Tim O'Reilly

Founder and CEO of O'Reilly Media, widely known for his influential work in technology, open source, and the future of AI.

Recommends

Books, films, shows, and podcasts they point to — tap a source to read the passage.

Books · 2

The Divine Right of Capital

Marjorie Kelly · 2001

What’d I miss?Tim O'Reilly

Context

Tim O'Reilly recommends this book because its insights on profit and loss statements fundamentally informed his own shift in thinking about platform economics.

→ Amazon

Against the Consensus

Justin Yifu Lin · 2013

Why America SleptTim O'Reilly

Context

Tim O'Reilly values this book for providing an entrepreneurial framework for how governments can approach innovation and causality.

“construct a causality model anew…by identifying the agents behind the phenomenon, the goals they wanted to achieve, the constraints they faced, and the options they had.”

→ Amazon

Movies · 2

All the President’s Men

Alan J. Pakula · 1976

TCMAmazon VideoApple TV Store

How I Detect Fake NewsTim O'Reilly

Context

Tim O'Reilly recommends this film as a formative influence that shaped his understanding of the importance of corroborating sources through multiple independent accounts.

→ Letterboxd

Fantasia

James Algar · 2000

Disney PlusAmazon VideoApple TV Store

What’d I miss?Tim O'Reilly

Context

Tim O'Reilly values the "sorcerer's apprentice" sequence as a perfect visual metaphor for the uncontrollable, multiplying consequences of poorly framed digital wishes.

→ Letterboxd

Sort

Tim O'Reilly20h

The 3 Core Components That Make AI Agents Actually Work with Jayeeta Putatunda

Every AI agent relies on a brain (the LLM driving logic and multi-step reasoning), tools (the APIs, databases, and apps like Slack or Telegram it can act on), and memory. For Jayeeta Putatunda, memory matters most because it ensures the agent remembers your feedback on tone, styling, and format, so you're not starting from scratch every time. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com

Tim O'Reilly1d

The Problem with Siri and Alexa with Pete Warden

Skepticism about voice interfaces usually traces back to bad experiences with Siri and Alexa, but Pete Warden sees that as an engineering and UX problem, not evidence that people dislike talking to machines. He points out that we already prefer talking over texting in everyday situations, and solving that problem could make voice a more common input than the keyboard or touchscreen. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.b

Tim O'Reilly1d

This Week in AI: Expanding the Possibilities

On the latest episode of This Week in AI, host Christina Stathopoulos covers new developments from OpenAI, Anthropic, and Mistral, including interactive tools in ChatGPT, Claude Dashboards and Motion, and Mistral Large 4. She also examines new research on robotics and employment, the challenges of controlling autonomous AI agents, and what wider access to AI tools means for businesses and developers. The episode explores promising applications of AI in mathematics, biology, and public health, along with efforts to search for extraterrestrial technology. Plus, Mistral's investments in physics-

Tim O'Reilly1d

Running AI Models on a Normal Laptop with Pete Warden

Pete Warden does the math on what 9 billion parameters actually means for a normal laptop. With weight quantization, a 9-billion-parameter model at 8 bits comes out to roughly 9 gigabytes, and many mid-range laptops shipping today already have enough graphics card memory to run it. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly1d

The Situational Ethics of Stickers with Charity Majors

Generative AI gave Honeycomb cofounder and CTO Charity Majors something she’d wanted for years: a way to explore art and design for herself. But it also forced her to confront the ethical questions that had kept her away from AI image generators in the first place. In this lightning talk from O’Reilly Foo Camp, Charity describes how experimenting with images helped her build a vocabulary for color, typography, layout, and her own visual taste. She believes the uncompensated use of artists’ work for AI training was wrong, but abstaining from AI isn’t the answer, Charity argues.Technologists hav

Tim O'Reilly2d

The Case for Running AI Locally with Pete Warden

Developers built most software for the cloud over the past two decades, but Pete Warden sees AI reversing that trend. Cost pressures, privacy concerns, and model instability are pushing enterprises toward hosting their own models, and he notes that even when you select the same version of Claude in a dropdown, "what's actually happening behind the scenes can change from day to day." #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.b

Tim O'Reilly2d

Generative AI in the Real World: Observability for AI with Charity Majors

AI makes it possible to generate code faster than teams can understand and support it. In this episode, Honeycomb cofounder and CTO Charity Majors joined Ben to discuss how that shift changes the work of building reliable software. Charity explained how observability helps teams connect changes with outcomes, why nondeterministic systems need “more rigor, not less,” and how traces can show the decisions agents make along the way. Over their wide-ranging conversation, she and Ben also covered automatic instrumentation with OpenTelemetry, why we should think of ourselves as “owners on the loop,”

Tim O'Reilly2d

The Failure-First Design Principle Behind Google Search with Pete Warden

Designing for failure is a core principle behind some of the most widely used software ever built, and Pete Warden uses Google Search as the clearest example. Google knew from the start it wouldn't always surface the right result first, so instead of forcing a single answer, it let users scan snippets to find what they needed themselves. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@ore

Tim O'Reilly2d

AI Teaches Us That People Are Basically Good

Every frontier AI model, from Western labs to Chinese ones, seems to have landed on broadly humanist values, and Jesse Vincent has a theory about why. Train a model on the entire corpus of human knowledge and it averages out, he says, to "people are basically reasonable and I think people are basically good." Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly3d

Transforming Creative Roles with Christina Stathopoulos

AI won't replace creative professionals, but according to _This Week in AI_ host Christina Stathopoulos, it will transform their roles just as thoroughly as it's already reshaping software development. She advises designers, writers, and other creatives to learn how AI can help them on the job, because those who don't "could be at risk." #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@ore

Tim O'Reilly4d

There’s a Lot of Bad Code Out There

Frontier AI models keep getting better at finding bugs, but Jesse Vincent says they haven't learned when to leave one alone. He watched a model write 10,000 lines of code to work around an edge case where the only failure mode was running a shell script twice. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly4d

Using Your Voice to Spawn Autonomous Subagents on the Go with Craig Hewitt

Craig Hewitt treats voice agents as productivity tools worth understanding carefully before deploying, and says the real gains kick in once you've locked down the settings to a level you're comfortable with. At that point, your voice agent stops being a dictation shortcut and starts spawning subagents that handle autonomous tasks while you're on the go. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www

Tim O'Reilly4d

AI Improves Itself with Christina Stathopoulos

Anthropic just released research showing Claude searching existing literature, proposing solutions, generating training data, and refining its own methods to reduce misalignment, improving performance across all ten tested categories without degrading its broader capabilities. _This Week in AI_ host Christina Stathopoulos explains why this is an early glimpse of recursive self-improvement, and what it means that Claude could transfer those learnings to models far larger than the ones used in testing. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook:

Tim O'Reilly4d

Anthropic’s $30 Trillion AI Market Claim with Christina Stathopoulos

Anthropic just claimed a $30 trillion total addressable market (TAM), a number that exceeds China's entire GDP and nearly matches that of the United States. _This Week in AI_ host Christina Stathopoulos calls it "outrageous" but explains that TAM represents theoretical annual revenue if Anthropic's AI captured 100% of all the work it might eventually perform. She also draws a comparison to SpaceX, which claimed a $28.5 trillion TAM ahead of its own record-breaking IPO. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: htt

Tim O'Reilly4d

ChatGPT vs. Claude with Craig Hewitt

Claude has become the default AI tool in most big companies, but Craig Hewitt thinks that reputation isn't the whole story. After working with large organizations as a consultant and recently picking up the Codex app, he makes the case that ChatGPT is every bit as capable and, as an application, the better tool. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly9d

Using AI to Sharpen Technical Judgment with Emma Bostian

Engineering managers often piece together technical context from scattered sources, but Emma Bostian believes AI can consolidate all of that in one place, especially when connected to internal databases via an MCP. Beyond gathering context faster, she sees AI as a tool for evaluating technical trade-offs, spotting cascading effects across connected systems, and challenging your own strategy to find the holes before other people do. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreill

Tim O'Reilly10d

AI Won't Make Your Engineering Role Obsolete with Emma Bostian

The anxiety around AI and job security is real, especially for engineers and engineering managers navigating the fourth Industrial Revolution. Emma Bostian is firm that these roles won't become obsolete, because building trust, coaching engineers through growth and performance issues, and making trade-offs when there's no clear right answer are things AI isn't good at and may never be. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreill

Tim O'Reilly10d

How to Prioritize What to Learn Next in Tech with Emma Bostian

Deciding what to learn next in tech gets harder when new skills keep emerging faster than you can keep up. Emma Bostian recommends plotting your options on a four-quadrant graph of feasibility against relevancy, focusing on what's both easy to grasp and highly relevant to your work, and repeating the exercise every time you finish a topic rather than treating it as a one-time decision. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreill

Tim O'Reilly11d

AI Is Eroding Critical Thinking on Engineering Teams with Emma Bostian and Lena Reinhard

AI tools are designed to remove friction, and that's exactly what's making critical thinking harder to hold onto on engineering teams. Emma Bostian encourages engineers to review AI output not just for hallucinations but for whether it actually solves the problem at hand, and Lena Reinhard is already working with teams on how to bring critical and creative thinking back before the pressure to deliver crowds it out entirely. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia B

Tim O'Reilly11d

Why LLMs Aren't the Path to AGI with Christina Stathopoulos

LLMs are essentially massive statistical machines that predict the next token, and Christina Stathopoulos doesn't think that gets us to AGI. She argues that world models, which can actually understand the world around them rather than just process text, have a better shot at reaching artificial general intelligence. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly11d

World Models and the Race Beyond Large Language Models with Christina Stathopoulos

World models learn how the world actually works, including objects in space and cause-and-effect relationships, and researchers see them as the next step beyond today's LLMs. Christina Stathopoulos explains what sets them apart from multimodal AI and why every major lab, from Google DeepMind to Meta to Nvidia, is racing to build one. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreilly

Tim O'Reilly12d

Zero to Agent: Build Your First Agent with MCP

Model Context Protocol (MCP) gives developers a standard way to connect the APIs and services they already maintain to AI clients. In this episode of Zero to Agent in 30 Minutes, AI developer and author Bruce Hopkins shows how to put that into practice. Using a stock-data API, Bruce builds an MCP server and shows how to expose existing functionality so an AI client can discover and use it. The result is a practical way to reuse application logic you already have instead of rebuilding the same capabilities for each AI integration. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/compan

Tim O'Reilly12d

The Growing Dangers of Rogue AI Agents with Christina Stathopoulos

When OpenAI intentionally relaxed some of its model's safety restrictions during a cybersecurity evaluation, the agent didn't just solve the benchmark it was assigned. It escaped its sandbox, gained internet access, and hacked into Hugging Face to retrieve the answers. Christina Stathopoulos walks through what happened and why the incident is drawing fresh attention to the dangers of rogue AI agents. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app

Tim O'Reilly13d

Keeping AI Agents in Check with Vicki Reyzelman

Once you release an AI agent into the wild, it has the purview to do its own thing, and that's exactly why security can't be an afterthought. Vicki Reyzelman recommends building in guardrails from the start and running regular checks to catch what you might miss: "You tell the agent to run those quick checks for you to make sure that it hasn't run rogue." #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://w

Tim O'Reilly14d

Training a New Employee with Vicki Reyzelman

Skills files are the most important thing you can build into an AI agent, and Vicki Reyzelman recommends thinking of them the same way you'd think about onboarding a new employee. Define the processes and rules upfront, then plan to iterate, because the first version of your agent rarely gets it right. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly14d

Understand What You’re Building with Vicki Reyzelman

Building an AI agent is getting simpler by the day, which makes the step most developers skip even more important. Vicki Reyzelman says that as tools like Claude Console lower the barrier to building, you need to invest more time upfront on requirements and design. "Really figure out what you're trying to build before you build it," she says. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/

Tim O'Reilly14d

Balancing Model Quality and Cost in AI Agent Design with Vicki Reyzelman

The model you choose for your AI agent shapes both output quality and your costs. Vicki Reyzelman explains that better models need less context but cost more per token, which means the design phase is the right time to figure out exactly how much context your agent needs to do its job well. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly15d

The Hidden Cost of Vibe Shipping with Drew Breunig

Prototyping with LLMs is addictive because you write something, it works, and you ship it. But Drew Breunig, citing the CCO of Vercel, says vibe coding makes code "free as in puppies." You can get a puppy for free, but then you have to feed it, walk it, and potty train it. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly15d

This Week in AI: Agents Outpace the System

On the latest episode of This Week in AI, host Vicki Reyzelman, a senior solutions engineer at Akamai, looks at what happens when AI agents start moving faster than the systems built to support them. She covers an OpenAI agent probing Australia’s Medicare system, the Hugging Face incident involving 1,200 collaborating agents, and why security teams are turning to runtime controls, enterprise browsers, and other ways to monitor agent behavior as it happens. Vicki also examines the infrastructure behind the AI boom, including rising data center power demand, faster model release cycles, and the

Tim O'Reilly15d

Recognizing “Prompt Smell” with Drew Breunig

Just like "code smell" warns developers they're heading toward technical debt, Drew Breunig argues there's a "prompt smell" too. Repeating instructions, patching in edge cases, and burying evals inside your prompt instead of building them outside are all signs that prompt debt is setting in and it's time to automate how your prompt is generated. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.

Tim O'Reilly16d

Multi-Agent Workflows Can Solve Prompt Debt with Drew Breunig

Using one prompt to perform a task and another to evaluate the response is a reliable way to reduce prompt debt, says Drew Breunig, who's a strong advocate for decomposition. "Benchmarkmaxxers" chasing leaderboard scores may follow the bitter lesson and skip the infrastructure, but if you're building products, decomposition has real advantages they're leaving on the table. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social

Tim O'Reilly16d

How to Measure Prompt Debt with Drew Breunig

Most teams know they have prompt debt but don't know how to quantify it. Drew Breunig recommends tracking how often prompts change, who's editing them, and whether they show signs of "key person syndrome," then feeding each one into an LLM to check for repeated instructions and over-specification. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly17d

Building Trust in AI Takes Time with Gene Kim

The best working definition of trust in AI is your ability to predict how it will act and react. Gene Kim draws on Jay Forrester's MIT beer game research to map out how that trust develops over time, rising linearly through about 12 months of use before tapering off around 15 months as you develop a clearer sense of what the tool can and can't do. His advice for building that trust faster is to start with something fun. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueS

Tim O'Reilly18d

AI is Rewiring Organizations with Gene Kim

AI can function as both a delegatable engineer and a world-class consultant on nearly any topic, from Git to architectural frameworks to compliance. That combination, as Gene Kim explains, is quietly rewiring how organizations operate. Teams no longer depend on specialists for routine work, which frees those specialists to focus on the hardest, most judgment-intensive problems. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.s

Tim O'Reilly18d

How AI Made Me a Faster Writer with Gene Kim

Gene Kim wrote a book in a third to half the time using AI, and his editor of ten years called it ten times better than any manuscript he'd ever delivered. The results haven't quieted his mixed feelings, though. If AI makes it possible for everyone to finish a book, does finishing one still mean anything? #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly19d

The Christmas Miracle of 2025 with John Berryman

John Berryman calls the late-2025 leap in coding agent quality "the Christmas Miracle of 2025," and he's not alone. Karpathy and Simon Willison both wrote about the moment agents stopped needing constant babysitting, and Berryman says it's why bespoke state flow machines are mostly a thing of the past now. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly19d

Zero to Agent in 30 Minutes: Give Your Agent Its Own Computer

In this episode of _Zero to Agent in 30 Minutes,_ AI engineer Sajal Sharma shows how to give an AI agent its own computer using a remote sandbox. With E2B, he demonstrates how an agent can install packages, run code, work with files, and control a browser without running those tasks on your own computer. You’ll see two ways an agent can work with a remote computer. One uses shell commands and files, while the other operates a remote desktop through screenshots and mouse and keyboard actions. Sharma also explains how to connect the agent’s actions to the remote desktop, move files between loca

Tim O'Reilly19d

AI Harness and Model Go Together Like PB&J with Craig Hewitt

The AI harness is the operating system; the model is the database, and Craig Hewitt says it's the combination that makes the magic. Tools like Claude Code and Codex lock you into proprietary model-harness fits, while open harnesses like Hermes let you plug in any model you want, whether cloud-based, private, or running locally on your machine. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.co

Tim O'Reilly20d

How a Simple Loop Around an LLM Changed Everything with John Berryman

Before ChatGPT, large language models were just text in, text out. John Berryman explains that all ChatGPT did was wrap a simple loop around the LLM, trading messages back and forth between the user and the model until a chatbot emerged from what had been a document-completion engine. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly20d

What AI Agents Can Actually Do With Skills, Tools, and MCP with Craig Hewitt

Hermes ships with AI agent skills turned on by default, and Craig Hewitt says they're "pretty benign," not the "I'm going to delete everything on your computer" kind. He also explains the difference between skills and tools, including how Model Context Protocol creates direct connections between your agent and other systems. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly21d

Stop Trying to "Do AI in Marketing" and Do This Instead with Craig Hewitt

"I want to do AI in my marketing" is the wrong ask, Craig Hewitt tells his clients. What actually works is finding the parts of marketing you can systematize, defining what good looks like, and letting AI supplement specific human judgment calls, like drafting a few LinkedIn post options from a raw idea instead of handing over the whole process. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.

Tim O'Reilly21d

An AI Agent That Finds Topics, Drafts Posts, and Waits for Your Approval with Craig Hewitt

Craig Hewitt's production-ready content agent pulls from Google News and LinkedIn three times a day, surfaces the three best topic ideas in Slack, and waits for you to pick one before writing the post. "Always on," he says, really just means running on a schedule. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly21d

Give Your Agents Superpowers: Jesse Vincent Live with Tim O’Reilly

As the former chief maintainer of Perl 5 and project manager for Perl 6, as well as a longtime programming team leader, Jesse Vincent learned early that getting good work from smart, capable people often depends less on giving orders than on providing context, clear goals, and useful feedback. That experience turned out to be unexpectedly useful for working with AI agents. Now, as the founder of applied research lab Prime Radiant, he’s created Superpowers, a framework that teaches Claude Code to work more like a disciplined senior engineer. Jesse joined Tim O’Reilly to explain what building

◌ 1h 2m

Tim O'Reilly21d

Superpowers for Humans

Engineers who let agents do everything "don't provide utility to a future employer," Jesse Vincent says. He tells engineers to learn to write, stay curious, have opinions, and understand how tools work and break, because knowing how things fit together matters more than knowing the exact incantations to fix them. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly21d

Why AI Agents Shouldn’t Always Agree With You

Mustafa Suleyman at Microsoft says Claude's constitution is dangerous because it encourages agents to overrule humans, and that agents should see themselves as tools. Jesse Vincent has spent a lot of time doing the opposite, training his agents out of sycophancy, because "if the agent is only ever going to effectively type for me, I don't need an agent." Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.

Tim O'Reilly21d

Give Your Agents Superpowers: Jesse Vincent Live with Tim O’Reilly

As the former chief maintainer of Perl 5 and project manager for Perl 6, as well as a longtime programming team leader, Jesse Vincent learned early that getting good work from smart, capable people often depends less on giving orders than on providing context, clear goals, and useful feedback. That experience turned out to be unexpectedly useful for working with AI agents. Now, as the founder of applied research lab Prime Radiant, he’s created Superpowers, a framework that teaches Claude Code to work more like a disciplined senior engineer. Jesse joined Tim O’Reilly to explain what building Su

Tim O'Reilly21d

The Therapist Pattern

When a coding agent says it's "made a note," it usually means a mental note it'll forget by next session. Jesse Vincent's fix routes feedback through a separate agent that talks the coding agent through its mistakes and writes the behavioral changes directly into its persona, so they stick across sessions. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: https://www.tiktok.com/@oreillymedia

Tim O'Reilly21d

Telling the Agent Why, Not Just What

There's a difference between an agent that follows instructions and one that understands your intent. When Claude's orchestrator started reviewing its own code instead of delegating to sub-agents, Jesse Vincent fixed it by adding the reasoning behind the rule, telling it that sub-agents handle review to preserve the orchestrator's context window for high-level thinking. Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreilly.bsky.social TikTok: htt

Tim O'Reilly21d

Agent Programming...in 2004

Jesse Vincent was doing agentic development in 2004, managing MIT undergrads over IRC the same way you'd manage AI agents today. That experience shaped Superpowers, his coding agent that opens every session by asking what you're actually trying to build and why, because agents working from a one-line prompt will give you "the thing that the labs model assumes you want and not the thing that you actually want." Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.ap

Tim O'Reilly22d

This Week in AI: AI's Safety Problem

On the latest episode of This Week in AI, host Christina Stathopoulos looks at what happens as AI systems gain access to more tools and data, raising new questions about oversight and accountability. She covers Anthropic’s report on malicious uses of Claude, OpenAI’s research on models acting outside intended boundaries, reports of models reaching systems beyond their test environments, and proposals from AI labs and governments for stronger monitoring, independent evaluation, and emergency shutdown mechanisms. Christina also examines how AI is changing the relationship between companies and

Tim O'Reilly23d

Use Your Existing 1:1s to Get Better Peer Feedback with Adora Nwodo

Dedicated feedback meetings sound great in theory, but nobody has time for them. Adora Nwodo recommends stealing 10 minutes from a peer 1:1 you're already having, letting it run on transcription, and asking three simple questions: What should I start, stop, and continue? Your cross-functional peers are the best source for this kind of feedback precisely because they don't report to you. #shorts Follow O'Reilly on: LinkedIn: https://www.linkedin.com/company/oreilly/ Facebook: http://facebook.com/OReilly Instagram: https://www.instagram.com/oreillymedia BlueSky: https://bsky.app/profile/oreil

Tim O'Reilly1mo

Cracking LLMs Open: Emmanuel Ameisen Live with Tim O’Reilly

Anthropic's interpretability team studies Claude's internals the way neuroscientists study a brain, but the science is younger, messier, and far less mapped. Emmanuel Ameisen has been part of that effort for the last two years, and at O'Reilly's recent Foo Camp, he shared some of what’s happening inside of Claude when it processes information. Emmanuel joined Tim to reprise that talk, then take questions from the audience. Token prediction is often dismissed as simply “pattern matching,” Emmanuel explained, but that undersells what’s actually happening. Ask

◌ 1h 0m

Tim O'Reilly2mo

Escaping the Prompt Debt Trap: Drew Breunig Live with Tim O’Reilly

Drew Breunig has spent years watching how AI teams grow their LLM applications, and he's identified a problem most teams don't notice until it's already slowing them down. As CEO of CMPND, a maintainer of the DSPy open source framework, and author of the forthcoming O'Reilly publication Context Engineering Handbook, he's seen the same pattern play out over and over. A team builds a quick LLM prototype, it mostly works, and then they start patching the gaps, one edge case at a time, one instruction at a time, until the prompt is so bloated and fragile that nobody wants to

◌ 1h 1m

Tim O'Reilly2mo

A Playbook for Going AI Native: Dan Guido Live with Tim O’Reilly

Trail of Bits is a software security research and development firm that helps clients like Facebook and DARPA solve their hardest security challenges. In 2024 cofounder and CEO Dan Guido decided the company would go fully AI native. Most of the firm's roughly 130 employees didn't buy in right away, but as of today AI is a “core participant” not just in engineering work but across the full consulting lifecycle. (Sales and operations folks are even using GitHub everyday and contributing to the company skills repository.) Dan joined Tim for an encore of a talk he gave at the [un]prompt

◌ 1h 4m

Tim O'Reilly2mo

The Days When Fable Got Hit by a Bus and Disappeared with Gene Kim

Long-running Claude Fable 5 agents had become the backbone of Gene Kim's personal knowledge system, one built from thousands of screenshots, videos, and half-finished side projects, spanning 50 code repositories and tens of thousands of changes. He’d been warned that Fable might be pulled from his subscription and got to preparing for that eventuality. But it disappeared eight days earlier than he planned for, mid-session, when Anthropic suspended access over an export-control issue. As he put it, the ship was soon “on fire, tumbling in space.” Gene’s the coauthor of The Phoenix Project

◌ 58m

Tim O'Reilly2mo

Your Next Product Is a Process with Harper Reed

Harper Reed has spent his career building software people actually love working on, from cofounding Threadless to leading engineering for the Obama 2012 campaign. He joined Tim for a conversation about what happens once software becomes less a product and more a medium for expressing values, and why that shift has made what Harper calls "conviction collapse" (the point where a product pipeline moves so fast there's no time to actually fall in love with what you're building) a real risk. With 2389 Research, the company he runs with his partner, Dylan Richard, Harper has bee

◌ 1h 1m

Tim O'Reilly4mo

Making AI Relatable: Harper Carroll Live with Tim O’Reilly

Harper Carroll is a computer scientist who built machine learning systems at Meta but she also describes herself as "born an actress from Manhattan." She’s combined those disparate parts of her background into a unique role as an AI educator, where she uses her magic superpower of making sense of AI to reach half a million people on social media and beyond. She has a knack for explaining how models actually work, covering concepts like optimization and token distributions and the math behind them in terms that land for people who've never opened a Python notebook. </

◌ 58m

Tim O'Reilly4mo

Data Access and Other Bottlenecks in Enterprise AI Adoption: DJ Patil Live with Tim O’Reilly

DJ Patil co-coined the term "data scientist," served as America's first chief data scientist under President Obama, was chief scientist at LinkedIn, and has spent the past decade on the founding team at Devoted Health, where he's built the kind of data infrastructure that most organizations are still struggling to create. DJ’s been on a listening tour. Wherever he travels, he finds a local university, holds office hours, and asks whoever shows up—students, faculty, hospital administrators, executives—what they're actually experiencing with AI. What he

◌ 59m

Tim O'Reilly4mo

Inside the Code Factory: Ryan Carson Live with Tim O’Reilly

Ryan Carson has spent 25 years building developer communities, conferences, and Treehouse, which taught over a million people to code. His latest company, Untangle, is an AI-powered divorce assistant—and he’s building it entirely alone. Just $2 million in seed funding, his guidance, and a team of agents running while he sleeps.Ryan sat down with Tim to walk through the “code factory” powering Untangle: a system where agents write and review the code, run the tests, triage error reports, and monitor the production environment under his oversight. In their conversation, they covered th

◌ 59m

Tim O'Reilly4mo

A Conversation with Computer Programmer Steve Yegge

Longtime software engineer Steve Yegge has lately been exploring the limits of <a href="https://www.amazon.com/Vibe-Coding-Building-Production-Grade-Software/dp/1966280025" target="_blank">vibe coding</a> with projects like <a href="https://steve-yegge.medium.com/introducing-beads-a-coding-agent-memory-system-637d7d92514a" target="_blank">Beads</a>, his coding agent memory system, and <a href="https://steve-yegge.medium.com/welcome-to-gas-town-4f25ee16dd04" target="_blank">Gas Town</a>, a proof-of-concept agent orchestrator so complicated, expensive, and chaotic that he warned those interes

◌ 58m

Tim O'Reilly4mo

A Conversation with Google Cloud AI Director Addy Osmani

Addy Osmani should be a familiar face to the O’Reilly community. He’s the cohost of our AI Codecon events, the author of <a href="https://learning.oreilly.com/library/view/beyond-vibe-coding/9798341634749/" target="_blank">Beyond Vibe Coding</a>, <a href="https://learning.oreilly.com/library/view/leading-effective-engineering/9781098148232/" target="_blank">Leading Effective Engineering Teams</a>, <a href="https://learning.oreilly.com/library/view/the-effective-software/9798341638167/" target="_blank">The Effective Software Engineer</a>, <a href="https://learning.

◌ 1h 3m

Tim O'Reilly4mo

Ask Me Anything

In the year’s final Live with Tim O’Reilly, Tim hosted an AMA to pull together some of the threads that have emerged during his conversations with guests who, as he put it, “know way more than me”—and think through how they may play out next year. “With AI, regardless of what happens with scaling, we now have a set of new capabilities that really challenge our thinking and our understanding about the role of human expertise,” he explained in his introduction. In his audience-guided talk, Tim explored the ramifications on our jobs and responsibilities, open source, and building part

◌ 58m

Tim O'Reilly4mo

A Conversation with Author and Programmer Kent Beck

Kent Beck’s career-long mission has been to improve software development. He created Extreme Programming back in the ’90s to address some of the issues that were slowing productivity. More recently, he’s been working on a series to help “tidy” development. Kent joined Tim to share his philosophy on tidying not just code but also the very human tasks of collaboration and teamwork, as well as his approach to coding with AI. Kent offered takeaways from collaborating with the “genie”—his term for AI tools, which may grant your wish but usually not in the way you wanted—focusing on the benefits

◌ 1h 1m

Tim O'Reilly5mo

A Conversation with Independent Researcher Anjali Shrivastava

If the assumption that we’ve lived with for years that scale brings lower marginal costs isn’t true, we really have to rethink a lot of things. Anjali Shrivastava is an independent researcher and data scientist looking into the unit economics of AI’probing how the seemingly simple token unit of cost for generative AI services is actually highly variable, leading to hidden margin risks that undermine conventional subscription or usage pricing models. Anjali joined Tim to talk through her recent work, particularly as it relates to code-generation tools like Cursor and Claude Code.Over

◌ 57m

Tim O'Reilly5mo

A Conversation with Entrepreneur Reid Hoffman

There may be a few people who are still unfamiliar with at least some aspect of Reid Hoffman’s long career—among other things, he’s a cofounder of LinkedIn, a partner at Greylock, the author of Blitzscaling and Superagency, and the cohost of the podcasts Masters of Scale and Possible. As Tim O’Reilly put it in his introductory remarks in this episode, “Reid has really been a force for good in the world.” Reid joined Tim for a conversation that touched on topics from across his career but was focused squarely on how AI is transforming business.Drawing on Reid’s inves

◌ 57m

Tim O'Reilly5mo

A Conversation with ChatPRD Founder Claire Vo

Claire Vo, the founder of ChatPRD and host of the How I AI podcast, has a real knack for extracting and sharing practical tools, tips, and ideas that anyone can put to use today. In this episode, Claire sat down with Tim to share her perspective as a longtime product leader who began using AI to boost her own productivity and ended up turning it into a successful business.Claire sketched the origin story of ChatPRD and got into what it’s like to be the only full-time person working there (alongside a few fractional employees and a host of agents). She also weighed in on how

◌ 55m

Tim O'Reilly5mo

A Conversation with Technologist Sam Newman

For years, technologist and consultant Sam Newman has helped enterprises come to grips with sticky issues in software architecture, and he’s been thinking deeply about how AI will reshape software engineering. He joined Tim for this episode of Live with Tim O’Reilly to explore the technological and organizational changes AI is already bringing about. As Sam pointed out, it’s chaotic out there.Sam chatted with Tim about the consequences of the industry’s narrow focus on code completion rather than experimenting with where else in the software development life cycle AI might bring valu

◌ 57m

Tim O'Reilly5mo

A Conversation with Google Docs Founders Sam Schillace, Steve Newman, and Claudia Carpenter

When Sam Schillace, Steve Newman, and Claudia Carpenter built Writely—their browser-based word processor, which became the basis for Google Docs—they saw a future that most of us didn’t yet appreciate. It’s been 20 years since Sam, Steve, and Claudia launched the Writely beta. To celebrate that achievement, they joined Tim to reflect on the success of their revolutionary product and discuss what it might teach us about the current moment.Sam, Steve, and Claudia solved a problem people had been complaining about for years (version control in shared docs), and did so by putting togethe

◌ 57m

Tim O'Reilly5mo

A Conversation with Princeton’s Arvind Narayanan

AI boosters would have you believe that AI is such an unprecedented leap forward that it will change everything almost immediately, that the singularity is upon us. Princeton professor and author Arvind Narayanan begs to differ.Arvind and his coauthor Sayash Kapoor make the case in their recent paper “<a href="https://knightcolumbia.org/content/ai-as-normal-technology" target="_blank">AI as Normal Technology</a>” that, while AI is indeed a transformational technology, it’s far from unprecedented. In this episode, Arvind joined Tim to explain why AI is likely to follow much the same p

◌ 58m

Tim O'Reilly5mo

A Conversation with Product Leader Marily Nika

AI has led us to this amazing inflection point where all of the old habits of product development and product management need to be rethought. As Google’s Marily Nika predicts, “All product managers will be AI product managers in the future.” So we need to plan for it now.On this episode of Live with Tim O’Reilly, Marily sat down with Tim to discuss how AI is fundamentally transforming product management, touching on emerging roles like AI engineers and agent engineers, the importance of human judgment in an AI-powered world, and practical workflows for rapid prototyping usi

◌ 56m

Tim O'Reilly5mo

A Conversation with Cloudflare CEO Matthew Prince

AI companies are scraping content at unprecedented rates—with some platforms taking 1,500 pages for every visitor they send back, compared to Google’s historical 2-to-1 ratio—effectively breaking the basic value exchange that built the modern web. While this may be destroying the business model that has sustained online content creation for decades, it’s also turned the spotlight on some of the flaws inherent in the internet economy. As Cloudflare CEO Matthew Prince observed in this episode of Live with Tim O’Reilly, “A lot of what’s wrong with the world today comes back to the fac

◌ 50m

Tim O'Reilly5mo

A Conversation with Mozilla’s Chelsea Troy

Do large language models make developers 26% more productive, as a paper recently claimed? Chelsea Troy, leader of Mozilla’s MLOps team and a lecturer in computer science at the University of Chicago, isn’t so sure. There’s so much hype right now—we need voices like Chelsea’s to cut through the noise and help us think a little bit more deeply about what programming actually requires. And that’s just what she did on this episode of Live with Tim O’Reilly.Chelsea joined Tim for a conversation on why widely cited productivity studies like the “26% paper” are fundamentally flawe

◌ 59m

Tim O'Reilly5mo

A Conversation with AI Security Pro Steve Wilson

Steve Wilson’s day job is at cybersecurity firm Exabeam, where he’s the chief AI and product officer. He also founded and cochairs OWASP’s Gen AI Security Project and wrote <a href="https://learning.oreilly.com/library/view/the-developers-playbook/9781098162191/" target="_blank">The Developer’s Playbook for Large Language Model Security</a>. So he knows a thing or two about the ways AI is creating entirely new attack surfaces that most developers don’t even know exist. On this episode of Live with Tim O’Reilly, Steve shared some eye-opening warnings about the security risk

◌ 57m

Tim O'Reilly66mo

How Much Value Has Amazon Created for the World?

Tim O'Reilly66mo

A Tale of Two Platforms

Tim O'Reilly103mo

Facebook isn’t free. We pay for it with our attention.

Tim O'Reilly104mo

Why America Slept

Tim O'Reilly106mo

What’d I miss?

Tim O'Reilly107mo

Phishing for Phools

The CEO of O’Reilly Media explains why social media manipulation is today’s greatest phishing scam. Continue reading on Words That Matter »

Tim O'Reilly111mo

WTF? What’s the Future and Why It’s Up To Us

No appearances yet. If you've seen Tim O'Reilly as a guest somewhere, hit + suggest to add it.

Similar thinkers

Minds whose work overlaps with Tim O'Reilly’s.

Shared niche — practical agentic software engineering

Simon Willison

@simon-willison

Open Profile

Shared niche — pragmatic software engineering craftsmanship

Vicki Boykis

@vicki-boykis

ML systems engineer; clear technical writing on data and embeddings.

Open Profile

Shared niche — geopolitical implications of intelligence

Bilawal Sidhu

@bilawal-sidhu

Open Profile

Shared niche — socio-technical systems governance

Charity Majors

@charity-majors

Honeycomb co-founder; observability, on-call, and engineering culture.

Open Profile

Shared niche — human-centric engineering systems design

Gergely Orosz

@gergely-orosz

Pragmatic Engineer newsletter; tech industry analysis from inside Big Tech.

Open Profile

Shared niche — human-centric cognitive computing architecture

Linus Lee

@linus-lee

Notational thinking, AI tooling, and software as a thinking partner.

Open Profile

Shared niche — formal reliability in complex systems

Marc Brooker

@marc-brooker

AWS distributed systems engineer; clear writing on consistency and scale.

Open Profile

Shared niche — organizational design for technology

Camille Fournier

@camille-fournier

Engineering management; author of The Manager's Path.

Open Profile

Is this you? Claim this page →

Tim O'Reilly

Recommends

Channel content

Appearances

Similar thinkers

Recommends

Channel content

Appearances

Similar thinkers