Agent Use Interface (AUI) – let users bring their own AI agent
As I started building AI integrations, I came to realize that for many projects, the the best agentic experience is one that simply enables the user's personal agent to take actions within your app. The existing options like MCP or A2A are quite involved and for simple apps that are already URL parameter driven, those options seem like overkill. This led me to prototype the Agent Use Interface (AUI) spec. The idea is simple: a lightweight, open spec that makes any app "agent-navigable." You drop an XML file at /agents/aui.xml that describes the URL-parameter-driven actions your app…
What it does
In the maker’s words, at launch
As I started building AI integrations, I came to realize that for many projects, the the best agentic experience is one that simply enables the user's personal agent to take actions within your app. The existing options like MCP or A2A are quite involved and for simple apps that are already URL parameter driven, those options seem like overkill. This led me to prototype the Agent Use Interface (AUI) spec. The idea is simple: a lightweight, open spec that makes any app "agent-navigable." You drop an XML file at /agents/aui.xml that describes the URL-parameter-driven actions your app supports, like search, create, filter, etc. And that way any AI agent can read aui.xml, understand what's possible, and construct URLs on behalf of the user. That's it. No SDK. No auth flow. No API keys. Just a catalog of what your app can do, written for LLMs to understand. Is there something like this that already exists? Is the approach too simple to be useful? If your app already supports Universal Links or is otherwise URL parameter driven you could probably add support for AUI in an afternoon. See a working example: https://habittiles.app/agents/aui.xml
Does the same job
all alternatives →
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, March 2026
the whole month →
Switch from ChatGPT to Claude with import memory feature
AI · Mar 2026 · claude.com
