Why these AI updates could change your job forever
Download MP3Hold onto your hats, because the AI world just went into hyperdrive! This week, we try to make sense of the absolute deluge of new model releases that have dropped faster than Brad can guess Ben's secret words in our new AI Word Nerd game. From OpenAI finally cracking the code on actually usable text in images (goodbye, alien script!) to Google's confusingly named but seriously powerful Gemini 2.5 Pro storming the leaderboards for coding and reasoning, it feels like a year's worth of updates landed in just two weeks. We'll break down the key players, including the latest from open-source champions like DeepSeek and Sesame, covering advancements in speech-to-text, text-to-speech, and even AI that can see. If you're feeling overwhelmed by the pace, you're not alone – join us as we navigate the flood!
But it's not all just news and benchmarks! Ben dives deep into his latest "vibe coding" experiment: building a personal AI agent to tame his unruly email inbox. Hear about the highs and lows of prompting an AI to sort, prioritize, and even connect with other tools like QuickBooks, all while trying not to break the API limit bank. Discover the quirks of working with these powerful models, the surprising difficulty of asking an AI to remove code it added, and the critical lessons learned from tinkering on the front lines. If you want practical insights into building useful AI tools and staying ahead of the curve without needing a PhD, this episode is packed with real-world experience and essential updates.
Links:
- Google AI Studio
- OpenAI Text-to-Speech Demo
- Sesame Text-to-Speech Model
- OpenAI Agent Building Framework
- Eric's Context Window Chart & RepoPrompt
- Santiago's MCP Tweet/Video
But it's not all just news and benchmarks! Ben dives deep into his latest "vibe coding" experiment: building a personal AI agent to tame his unruly email inbox. Hear about the highs and lows of prompting an AI to sort, prioritize, and even connect with other tools like QuickBooks, all while trying not to break the API limit bank. Discover the quirks of working with these powerful models, the surprising difficulty of asking an AI to remove code it added, and the critical lessons learned from tinkering on the front lines. If you want practical insights into building useful AI tools and staying ahead of the curve without needing a PhD, this episode is packed with real-world experience and essential updates.
- (00:00) - Introduction
- (06:01) - Recent Developments in AI Models
- (10:25) - OpenAI's Image Generation Innovations
- (13:19) - Google's Gemini 2.5 Release
- (17:47) - DeepSeek's New Model and Open Source Impact
- (18:58) - Advancements in Speech-to-Text Technology
- (21:23) - Innovations in AI: Leading the Charge
- (28:27) - Exploring New AI Models and Their Applications
- (38:29) - Building AI Agents: Insights and Experiences
Links:
- Google AI Studio
- OpenAI Text-to-Speech Demo
- Sesame Text-to-Speech Model
- OpenAI Agent Building Framework
- Eric's Context Window Chart & RepoPrompt
- Santiago's MCP Tweet/Video
#AIUpdates #GoogleGemini #OpenAI #AIAgents #AICoding #VibeCoding #OpenSourceAI #AIProductivity #TechPodcast #ContextWindow
Creators and Guests
Host
Bennett Bernard
Mortgage Accounting & Finance at Zillow. Tweets about Mortgage Banking and random thoughts. My views are my own and have not been reviewed/approved by Zillow
Host
Bradley Bernard
Coder, builder, mobile app developer, & aspiring creator. Software Engineer at @Snap working on the iOS app. Views expressed are my own.
