error rates by up to 60% and drastically lower inference costs.

• The critical difference between supervised fine-tuning and RL for agentic workflows, and why RL is essential for true agent reliability.

• A practical, code-level walkthrough of building and training an email search agent that outperforms OpenAI’s GPT-3.5 on a 14-billion-parameter open-source model.

• Strategies for generating high-quality synthetic data and designing nuanced reward functions with ‘partial credit’ to effectively train your agents.

• Key use cases where RL fine-tuning delivers the most significant benefits, including real-time voice agents and high-volume applications.

Kyle Corbett is the founder of OpenPipe, a platform dedicated to helping enterprises build and deploy customized AI models using advanced fine-tuning and reinforcement learning. He’s a seasoned builder who has been working at the frontier of fine-tuning since before public APIs existed.

Key topics covered:

• The limitations of off-the-shelf LLMs for agent reliability and how RL solves them.

• The importance of latency and cost optimization in real-world AI deployments.

• Detailed explanation of the agentic workflow and tool calling in an email search bot.

• The Enron email dataset as a realistic environment for agent training.

• OpenPipe’s open-source Agent Reinforcement Trainer (ART) library for building RL agents.

• The iterative process of data generation, rubric-based scoring, and model updates.

This episode of AI Tinkerers One-Shot goes under the hood with Kyle to share practical learnings for the community.

💡 Resources:

• OpenPipe Website - https://openpipe.ai

• Kyle Corbett LinkedIn - https://www.linkedin.com/in/kcorbitt/

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://one-shot.aitinkerers.org/

Social Media: @AITinkerers @OpenPipeAI @corbtt

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction

01:09 Welcome Kyle Corbett, Founder of OpenPipe

01:55 What OpenPipe Does

02:31 OpenPipe’s Journey and YC Experience

00:04:13 Email Search Bot Project Overview

00:05:19 Why Fine-Tuning for Email Search

00:06:22 Email Search Bot: Queries and Results

00:09:23 On-Premise Deployment and Data Sensitivity

00:10:45 Agent Trace Example and Tooling

00:13:55 Using the Enron Dataset

00:15:13 Reinforcement Learning Fundamentals

00:17:01 Synthetic Data Generation with Gemini 2.5 Pro

00:18:51 Reliable Q&A Pairs and Data Scale

00:21:59 Fine-Tuning Impact on Model Performance

00:22:25 RL Adoption in Industry and Community

00:24:37 Rollout Function and Agent Implementation

00:27:52 Rubric and Reward Calculation for RL

00:30:39 Training Loop and Model Updates

00:33:52 RL Fine-Tuning vs. OpenAI’s Fine-Tuning

00:40:38 Time Commitment for RL Projects

00:41:55 Use Cases for RL Fine-Tuning

00:45:37 OpenPipe’s Offerings: Open Source, White Glove Service

00:47:07 Kyle’s Side Tinkering and Future of AI

00:49:59 Discovering AI Tinkerers

Discover a groundbreaking approach to optimizing Large Language Models with Tomasz Kolinko, a true OG tinkerer and entrepreneur. In this One-Shot interview, Tomasz unveils his 'Effort Engine,' a novel algorithm that dynamically selects which computations are performed during LLM inference, allowing for significant speed improvements while maintaining surprising output quality. Learn how this method goes beyond traditional quantization by dynamically managing computations and even enabling partial model loading to save VRAM.

Tomasz shares his unique benchmarking techniques, including the use of Kullback-Leibler divergence and heat maps, offering a new lens to understand how models behave under reduced 'effort.' This conversation provides practical insights into the underlying mechanics of AI models and offers a fully open-source project for practitioners to experiment with.

💡 Resources:

• Tomasz Kalinko's GitHub - https://kolinko.github.io/effort/about.html

• The Basics - https://kolinko.github.io/effort/equations.html

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://one-shot.aitinkerers.org/

Social Media Tags: @AITinkerers @kolinko

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction

00:01:07 Welcome Tomasz Kalinko

00:02:11 Introducing Effort Engine

00:03:10 Dynamic Inference Explained

00:05:56 How the Algorithm Works

00:08:07 Speed vs. Quality Trade-offs

00:11:37 Dynamic Weight Loading & VRAM

00:15:24 Effort Engine Demo

00:26:01 Model Breakdown Observations

00:29:49 Architecture & Benchmarks

00:32:17 Kullback-Leibler Divergence

00:39:22 Heat Map Visualization

00:41:07 Community & Future Work

Discover how to build powerful AI agents that integrate with your personal communication, as Luke Harris, Head of Growth at ElevenLabs, shares his journey and groundbreaking projects. Luke, a true tinkerer, reveals how he built the most popular WhatsApp MTP server and an ingenious iPhone shortcut for superior voice transcription.

What you’ll learn:

• The surprising gap in consumer AI and how to build solutions for it.

• Practical insights into building secure AI agent systems and managing data privacy.

• How ElevenLabs’ new speech-to-text (Scribe) and conversational AI APIs enable advanced voice agents.

• The importance of launching personal projects and the unexpected opportunities they create.

• Emerging growth areas for AI in education and niche industries.

Luke Harris leads growth at ElevenLabs, a leading company in generative voice AI. With a background spanning software engineering, biotech ML, and entrepreneurship, Luke is a prolific builder and a Y Combinator alum.

Key topics covered:

• Building the WhatsApp MTP server and its technical challenges.

• The power of Apple Shortcuts for AI-powered mobile workflows.

• ElevenLabs’ Scribe model for highly accurate speech-to-text and diarization.

• Designing conversational AI agents with low latency and emotional expressiveness.

• AI-first workflows for product development, content creation, and sales intelligence.

This episode of AI Tinkerers One-Shot goes deep under the hood with a builder at the forefront of voice AI.

💡 Resources:

• ElevenLabs - https://elevenlabs.io

• Luke Harris’s Blog - https://harrys.co

• Luke Harris’s LinkedIn - linkedin.com/in/luke-harris-ai

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://aitinkerers.org/podcast

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction to Luke Harris

00:01:51 Luke’s Background and Growth at ElevenLabs

00:04:59 The WhatsApp MTP Server Project

00:11:17 Data Privacy and Security in AI Agents

00:16:15 Automation vs. Human-in-the-Loop

00:27:17 ElevenLabs Speech-to-Text and iPhone Shortcut

00:31:34 ElevenLabs Scribe: Advanced Transcription Features

00:35:03 Conversational AI and Voice Agents

00:44:08 Building Voice Agents with ElevenLabs

00:51:41 Unexpected Growth Drivers for Voice AI

01:00:15 Luke’s AI-First Workflow and Resources

01:05:22 Keeping Up with AI & AI Tinkerers Community

Discover how to build powerful AI agents that integrate with your personal communication, as Luke Harris, Head of Growth at ElevenLabs, shares his journey and groundbreaking projects. Luke, a true tinkerer, reveals how he built the most popular WhatsApp MTP server and an ingenious iPhone shortcut for superior voice transcription.

What you’ll learn:

• The surprising gap in consumer AI and how to build solutions for it.

• Practical insights into building secure AI agent systems and managing data privacy.

• How ElevenLabs’ new speech-to-text (Scribe) and conversational AI APIs enable advanced voice agents.

• The importance of launching personal projects and the unexpected opportunities they create.

• Emerging growth areas for AI in education and niche industries.

Luke Harris leads growth at ElevenLabs, a leading company in generative voice AI. With a background spanning software engineering, biotech ML, and entrepreneurship, Luke is a prolific builder and a Y Combinator alum.

Key topics covered:

• Building the WhatsApp MTP server and its technical challenges.

• The power of Apple Shortcuts for AI-powered mobile workflows.

• ElevenLabs’ Scribe model for highly accurate speech-to-text and diarization.

• Designing conversational AI agents with low latency and emotional expressiveness.

• AI-first workflows for product development, content creation, and sales intelligence.

This episode of AI Tinkerers One-Shot goes deep under the hood with a builder at the forefront of voice AI.

💡 Resources:

• ElevenLabs - https://elevenlabs.io

• Luke Harris’s Blog - https://harrys.co

• Luke Harris’s LinkedIn - linkedin.com/in/luke-harris-ai

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://aitinkerers.org/podcast

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction to Luke Harris

00:01:51 Luke’s Background and Growth at ElevenLabs

00:04:59 The WhatsApp MTP Server Project

00:11:17 Data Privacy and Security in AI Agents

00:16:15 Automation vs. Human-in-the-Loop

00:27:17 ElevenLabs Speech-to-Text and iPhone Shortcut

00:31:34 ElevenLabs Scribe: Advanced Transcription Features

00:35:03 Conversational AI and Voice Agents

00:44:08 Building Voice Agents with ElevenLabs

00:51:41 Unexpected Growth Drivers for Voice AI

01:00:15 Luke’s AI-First Workflow and Resources

01:05:22 Keeping Up with AI & AI Tinkerers Community

Learn how to integrate AI agents directly into your existing applications and unlock new levels of user experience and developer productivity with Atai Barkai, CEO of CopilotKit. Discover how CopilotKit provides the essential infrastructure to bridge advanced AI models with your current software stack, making your applications smarter and more intuitive.

What you’ll learn:

• How CopilotKit’s CoAgents and AGUI protocol simplify the integration of AI agents into any application, regardless of framework.

• The practical benefits of implementing ‘SaaS Copilots’ to reduce learning curves and enhance user interaction in complex software.

• Real-world strategies for driving significant internal efficiency gains within large enterprises using AI agents.

• Why the ‘human-plus-AI’ mental model is crucial for the foreseeable future of intelligent systems.

Atai Barkai is the founder and CEO of CopilotKit, a leading open-source framework for building production-ready AI copilots. With extensive experience in AI agent user experience, Atai shares insights from working with indie developers to Fortune 100 companies, offering a unique perspective on the evolving AI landscape.

Key topics covered:

• Bridging AI agents with existing application UIs for enhanced functionality.

• Understanding the AGUI (Agent User Interaction Protocol) for seamless agent-user communication.

• Implementing intent-based interfaces for complex SaaS applications.

• Achieving ‘industrial evolution level productivity gains’ with AI co-agents.

• The open-source model of CopilotKit and its ease of self-hosting.

This episode of AI Tinkerers One-Shot goes under the hood with Atai Barkai to share practical learnings for the community.

💡 Resources:

• CopilotKit Website - https://copilotkit.ai

• Atai Barkai LinkedIn - https://www.linkedin.com/in/atai-barkai

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://aitinkerers.org/podcast

Social Media: @AITinkerers @copilotkit @ataiiam

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction

00:01:20 Welcome to One-Shot

00:01:55 How Joe Met Atai

00:03:29 Getting Started with CopilotKit

00:03:52 CopilotKit Components: Standard Agent & CoAgents

00:04:50 AGUI Protocol & Events Explained

00:08:05 CopilotKit UI & Shared State Demo

00:11:34 Open Source vs. Cloud Model

00:13:18 Integrating CoAgents into Your App

00:14:38 Why Bring Agents into Applications?

00:15:39 Practical Agent Adoption & Use Cases

00:17:42 Learning More About CopilotKit

00:18:15 What Atai is Tinkering With

00:21:26 Craziest & Most Impactful Use Cases

Learn how to integrate AI agents directly into your existing applications and unlock new levels of user experience and developer productivity with Atai Barkai, CEO of CopilotKit. Discover how CopilotKit provides the essential infrastructure to bridge advanced AI models with your current software stack, making your applications smarter and more intuitive.

What you’ll learn:

• How CopilotKit’s CoAgents and AGUI protocol simplify the integration of AI agents into any application, regardless of framework.

• The practical benefits of implementing ‘SaaS Copilots’ to reduce learning curves and enhance user interaction in complex software.

• Real-world strategies for driving significant internal efficiency gains within large enterprises using AI agents.

• Why the ‘human-plus-AI’ mental model is crucial for the foreseeable future of intelligent systems.

Atai Barkai is the founder and CEO of CopilotKit, a leading open-source framework for building production-ready AI copilots. With extensive experience in AI agent user experience, Atai shares insights from working with indie developers to Fortune 100 companies, offering a unique perspective on the evolving AI landscape.

Key topics covered:

• Bridging AI agents with existing application UIs for enhanced functionality.

• Understanding the AGUI (Agent User Interaction Protocol) for seamless agent-user communication.

• Implementing intent-based interfaces for complex SaaS applications.

• Achieving ‘industrial evolution level productivity gains’ with AI co-agents.

• The open-source model of CopilotKit and its ease of self-hosting.

This episode of AI Tinkerers One-Shot goes under the hood with Atai Barkai to share practical learnings for the community.

💡 Resources:

• CopilotKit Website - https://copilotkit.ai

• Atai Barkai LinkedIn - https://www.linkedin.com/in/atai-barkai

• AI Tinkerers - https://aitinkerers.org

• One-Shot Podcast - https://aitinkerers.org/podcast

Social Media: @AITinkerers @copilotkit @ataiiam

👍 Like this video if you found it valuable, and subscribe to AI Tinkerers One-Shot for more conversations with innovators building the future of AI!

00:00 Introduction

00:01:20 Welcome to One-Shot

00:01:55 How Joe Met Atai

00:03:29 Getting Started with CopilotKit

00:03:52 CopilotKit Components: Standard Agent & CoAgents

00:04:50 AGUI Protocol & Events Explained

00:08:05 CopilotKit UI & Shared State Demo

00:11:34 Open Source vs. Cloud Model

00:13:18 Integrating CoAgents into Your App

00:14:38 Why Bring Agents into Applications?

00:15:39 Practical Agent Adoption & Use Cases

00:17:42 Learning More About CopilotKit

00:18:15 What Atai is Tinkering With

00:21:26 Craziest & Most Impactful Use Cases

What if you could skip half of your LLM’s computations—and still get the same output?

In this episode of One-Shot, we sit down with Tomasz Kolinko, the Warsaw-based founder of Effort Engine—a new AI inference algorithm that dynamically adjusts precision in real time.

This isn’t quantization. It’s something weirder—and maybe more useful.

Tomasz walks us through how he:

- Built a custom algorithm that runs 2–3x faster on MacBooks

- Developed a system that can skip 50%+ of model computations dynamically

- Created heatmaps to visualize token-level divergence

- Benchmarked everything himself… and shared the code

You’ll also see:

- Live demos of inference tuning from 100% to 5%

- Why AI models still work (sometimes better!) with just 30% effort

- How a DIY hacker space in a car shop led to one of the most creative AI projects in Europe

If you’re building with LLMs, pushing inference limits, or just obsessed with optimization — this episode will change how you think about AI computation.

In this episode of AI Tinkerers "One-Shot", we go deep with Steve Krenzel, founder of Logic, on how his company turns standard operating procedures (SOPs) into fully functioning APIs. Dive deep with us on schema generation, test cases, structured outputs, and backtesting.

We break down:

1. Why Steve avoids agentic frameworks

2. How Logic automates 100K+ tasks/month for real customers

3. The power of structured output for reasoning and reliability

4. How prompt caching and append-only templates unlock scale

5. His open-source coding agent that builds software from scratch

6. How they achieved less than 2% error rates beating human teams

7. His famous Prompt Engineering Guide that went viral in 2023

If you’re building with LLMs, designing autonomous workflows, or just want to see what the future of developer productivity looks like—this is a must-watch.

Relevant Links:

Follow Steve: https://www.linkedin.com/in/stevekrenzel/

Follow Logic: https://www.linkedin.com/company/with-logic

From the episode:

- http://github.com/stevekrenzel/pick-ems

- http://app.staging.logic.inc/

What if building complex AI agents felt as natural as composing React components—and they could even rewrite their own code? 🤯

In this episode of One Shot / AI Tinkerers, host Joe sits down with Evan Boyle, founder of GenSX, to explore a radically new way to design, run, and ship long-running agent workflows:

🔑 Key takeaways

- React-inspired component model for agents – why JSX-style, type-safe functions beat static graphs for scalability and code reuse.

- Traces, telemetry & evals baked-in – see every prompt, variable, and LLM call in real time.

- $4 self-modifying coding agent – Evan demos an agent that checks out its own repo, refactors 3 K lines, runs tests, and pushes to GitHub… iteratively.

- Real-world production use cases – from million-document legal discovery to inbox-wide entity extraction and analytics.

- Durable execution & infra shift – why 5-second latencies and massive parallelism are forcing a rethink of serverless, queues, and caching.

- Developer experience first – faster dev loops with component-level caching, cursor rules, and LLM “rubber-duck” debugging tricks.

🛠️ Tools & frameworks mentioned

GenSX, React/JSX, OpenAI & Anthropic models, Temporal, Pulumi, Cursor, LangChain, LlamaIndex, Crew AI…and more.

🔗 Try GenSX → https://www.gensx.com

💬 Join the community → https://github.com/gensx-inc/gensx

🐦 Follow Evan on X/Twitter → https://x.com/_Evan_Boyle

🙌 Enjoyed the conversation?

👍 Like, 🔔 subscribe, and drop your questions or aha moments in the comments. It helps more builders discover the pod!

📍 Chapters

00:00 Intro & Evan’s background

04:28 Why existing agent frameworks break at scale

12:55 Inside the React-style component model

23:10 Live demo: Hacker News Analyzer (1,000 LLM calls in parallel)

32:45 Tracing, telemetry, and evals

38:20 The self-modifying code agent ($4/iteration)

50:40 Real production agent use cases

59:05 Dev-tooling tips: caching, logging-only debug loops

1:08:30 The future of AI infrastructure & closing thoughts

#GenSX #AIAgents #DeveloperExperience #React #SelfModifyingCode #AIWorkflow #OneShotPodcast