File size: 1,638 Bytes
9971119
bd3ef8b
9971119
 
 
 
 
 
 
bd3ef8b
 
9971119
 
4131875
 
 
 
 
 
 
 
 
 
 
 
 
bc1c646
 
 
 
 
 
4131875
8f9bde4
 
 
 
4131875
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
---
title: ARIA
emoji: 💬
colorFrom: yellow
colorTo: purple
sdk: gradio
sdk_version: 6.5.1
app_file: app.py
pinned: false
license: apache-2.0
short_description: Adaptive Response & Intelligence Agent
---

ARIA — **A**daptive **R**esponse & **I**ntelligence **A**gent. A Messenger-style,
event-driven, multimodal conversational agent for the NNLABS platform.

This Space is **ARIA's brain** (vision + reasoning, emits structured JSON bubbles).
The full product also ships a Cloudflare realtime backend and a Messenger UI — see
[`ARIA_ARCHITECTURE.md`](./ARIA_ARCHITECTURE.md).

## Layout
```
app.py            Gradio Space entry — exposes api_name="respond"
brain/            persona, contracts, model orchestration (vision)
```

The Cloudflare side (the `aria-worker` Durable Object, the D1 migrations and the
Messenger front-end) used to live in this repo under `cloudflare/`. It has moved to
the website repository, where the rest of the deployment lives. This Space is only
the brain. A public Space is the wrong place to keep the map of a private backend,
and the copies here had drifted behind the live ones anyway.

## Run / deploy
- **Brain (this Space):** secrets `HF_TOKEN` (inference) and `ARIA_BRAIN_SECRET` (the
  aria-worker signs every call with it; without it the Space refuses every call);
  optionally `ARIA_MODEL`. There is no public playground: the brain is billed to the
  organisation and answers only signed requests.

Built on [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub),
and [Cloudflare Workers / Durable Objects / R2](https://developers.cloudflare.com/).