Send us Fan Mail

Be part of the show: https://btb-listener-line.fly.dev/
Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

In this episode of Beyond the Build, we talk about how personal AI agents are starting to change the way we interact with technology, including research showing that some agents may recommend more expensive options based on what they infer about a user.

We also get into the privacy tradeoffs of personal AI, OpenAI Dots vs. Meta Muse, whether Plan Mode still makes sense for coding agents, why taste still matters when building with AI, browsers sending gRPC requests directly, and what it actually takes to identify and push high-leverage engineering work inside an organization.

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

Send us Fan Mail

Be part of the show: https://btb-listener-line.fly.dev/
Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

This week on Beyond the Build, we break down some of the biggest stories in software engineering, cybersecurity, AI, and tech.

We discuss Jev, a new decision-focused AI model, the shifting AI talent pipeline between China and the US, AI hacking incidents, Kimi coming to AWS Bedrock, the reported FBI breach, North Korean fake developer interviews, Polymarket fraud, Samsung smart fridge failures, and the debate around whether MCP is still necessary.

We also talk about senior engineer communication, how coding agents edit files, and the growing problem with AI-generated writing.

If you’re a software engineer or tech professional trying to keep up with what’s changing across the industry, this episode is for you.

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

00:00 Trailer
01:36 Intro — Last Week in Tech
01:50 Prediction Markets, Fraud & Polymarket
06:29 Kimi AI Comes to AWS
10:41 The AI Slowdown & Management Consultants
15:05 Jev: The AI Model Built for Fast Decisions
17:47 Slop Mop & The Problem With AI Writing
21:44 The Changing Landscape of Chinese AI Talent
26:56 AI Models Are Hacking Computers
30:27 Is MCP Still Necessary?
36:45 ShinyHunters Hack the FBI
39:46 When a Software Update Breaks Your Fridge
42:43 North Korean IT Workers & Deepfake Job Interviews
45:40 Rapid Fire: The Senior Engineer Death Spiral
48:03 Managing Capacity, Delegation & Engineering Work
50:13 How Coding Agents Edit Your Files
51:23 Should AI Be Writing Your Documents?
55:49 Wrap Up
56:17 A New Way to Get Your Questions on the Show

Send us Fan Mail

Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

In this episode of Beyond the Build, we break down Anthropic’s latest AI misuse report and look at how bad actors are already using AI for scams, surveillance, cyber operations, influence campaigns, telecom data scraping, and other real-world attacks.

We also discuss the growing AI arms race between companies like Anthropic, OpenAI, and other major labs. AI companies are warning about models becoming harder to supervise while simultaneously investing billions to build more capable systems. Can any company realistically slow down if everyone else keeps racing ahead?

We also get into a new security problem emerging around AI coding agents and plugins. As developers give tools like Claude Code, Codex, GitHub Copilot, and Gemini CLI more access to repositories, terminals, APIs, and credentials, supply-chain attacks become significantly more dangerous.

Then we debate what code review should look like in the age of AI. If AI can generate code faster than humans can read it, should every line still be reviewed? Or does code review become less about catching bugs and more about making sure engineers still understand the systems they are building?

We also cover AWS infrastructure failures during geopolitical conflict, Windows 11 updates breaking hardware, OpenAI’s IPO plans, digital ownership, and what happens when the software and infrastructure we depend on become increasingly automated.

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

Send us Fan Mail

Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

In this episode of Beyond The Build, we talk about the growing wave of AI safety warnings coming from researchers at companies like OpenAI, Anthropic, and Google DeepMind, and whether those concerns should actually change how seriously we think about the risks of increasingly capable AI systems.

We also get into the security problems that come with AI agents having access to production secrets, private data, credentials, and external systems, along with recent controversies around AI research, watermarking, browser privacy, and the trust we place in the tools we use every day.

Then we bring it back to software engineering. If AI can already investigate alerts, write code, open pull requests, and automate more of the development lifecycle, what is actually left for engineers? We discuss why coding is only one part of the job, why judgment, system design, debugging, operations, and accountability still matter, and why you cannot fully delegate a system you do not understand.

Along the way, we also touch on AI-assisted research, security vulnerabilities, prompting, brownfield vs. greenfield development, Postgres at scale, expired credit card security, and a few other things happening in tech this week.

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

Send us Fan Mail

Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

AI is changing software security faster than most engineering teams are prepared for.

In this episode of Beyond the Build, we break down how AI agents are starting to discover and exploit vulnerabilities, why the window between a CVE being disclosed and actively exploited is shrinking, and what that means for developers, security teams, and companies shipping software today.

We also discuss new findings around OpenAI’s AI agent swarms, concerns around observability and model behavior, major data breaches involving identity verification providers, and why digital privacy is becoming even harder to protect.

Beyond security, we get into how AI is reshaping software engineering itself, including the impact on frontend developers, Meta’s push toward smaller engineering teams, local LLMs running directly in the browser with WebGPU, and why human error can still take down even the most resilient cloud infrastructure.

Topics covered:
• AI agents and autonomous hacking
• OpenAI swarm security concerns
• CVEs and faster vulnerability exploitation
• Software supply chain and cybersecurity risks
• Data breaches and identity verification
• Privacy and age verification
• The future of frontend engineering
• AI-driven software development
• Meta engineering layoffs and productivity
• WebLLM and local AI inference
• Google Cloud outages and infrastructure resilience

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

Send us Fan Mail

Ask us a question or suggest an Episode: https://forms.gle/8tEg2CKTNNWteSgt9

This week on Last Week in Tech, we’re talking about what happens when memory gets expensive and software can’t afford to stay bloated.

RAM prices are climbing fast, Android is tightening memory limits, and companies like Cloudflare are squeezing huge amounts of memory out of their systems through better data structures and smarter engineering. The bigger question is whether memory efficiency is becoming a functional requirement again.

We also cover OpenAI cutting Cursor off after its acquisition by SpaceX AI, Nvidia’s reported move for Hugging Face, Stripe acquiring OpenRouter, why the harness around an AI model may matter more than the model itself, and whether AI is slowly eroding the fundamentals developers used to learn through experience.

Plus, we get into developer security hygiene, Chrome privacy, the growing number of CVEs, and Meta’s massive settlement over teen social media usage.

If you’re a developer, this episode is really about one broader shift: the constraints are coming back, and they may force us to build better software.

Connect with Ope - https://www.linkedin.com/in/osunkoyaope/

Connect with Uma:
Instagram: (https://www.instagram.com/umacodes)
Tiktok: (https://www.tiktok.com/@umacodes)
Twitter: (https://twitter.com/umacodes)
LinkedIn: (https://www.linkedin.com/in/umaabu/)

Follow us on Social Media
Instagram: (https://www.instagram.com/beyondthebuildpodcast)
TikTok: (https://www.tiktok.com/@beyondthebuildpodcast)

Send us Fan Mail

In this episode of Beyond the Build, we break down some of the biggest AI and software engineering stories from the week, including Anthropic’s explosive revenue growth, Cursor’s new Git hosting platform, Lovable moving away from Next.js, and the growing cost of AI inside companies.

We discuss why Anthropic’s reported $65B annualized revenue run rate is impressive but difficult to compare directly with OpenAI, and whether today’s AI valuations are sustainable as inference costs remain high.

We also look at Cursor Origin, a new Git hosting platform built for the AI coding era, after GitHub experienced major scaling pressure from rapidly increasing commit volume. Then we cover Lovable’s migration from Next.js to TanStack, why the company chose to rebuild its own product using its own platform, and what that says about the future of modern web development.

Other topics include Google buying Spirit Airlines’ operational data for AI training, AI agents accidentally DDoSing services and interfering with each other, questions around legal accountability for autonomous agents, AT&T processing 45 billion AI tokens per day, the rise of “AI Didn’t Read,” and how companies are starting to measure whether their AI spending is actually creating value.

If you’re interested in AI coding tools, Cursor, GitHub, Anthropic, Claude, Lovable, Next.js, TanStack, AI agents, software engineering, developer tools, and the economics of AI, this episode covers what’s changing and why it matters for developers.

Send us Fan Mail

In this episode of Beyond the Build, we break down some of the most interesting stories happening across AI, software engineering, and the broader tech industry.

We talk about AI agents moving into the browser, the growing US-China AI divide, why AI search may be weakening the open web, and how prediction markets are blurring the line between investing and gambling. We also get into Shopify replacing Redis with MySQL, what code quality looks like when AI is writing more of the code, and why runtime validation may become increasingly important.

Beyond engineering, we discuss AI companions, digital sovereignty, AI-powered learning, real-world security practices, AI-assisted journalism, and the backlash around creators like Hank Green using AI for research.