North Meets South Web Podcast

Hosted by Jacob Bennett and Michael Dyrynda

TechnologyExplicit

Website RSS feed

Episodes

196

Latest episode

Jun 2026

Language

About the show

Jake Bennett and Michael Dyrynda conquer a 14.5 hour time difference to talk about life as web developers

Listen to episodes

60 recent

June 4, 2026Episode 19456 min

Fast Laravel with Jason McCreary

Michael and Jake are joined by Jason "JMac" McCreary to talk the impact of AI on Laravel Shift and modern upgrade workflows, and his latest Fast Laravel course focused on edge caching and application performance.Jason shares how Laravel Shift has evolved alongside AI-assisted development, why recent Laravel releases have changed the upgrade landscape, and why he still believes there's value in keeping applications aligned with the latest framework conventions rather than simply running composer update. The conversation explores how AI tools are influencing developer workflows, the future of upgrade automation, and new ways Shift is integrating with agentic coding tools. The second half of the episode dives deep into Fast Laravel, Jason's course on making Laravel applications dramatically faster using Cloudflare edge caching. Drawing on decades of web development experience, he explains why page caching remains one of the most effective performance techniques available, how Laravel's default stateful behaviour can prevent effective caching, and the practical steps required to achieve cache rates approaching 99% on real-world applications.Show LinksLaravel ShiftShift AI SkillsFast LaravelSeparate `static` middlewareManaged queues on Laravel CloudLaravel Cloud

May 21, 2026Episode 19346 min

Laracon AU CFP, developer storytelling, and audience engagement

Michael shares a behind-the-scenes look at organising Laracon AU 2026, including the new committee-based CFP review process, the tooling built to manage the talk submissions, and how AI-assisted workflows helped shape the final conference schedule. The conversation dives into balancing technical depth with audience engagement, designing conference cadence to avoid cognitive overload, and why advanced technical talks are so difficult to execute well.Jake and Michael also discuss the realities of crafting technical presentations, from simplifying code examples and avoiding "proof of expertise" syndrome, to using AI tools as collaborative thought partners when preparing talks. Along the way, they explore how conference organisers think about audience fit, production experience, practical takeaways, and keeping attendees engaged during deeply technical sessions.Show linksLaracon AUModel Context Protocol (MCP)Riff & Refine: Trust the Process

May 7, 2026Episode 19236 min

Unused APIs, Passport testing traps, and local AI bottlenecks

In this episode, Michael shares details from a major internal platform shift at work, including the decision to completely remove an underused public JSON API and rebuild integrations around real customer needs instead of hypothetical use cases. The conversation dives deep into Laravel Passport, Sanctum, OAuth flows, request authorisation, and some tricky edge cases around testing authenticated APIs.Jake then broadens the conversation into AI infrastructure, local model hosting, security implications of autonomous AI systems, NVIDIA hardware demand, and the future potential of photonic processors as a solution to the growing power and cooling bottlenecks facing AI workloads.Show linksLaravel PassportLaravel SanctumLaravel Passport actingAs testing helpersPHP enumsPHPStanLarastanZapierClaudeNVIDIA DGX systemsPhotonic processors

April 23, 2026Episode 1911 hr 0 min

Gents on Gent with David Hemphill

Michael and Jake are joined by David Hemphill to discuss David's macOS app Gent, a task runner built on the "Ralph loop" pattern for AI-powered coding workflows.The conversation covers how Gent takes a project requirements document (PRD), breaks it into small tasks that fit within a single context window, and runs them sequentially or in parallel using copy-on-write clones and Git worktrees.We discuss our own evolving workflows with Claude Code, including plan mode, the "Grill Me" skill for stress-testing plans, managing context windows, and the /rewind command.Show LinksDavid HemphillGentRalph loopConductorPolyscopeChief"Grill Me" skillMatt Pocock / AI HeroSoloThe Eternal Promise: A History of Attempts to Eliminate ProgrammersLaracon

April 9, 2026Episode 19049 min

Flight booking mistakes, Laracon AU, and dead letters

In this episode, Michael and Jake catch up ahead of Laracon and share a wild travel story involving flight changes, third-party booking headaches, and expensive rebooking.Jake then shares a fun personal highlight: attending the NCAA Men's Basketball Championship and watching Michigan win.The conversation shifts into development work, where Jake dives into building a centralised system for managing failed Laravel jobs across multiple applications. He explains the challenges of aggregating failed jobs without Horizon, how they built a custom package to expose APIs for inspecting and retrying jobs, and the nuances of Laravel's queue system.They also explore ideas for turning this work into a Laracon talk, emphasising practical, experience-driven content over purely technical deep dives.Show LinksLaracon AULaravel HorizonSentryDead Letter Queue (00:00) - Introduction and road to episode 200 (01:00) - Laracon plans and travel setup (02:00) - Flight booking disaster and schedule change (06:00) - Rebooking flights and unexpected costs (09:00) - Lessons learned with third-party bookings (10:00) - Michigan wins NCAA championship (12:30) - Midwest geography and personal background (12:45) - Building a centralized failed jobs system (15:30) - Challenges with retries and tracking failures (16:40) - The "Dead Letter" package and API approach (23:20) - Turning real-world problems into Laracon talks (48:20) - Wrapping up and outro

March 19, 2026Episode 18938 min

OIDC, bastion hosts, and production safety

In this episode, Jake and Michael dive into modern infrastructure security practices, sparked by an annual audit and the painful process of rotating AWS IAM tokens. That experience leads into a broader discussion on why long-lived credentials in GitHub Actions are risky, and how OIDC (OpenID Connect) enables a more secure, short-lived, role-based alternative.Show linksScout SuiteOpenID Connect (OIDC)Laravel ForgeLaravel HorizonScrambleClaudeLoRA (Low-Rank Adaptation)

March 5, 2026Episode 18844 min

Worktree structures, workflow events, and enum metadata

In this episode, we discuss using `claude --worktree` to spin up parallel feature work, and the unexpected friction that can arise when your editor doesn’t play nicely with nested worktrees.Jake shares his experience running multiple Claude agents in parallel and problems that surfaced in PhpStorm. Michael explains how he structures worktrees differently, avoiding those issues, and the two compare workflows between PhpStorm and Neovim.Show Linksclaude --worktreeGit worktreesPhpStormNeovimTmuxLazyGitUsing GitHub CLI in workflowsarchtechx/enumsArborLaracon USGit Worktree Hub plugin for PhpStorm

February 19, 2026Episode 18742 min

Charging chaos, corona discharge, and vector embeddings

Michael and Jake discuss Jake's device charging chaos, household optimisation, international power outlets, and vector embeddings.Show linksGitryinMagnetic 3-in-1 wireless chargerDesktop charging station 12-in-1Corona dischargeLaracon US ElevenLabsDaily Dose of DS (Data Science)MstyRetrieval-Augmented Generation (RAG)Laravel AI

February 5, 2026Episode 18639 min

OpenClaw, Arbor, and horseless carriages

Michael and Jake catch up on what’s been occupying their time lately, from AI tooling experiments to new developer workflows, before closing with a broader reflection on how new technologies are often misunderstood at first.Show linksOpenClaw / Clawd Bot / MoltbotArborAI horseless carriages

January 22, 2026Episode 18546 min

When AI clicks, automation at home, and developer workflows

Jake and Michael return for 2026 and talk about their evolving experiences with AI; what it’s good at, what it’s not, and how it’s changing the way they work.Show linksOpenAI / ChatGPTAnthropic / Claude (Sonnet & Opus)OpenCode (multi-provider AI coding interface)Home AssistantZigbee temperature sensorsGitHub CopilotOllama (local LLM runner)NVIDIA DGX SparkAmp CodeMiniMax M2.1 modelSoftware for an audience of oneArborOpenCode Desktop has workspaces supportOpus 4.5 is going to change everything