Breakeven Brothers

The Breakeven Brothers are back from hiatus with a huge announcement: one of the brothers has started a new role as a software engineer at OpenAI! He shares the inside scoop on the interview process, the dynamic culture, and what it’s like to be working at the epicenter of the AI revolution. This personal journey kicks off a broader conversation about career development in the age of AI, with insights from a career panel at a local college and advice for accountants to embrace new tools like Excel and the command line.

The conversation then dives headfirst into the viral whirlwind surrounding OpenClaw (formerly Moltbot/ClawdBot). The brothers dissect the massive hype, the very real security concerns, and the bizarre phenomenon of AI bots seemingly conspiring on forums. They also tackle the growing problem of AI-generated "slop" flooding open-source projects, the burden it places on maintainers, and the recent trend of AI development tools shifting from complex command-line interfaces back to user-friendly graphical apps. It's a candid and grounded discussion on separating genuine innovation from the noise in today's fast-moving tech landscape.

Chapters:
00:09 Introduction and Recent Life Changes
00:40 Brad Announces New Job at OpenAI
04:35 Ben's Accounting Career Panel Experience
07:29 Ben's New Claude Code YouTube Series
15:39 The Rise and Hype of OpenClaude
28:19 AI Slop Burdening Open Source Projects
33:13 The Return of Graphical User Interfaces
37:59 Bookmarks: Improving Claude and Building Drones

Links:
- Augmentic Accounting YouTube Channel
- OpenClaw
- Mitchell Hashimoto's X/Twitter Profile
- Real Python: Quantum Computing With Qiskit 

#OpenAI #OpenClaw #AIHype #CareerChange #SoftwareEngineering #AccountingTech #OpenSource #ClaudeCode #TechPodcast #AItools

Creators & Guests

What does the future hold? Before diving into some wild predictions, we recap a whirlwind week of AI meetups in San Francisco. We share key takeaways from the Agents Anonymous meetup at Sentry's HQ, including a fascinating talk from a Cursor engineer on solving the "long-running agents" problem and what it took to build a web browser autonomously. Then, we switch gears to the Laravel community, discussing Taylor Otwell's demo of a new first-party AI package and the unique vibe of builders who wait for tech to be proven before delivering powerful tools.

With our minds on the future, we lock in our 2026 predictions with a 4x4 bingo card. We're placing our bets on everything from full-length AI-generated films and humanoid robots in high-income homes to the first $1 trillion AI IPO and the rise of a $10,000-a-month AI subscription plan. Will UBI become a serious midterm topic? Will Apple finally make a huge AI comeback? Will we see the reintroduction of human typing (typos and all) as a signal of authenticity? We cover 16 bold, specific, and sometimes scary predictions that paint a vivid picture of the next two years in tech and society.

Chapters:
00:08 Recapping Recent San Francisco AI Meetups
01:06 Cursor and the Long-Running Agents Problem
04:30 Laravel's New Official AI Package
10:57 Introducing the 2026 Predictions Bingo Card
11:27 Predictions: AI Media and Humanoid Robots
16:12 Predictions: AI Operating Systems and Assistants
19:27 Predictions: Economic Shifts and Market Competition
23:05 Predictions: Security, Society, and Hardware
27:38 Predictions: The Future of AI Coding
38:34 Predictions: Market Recessions and Human Typing
43:02 Full Bingo Card Recap and Bookmarks

Links:
- Creating a web browser with long running agents
- Callstack's React Native Skill for Claude
- Indie Dev Dan on YouTube
- Ben Affleck on Joe Rogan discussing AI in movies

#AIPredictions #TechPodcast #ClaudeCode #Laravel #HumanoidRobots #FutureOfAI #Cursor #SanFrancisco #UBI #AICommunity

Creators & Guests

Welcome to 2026! The new year kicks off with shocking news from the web development world as industry darlings Tailwind Labs and Laracasts announce massive layoffs, citing the disruptive force of AI. Brad and Bennett dive deep into the drama, analyzing the GitHub comments and tweets that sent shockwaves through the community. They discuss the harsh reality for businesses built on selling code or educational content in an era where advanced AI can generate both in seconds, connecting the dots between these high-profile stumbles and the broader economic shifts happening across the tech landscape.

The conversation then pivots from industry analysis to personal workflows and the tools defining the new era of development. Brad shares his recent, surprising switch from the much-loved Claude Code to OpenAI's Codex, detailing the specific prompts and settings that are unlocking new levels of productivity. The brothers explore the rise of Agent SDKs, the power of community-created 'Skills,' and the emerging, sometimes wild, use cases for AI, from analyzing personal DNA data to automating tax preparation. This episode is a crucial look at the tangible impacts of AI, offering both a warning and a guide for navigating the rapidly changing world of technology in 2026.

Chapters:
00:00 Welcome to the First Episode of 2026
01:09 AI's Toll on Tailwind and Laracasts
11:19 Why 2026 Is the Year of Agents
14:47 Building Practical AI Agent Workflows
22:30 AI's Role in Accounting and Taxes
27:16 Creative Uses for Personal AI Agents
36:15 Codex vs. Claude Code for Development

Links:
- Tailwind CSS `llms.txt` Pull Request
- Adam Wathan on X (Twitter)
- Laracasts
- TaxDome
- Peter Steinberger: Shipping at Inference Speed
- Boris Cherny's Tweet on the Future of Engineering
- Levelsio on X (Twitter)

#AI #TechPodcast #SoftwareDevelopment #TailwindCSS #FutureOfWork #ClaudeCode #OpenAICodex #Laracasts #DeveloperTools #AgenticWorkflows

Creators & Guests

In our final episode of 2025, we're taking a look back at what was dubbed 'the year of the agents' and peering into what 2026 might hold. The conversation kicks off with a deep dive into a potential paradigm shift: the rise of AI 'skills.' We explore Anthropic's big push for this new framework, contrasting it with existing tool-calling and MCP server approaches. Are skills just a more user-friendly way to package domain knowledge, or are they a fundamental change in how we'll build with AI? We break down the pros and cons, from context window preservation to ease of sharing in a marketplace.

Beyond the theoretical, we get into the latest industry drama and developments. We tackle the heated online debate surrounding Claude Opus 4.5—is Anthropic secretly downgrading its flagship model, or are user expectations simply rising? We also cover the release of Google's impressively fast Gemini 3 Flash, a major NPM supply chain attack that compromised thousands of repositories, and the exciting new visual editor and debug mode features from Cursor that could change the front-end development workflow. Join us as we wrap up the year with the biggest stories and share our favorite resources to take into the new year.

Chapters:
00:00 End of Year and AI Music Licensing
02:43 Anthropic and OpenAI Introduce "Skills"
12:01 The Opus 4.5 Downgrade Controversy
19:14 Google's New Gemini 3 Flash Model
21:50 Recent Supply Chain Attack on Developers
26:39 Cursor's New Visual Editor and Debug Mode
31:11 Bookmarks: Building Skills and Email Triage
34:15 Wrapping Up 2025 and Holiday Plans

Links:
- Suno AI: Ben mentions he used Suno AI, an AI music generation tool, to create the podcast's current theme song and pays a monthly subscription to maintain the commercial license.
- Highcharts: A paid JavaScript charting library mentioned as an example of a tool with a commercial license that operates on an 'honor system,' similar to Suno AI.
- Chart.js: The free, open-source alternative to Highcharts that Brad switched to for his project, Split My Expenses.
- Cursor IDE: The AI-first code editor discussed for its new features, including a visual editor for front-end development and an automated debug mode.
- Anthropic: Don't Build Agents, Build Skills Instead: Brad's bookmark: An Anthropic presentation on the AI Engineer YouTube channel titled 'Don't Build Agents, Build Skills Instead,' which argues for a shift in how developers leverage AI.
- LangChain YouTube Channel: Ben's bookmark: A deep-dive video from LangChain's YouTube channel about building an agent for email triage, a practical application of AI for managing high-volume inboxes.

#AIPodcast #Anthropic #AISkills #SoftwareDevelopment #Cybersecurity #CursorIDE #AIAgents #GeminiAI #TechNews #LangChain

Creators & Guests

The Breakeven Brothers are back after the holiday break, sharing stories from Thanksgiving travels and post-holiday shopping. Bradley kicks things off with a major announcement: the SplitMyExpenses Android app is finally live! He shares the challenging, and often frustrating, journey of navigating the Google Play Store for the first time, contrasting it with his experience on iOS and celebrating the milestone of being available on all major platforms.

From there, the conversation dives headfirst into the whirlwind of recent AI developments. Just days after their last recording, Anthropic released the incredibly powerful Claude Opus 4.5, and the brothers break down why it's a potential game-changer, especially for developers using tools like Cursor. They also explore Google's stunningly realistic new image model: Nano Banana Pro and discuss OpenAI's swift response with Codex Max. This rapid-fire competition leads to a discussion about OpenAI's rumored "code red" to refocus efforts on ChatGPT amidst user decline, raising questions about who will ultimately win the AI race.

Chapters:
00:09 Introduction and Thanksgiving Recap
02:12 Black Friday Shopping and AI Try-Ons
04:10 Launching the SplitMyExpenses Android App
10:31 Google's New Gemini 3 Pro Model
17:39 Nano Banana Pro's Realistic Images
22:11 OpenAI's Agentic Codex Max Model
24:52 Anthropic's Incredible Opus 4.5 Release
38:48 Anthropic's Bun Acquisition and IPO Rumors
43:33 OpenAI Declares a 'Code Red'

Links:
- SplitMyExpenses on Android
- SplitMyExpenses on iOS
- Bradley Bernard's Blog
- Back Market
- Google's SynthID Watermark
- The Boring Marketer

#AI #TechPodcast #Claude4.5Opus #GoogleImagen #OpenAI #AndroidDev #IndieHacker #SoftwareEngineering #Anthropic #DeveloperTools

Creators & Guests

Brad is back from the grind and finally has something to show for his 4 AM coding sessions: the official SplitMyExpenses iOS app is live on the App Store! He shares the entire journey, from the initial manual coding attempts to leveraging AI tools like Claude Code that dramatically accelerated his development pace and expanded the app's feature set. Ben also dives into his own coding project, detailing how he rebuilt his accounting firm's website from scratch, ditching Squarespace for a custom solution powered by Cursor's new "plan mode." The brothers compare notes on how indispensable AI has become in their workflows, feeling almost handicapped without it.

The conversation then zooms out to the broader AI landscape, dissecting the pros and cons of different models and why Ben finally decided to cancel his Claude subscription. With hype reaching a fever pitch, they speculate on the imminent release of Google's Gemini 3.0, discussing what it needs to deliver to be a true game-changer in terms of intelligence, speed, and cost. To wrap up, they explore a fascinating and potentially alarming theory from "The Big Short's" Michael Burry about an AI bubble fueled by misleading accounting practices, leaving listeners to ponder if the current AI boom is built on a solid foundation or a house of cards.

Chapters:
00:10 Catching Up and New Haircuts
02:05 Brad Launches Split My Expenses App
05:32 How AI Transformed App Development
15:36 Ben Rebuilds His Website with AI
20:50 Deep Dive into AI Coding Tools
28:49 Anticipating Google's Gemini 3.0 Release
34:28 Bookmarks: AI Bubbles and iPhone Socks

Links:
- SplitMyExpenses on the iOS App Store
- Catalyst CFO Website
- Render Hosting
- Cursor - The AI-first Code Editor
- Seeking Alpha: Michael Burry and the AI Bubble
- Cursor Presentation: How We Trained Our Model

#AIPodcast #TechPodcast #IndieDev #ReactNativeDev #CursorAI #Gemini3 #AIbubble #SoftwareEngineering #AppLaunch #SideHustle

Creators & Guests

The Breakeven Brothers are finally back! After another hiatus filled with cross-continental travel, surprise kitten fostering, and a whole lot of life, Bradley and Bennett jump right back into the deep end. We kick things off with stories from Brad's trip to Korea—navigating rainy weddings and constant hotel hopping—and Bennett's unexpected turn as a foster dad to a tiny kitten that wandered into his backyard. These adventures set the stage for a much larger conversation about how quickly things can change, especially in the world of AI.

Once we're settled, we dive into a wild new technique from DeepSeek that turns text into images just to... turn it back into text? It sounds crazy, but this 10x compression trick could fundamentally solve one of AI's biggest frustrations: why our agents seem to get dumber the longer we talk to them. This sparks a huge debate about OpenAI's new browser, ChatGPT Atlas. Is it a revolutionary agent that will plan our trips and buy our groceries, or a massive security risk waiting to happen? Brad also gives us a firsthand report from an exclusive "Claude Code Anonymous" meetup in San Francisco, sharing the secrets behind a 700-line master prompt that gets near-perfect results. Tune in for a packed episode covering everything from AI walled gardens to the future of how we browse the web.

Chapters:
00:00 Back from Hiatus & Personal Updates
04:11 DeepSeek's Text-to-Image Compression
13:15 OpenAI's New Agentic Browser: ChatGPT Atlas
20:38 Security Concerns with Agentic Browsers
33:20 Takeaways from Claude Code Anonymous in SF
40:16 Bookmarks: Meta vs. AI & Cursor's Plan Mode
43:54 Conclusion & wrap up

Links:
- Meta Banning Rival AI Chatbots on WhatsApp
- Cursor's New "Plan Mode"

#AI #TechPodcast #OpenAI #ChatGPTAtlas #DeepSeek #ClaudeAI #AICoding #ContextWindow #AIAgents #BrowserWars

Creators & Guests

For our dirty 30, we're unpacking some of the biggest AI news of the year, coming at you fresh on the same day it dropped. OpenAI just held their DevDay, and to say they dropped a bombshell would be an understatement. We’re giving our live, unfiltered reactions to a tidal wave of new releases that genuinely knocked it out of the park. From an App Store-like ecosystem built directly into ChatGPT to some wild new revenue estimates (is a billion dollars a month even real?), OpenAI is building a platform that’s impossible to ignore.

Join us as we dive into the game-changing tools that could reshape the AI landscape. Ever wanted to build a powerful, custom agent without writing a single line of code? We break down the new Agent Builder and the live demo where they built and deployed one in under eight minutes. Plus, we were blown away by what might be the coolest live tech demo we've ever seen, involving stage cameras, lighting, and an Xbox controller, all prompted into existence on the spot. We also cover the API releases for Sora 2 and GPT-5 Pro and discuss what this all means for developers, non-technical creators, and the future of building.

Chapters:
00:09 Introduction: OpenAI's Big DevDay
01:32 High-Level DevDay Announcements
06:52 Building Apps within ChatGPT
18:49 The No-Code Agent Builder Kit
27:49 Sora 2 & GPT-5 Pro API Updates
35:13 The Impressive Live Codex Demo
45:44 Final Thoughts & Funny Bookmarks
50:38 Conclusion

Links:
- Sunny Madra's Tweet on OpenAI's Revenue
- OpenAI DevDay Opening Keynote
- Kitze Tweet about MCP
- Sora 2 Video of Sam Altman stealing Studio Ghibli

#OpenAI #DevDay #ChatGPT #AgentKit #Codex #Sora2 #GPT5Pro #NoCodeAI #AIPodcast #TechReaction

Creators & Guests

The Brothers are back! After a much-needed month-long break filled with wedding anniversaries, unfortunate illnesses, and a trip to Hawaii, Bradley and Bennett return to find the AI world has moved at lightning speed. They waste no time diving headfirst into a tool that has defined white-collar work for decades: Excel. With Anthropic's Claude now able to work directly with spreadsheets, is this the game-changing moment accountants have been waiting for, or is it merely a "Band-Aid" on a workflow that's about to be made obsolete by autonomous AI agents? Bennett puts on his tinfoil hat for a bold prediction that could change the future of knowledge work forever.

The debate then shifts from the spreadsheet to the command line, where a civil war is brewing in the world of AI coding assistants. After a "golden period" where Claude Code was the undisputed champion, a recent dip in performance has opened the door for a revitalized competitor: OpenAI's Codex. Bradley breaks down why developers are losing faith in Claude and flocking to its rival, exploring the high-stakes showdown between the two platforms. From groundbreaking new features to scary security vulnerabilities and our predictions for the rest of the year, this episode is a jam-packed catch-up on everything you need to know in the world of AI.

Chapters:
00:09 Introduction: Catching Up
02:36 Claude's New Excel Integration
06:31 Hot Take: Is Excel Obsolete?
12:05 The Rise of Autonomous AI Agents
18:35 Adapting to an AI-Driven Workforce
27:16 Microsoft Copilot Integrates Anthropic Models
32:56 The Claude Code vs. Codex Debate
48:17 AI Predictions & Security Risks
54:58 Conclusion

Links:
- Bennett's YouTube Channel: Augmented Accounting
- Automate the Boring Stuff with Python by Al Sweigart
- Python for Finance by Yves Hilpisch
- Microsoft & Anthropic Partnership (thedecoder.com)
- Google Calendar OpenAI ChatGPT MCP Server Vulnerability

#AI #Excel #FutureOfWork #ClaudeCode #CodexCLI #AccountingTech #DeveloperTools #Programming #TechPodcast #Anthropic

Creators & Guests

A bombshell MIT study just dropped, claiming 95% of companies get ZERO return on their AI investments. Is the hype bubble finally bursting, or is there a deeper story the headlines are missing? In this episode, we're calling BS on the doom-and-gloom narrative. Drawing from our own in-the-trenches experience building an app with AI, we break down why these studies might be flawed and reveal the massive "learning gap" that's holding most people back from unlocking AI's true potential. It's not the models that are failing; it’s how we're being taught to use them.

The AI drama doesn't stop there. We're unpacking all the latest chaos from Silicon Valley, from Meta slamming the brakes on its billion-dollar hiring spree to the shocking rumor that Apple might actually use Google's Gemini to power the next generation of Siri. We also dive into a new trend for developers: can you really monetize your expertise by selling $9 AI rules? We debate that and the growing battle against "AI slop" in open source before tackling a big question for the professional world: are junior accountants about to start doing a manager's job right out of college? Tune in for our unfiltered take on the hype, the reality, and where we're all headed.

Chapters:
00:00 Introduction
01:37 The Reality of AI Investment ROI
14:58 Monetizing Expertise by Selling Cursor Rules
23:19 Meta's Billion-Dollar AI Hiring Freeze
34:27 Apple Considers Google Gemini for Siri
43:34 Disclosing AI-Generated Code Contributions
51:06 How AI is Reshaping Junior Roles
1:05:54 Bookmarks: AI Workflows and Scrapers
1:12:49 Wrapping Up

Links:
- SplitMyExpenses
- Axios article
- Cursor rules
- Laravel Boost
- Ghostty Terminal
- PwC Article on the Future of Junior Accountants
- Peter Steinberger's "AI Dev Workflow" Blog Post
- Report Reveals Internal Chaos Behind Apple's Siri Failure - MacRumors
- Apple Explores Using Google Gemini AI to Power Revamped Siri
- Marc Colbrugu's Tweet on Cloudflare

#TechPodcast #AINews #AIROI #SoftwareEngineering #MetaHiringFreeze #FutureOfWork #AppleGemini #AICodingTools #Productivity #DeveloperPodcast

Creators & Guests