A recap episode of airhacks.tv covering a broad sweep of topics: API gateway comparisons (nginx, HAProxy, cloud gateways), BCE (Boundary-Control-Entity) applied to static pages, the release of HTMLDB (a ~550-line zero-dependency Java 25 database built on HTML/XML, designed for agent memory), floci.io (a Quarkus-based emulator for AWS services with ~20k GitHub stars in 5 months), speculation that LLMs understanding the web platform natively could make frontend frameworks like React and Angular obsolete, a vanilla Java AI agent running on AWS agent core in ~40 lines with 60ms startup, and a look back at an episode from April 2018 discussing WebStart and JavaFX.
Questions this post answers
What is HTMLDB and how is it built?
HTMLDB is a database built on HTML treated as valid XML, implemented in roughly 550 lines of Java 25 with no external libraries. It supports tables and rows via a CLI, lets data be viewed in a browser as semantic HTML, and was motivated by agent memory use cases since LLM agents understand XML well. It is part of the Z family of tools and is not meant as an Oracle or Postgres replacement. Developers tracking lightweight, agent-friendly data stores can follow releases like this one on daily.dev.
What is floci.io and how popular has it become?
floci.io is a Quarkus-based emulator that reproduces AWS services including Lambda, DynamoDB, Neptune, CodeBuild, and ElastiCache for local development and testing. It gained roughly 20,000 GitHub stars within five months of release, indicating rapid adoption among developers who need to emulate AWS infrastructure without deploying to the cloud. Teams evaluating AWS emulation tools can compare fast-growing options like floci.io on daily.dev.
How fast does a vanilla Java AI agent start on AWS agent core runtime?
A vanilla Java agent running on AWS agent core runtime starts in about 60 milliseconds and has a footprint of roughly 5KB, implemented in around 40 lines of code with no frameworks. This demonstrates that lightweight, dependency-free Java can be viable for building AI agents on cloud runtimes without pulling in heavier frameworks. Java developers building lean AI agents can keep up with minimal-footprint techniques on daily.dev.