Small Decision Models, Terminal Web Apps, & Effect.ts | Austin JS Meetup

​Come and give a talk! Sign up for a presentation!

​Talk: Driving Typed UIs with Small Decision Models (and an LLM as Backup)
Speaker: Collier King
Abstract: Most agent UIs send every click-equivalent through an LLM. This talk shows a different pattern: a small classification model (an RLCD model such as Jev or Cloudflare's Clef) routes each chat request in under a second and drives a typed React workspace on the user's behalf. The LLM only steps in when the classifier isn't sure. We'll cover OpenAPI as one backbone for REST, MCP and the web app; the Flue agent framework and a browser bridge that run every agent command through the same controllers a click uses; and what makes a classification model succeed in production: deterministic pre-processing, per-model calibration gates, and replayable evaluation. Live demo included, plus what this means for where inference is heading.

Talk: Run Webapps in your Terminal
Speaker: Ryan X. Charles
Abstract: TermSurf is a terminal that supports running real web browsers for agentic web developers. We're not talking text-based browsers here, but real web browsers like Chromium. Not only can you browse the web for real inside TermSurf, but you can also build graphical apps based on web technologies inside, like a light version of Electron or Tauri. Demo apps will be presented like images viewers, messengers, and other apps to aid agentic web developers.

Title: From Functions to Effect
Speaker: Kevin Kipp
Abstract: Most TypeScript functions hide important facts: what can fail and what services they need. This talk builds from ordinary functions through Result types to Effect’s Effect<Success, Error, Requirements> model. Learn how Effect makes failures and dependencies visible in types, while keeping programs composable for retries, logging, testing, and more.

​Meetup Schedule:
6:00 - 7:00 -> Food and networking
7:00 - 8:00 -> Talks and Q&A
8:00 -> Walk to Zilker Brewing Company and hangout