Emotional intelligence API · invite-only beta

Give your AI a sense of the person on the other side.

One API reads the face, the voice and the words in a clip, then returns a small JSON reading: mood, energy, stress, attention and balance, each scored 0 to 100 with a plain label.

Request access → Quickstart

Priced by the minute of media · 10 free minutes a month · invite-only while we finish v1

REST + JSONScores 0–100English and ArabicMedia deleted after analysis
CLIP · 0:00video · audio · text expressionsneutralmomentsshift mood.label = neutral
POST /api/v1/readexample
{
"mood": { "label": "neutral" },
"balance": { "score": 74 },
energy.score42
stress.score30
attention.score78
}
Example reading on sample data

Your agent reads ten thousand tokens a second. It still can’t tell when the person on the other side is about to give up.

01 · Input

Send a clip or a sentence.

Post a short video or audio clip, or plain text. One signal or all three.

02 · Face

Expressions, gaze and steadiness.

Read from the video, about one frame a second. The clip is deleted once the reading is made.

03 · Voice & words

Read together.

Tone, energy and pauses from the voice; how positive or negative the words are, in English and Arabic.

04 · Output

One small JSON back.

Mood, energy, stress, attention and balance. Every score 0 to 100, every label from a short fixed list.

05 · Privacy

Media deleted after analysis.

Raw video and audio are removed when the reading is made. A field you did not send is left out, never guessed.

faceexpression · sad facelooked away voice · quieter"I guess…"voice · long pause"it's fine"words · leaning negative
POST /v1/read200 · example

reading · ses_8a3f

mood32
energy30
stress55
attention81
balance62
mood"label": "negative"
Media deleted after analysisScores 0–100
QUICKSTART

From a key to your first reading.

Four steps. The example runs here on sample data; the same calls work against the API today.

Errors, limits and every field: Read the docs →

.env
FOUR WAYS IN

One engine behind every door.

API

The endpoint.

One call for a sentence, one session for a clip. JSON in, JSON out.

POST /v1/readAPI reference → SDKs

SDKs, next.

A TypeScript SDK comes with v1, then Python. Any HTTP client works today.

curl · fetch · requestsRead the quickstart → SoonMCP

MCP server.

Planned after the SDK: the same JSON for Claude and other agents.

MCPWhat is planned → Studio

No code.

Invited developers upload a clip in the Console and see its reading.

feneri.xyz/consoleAbout Studio →
0–100every score, with a plain label
5languages and dialects: English (US, UK) and Arabic (UAE, Saudi, Egypt)
10free media-minutes a month on Sandbox
Deletedthe media, once the reading is made

Build with the engine.

Request access →Read the docs

Priced by the minute · invite-only beta