Syntax - Tasty Web Development Treats9 Feb 2024

728: AI Superpowers with Kevin Hou and Codeium

In this supper club, Scott and Wes welcome Kevin Hou, Head of Product Engineering at Codeium, a blazing fast AI-powered code completion and chat tool for developers. Show Notes 00:00 Welcome to Syntax! 05:31 An introduction to Codeium. 07:56 What information are you sending the AI to get such good completions? Codeium compliance 09:15 Codeium runs a 'Language Server'. 11:15 Crawling dependency tree and abstract syntax tree. 12:07 Using Codeium Live. 12:34 How big of a codebase can you run this on? 14:39 Sending select amounts of data to AI. 16:06 Does Codeium maintain codebase preferences and styling (ie. snake case)? 17:39 Will Codeium scan the dependency? 19:23 AI UI, have we found the best format? 21:55 Crazy ideas in tech. 22:53 Additional AI UI inputs. 24:14 How do you make an AI model? 28:42 How does Codeium manage the product roadmap? 32:09 Do AI models get worse over time? How does Codeium validate that it's not? Open AI Evals 35:39 How is Codeium THAT fast? 36:49 What programming langauges does Codeium use? 38:55 Codeium Playground. Codeium Playground 39:15 Caching as a performance improvement. 39:58 What is the pipe between Codeium and editor? 40:17 Codeium chat service. 41:44 A WebSocket system allowing push and pull communication. 43:13 Closed Beta for GPT 4.0. 46:12 The dreaded closing quote bug. 48:26 Sometimes bugs bug Wes. 49:49 Supper Club Questions 51:40 Perplexity.ai 52:35 What editor does Kevin use? 55:31 Sick Picks + Shameless Plugs Sick Picks Kevin: Lapse.com Shameless Plugs Kevin: Codeium.com Hit us up on Socials! Syntax: X Instagram Tiktok LinkedIn Threads Wes: X Instagram Tiktok LinkedIn Threads Scott: X Instagram Tiktok LinkedIn Threads Randy: X Instagram YouTube Threads

Denne episoden er hentet fra en åpen RSS-feed og er ikke publisert av Podme. Den kan derfor inneholde annonser.

Episoder(1025)

1025: The Open Web's second chance (w/ Dan Abramov)

Dan Abramov joins Scott and Wes to explain AT Protocol, the open standard quietly rebuilding the social web. They get into how it actually works, why it’s way bigger than just Bluesky, and why Dan cal...

29 Jul 1h

1024: Open Models Replace Big AI

A huge week for open models: Inkling (the first big US open-weight model since Gemma 4), Qwen 3.8, and Kimi 3 all dropped. Plus Vue 3.6 RC + Vapor Mode, the slow death of Stack Overflow, a decoy font ...

27 Jul 1h 7min

1023: Mosh, Caddy & Tailscale: A Remote Dev Deep Dive

Running your dev process off your main machine doesn’t have to mean the cloud. Scott and Wes break down how to host projects on a box you can reach from any device, close your laptop mid-build, and ke...

22 Jul 48min

1022: Bun re-written in Rust, Zig team big mad

CJ and Scott break down the biggest week in web dev: TypeScript 7 ships with a 10x-faster native port, Bun gets rewritten in Rust (much to the Zig team’s dismay), and Better Auth joins Vercel. Plus GP...

20 Jul 1h 15min

1021: We got addicted to an AI model we can't talk about

Dax Raad, co-founder of OpenCode, joins Scott and Wes to talk remote dev servers, OpenCode 2.0, and why his team is “addicted” to AI models they’re not even allowed to name yet. Show Notes 00:00 In...

15 Jul 51min

1020: Do You Read The Code?

Auto-accept on, brain off? We’re getting into whether anyone actually reads the code they ship anymore — then it’s HTTP’s first new method in 16 years (meet QUERY) and OpenAI’s compute-hungry “Sol Ult...

13 Jul 1h 27min

1019: LGTM, Ship It: The AI Code Review Problem

This episode tackles the growing pains of AI-assisted development, from the struggle of reviewing thousands of lines of agent-generated code to the mounting technical debt when teams merge PRs without...

8 Jul 39min

1018: Google fires Workspace CLI Creator

Google fires the engineer behind its Workspace CLI tool, OpenAI previews GPT-5.6 with three new model tiers, and Astro 7 lands with a full Rust rewrite. Plus: Coinbase cuts token costs with smarter ro...

6 Jul 1h 17min