Skip to main content
POST
/
functions
/
v1
/
chat
curl -X POST "https://api.myweave.ai/functions/v1/chat" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key-here" \
  -d '{
    "message": "Hello, I need help with leadership development",
    "context": {
      "coachId": "coach_xyz789",
      "userId": "user_123",
      "isAuthenticated": true
    }
  }' \
  --no-buffer

data: Hello! I'd be happy to help you with leadership development.
data:  Let's start by understanding your current goals...
data: 

[Response completes]

​
Overview

The Human Intelligence API enables real-time conversations with personas trained on real human expertise. It uses Server-Sent Events (SSE) for streaming responses and supports file attachments, course context, and authenticated/anonymous users.

​
Request Body

​
message
string
required
The user’s message content
​
thread_id
string
Existing thread ID. If not provided, a new thread will be created automatically. The thread ID is returned in the response headers (X-Thread-Id) and in the message data.
​
language
string
default:"en"
🧪 Experimental FeatureISO 639-1 language code for responses. The persona will respond naturally in the specified language while maintaining their authentic voice and real human expertise.Supported Languages (65): English, Spanish, French, German, Chinese, Japanese, Arabic, Hebrew, Hindi, Portuguese, Russian, Italian, Korean, and 52 more.RTL Languages: Arabic (ar), Hebrew (he), Persian (fa), Urdu (ur) automatically render right-to-left.Examples: en (English), es (Spanish), fr (French), zh (Chinese), ar (Arabic), he (Hebrew)
This feature is in experimental phase. Language quality may vary by language and coaching context.
​
context
object
required
Context information for the conversation
​
attachments
array
File attachments for the message
​
action
string
Special action to execute:
  • create_knowledge - Create knowledge content (requires coach authentication)

​
Response

The API returns a streaming response using Server-Sent Events (SSE). The thread ID is included in both:
  • Response header: X-Thread-Id
  • SSE message data
​
X-Thread-Id
header
Thread ID for the conversation (in response headers)
SSE Event Format:
  • Each chunk is prefixed with data:
  • Contains text chunks from the persona’s response (powered by real human knowledge)
  • Final chunk signals completion
curl -X POST "https://api.myweave.ai/functions/v1/chat" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key-here" \
  -d '{
    "message": "Hello, I need help with leadership development",
    "context": {
      "coachId": "coach_xyz789",
      "userId": "user_123",
      "isAuthenticated": true
    }
  }' \
  --no-buffer

data: Hello! I'd be happy to help you with leadership development.
data:  Let's start by understanding your current goals...
data: 

[Response completes]

​
Features

  • Real-time Streaming: Server-sent events for live responses
  • Multilingual Support: Personas respond in 65 languages including RTL languages (Arabic, Hebrew, Persian, Urdu)
  • Human Expertise: Each persona trained on real human knowledge and expertise
  • Automatic Thread Management: Threads created automatically, ID returned in headers
  • File Attachments: Support for images, PDFs, and documents
  • Course Integration: Connect conversations to structured courses and lessons
  • Mode-based Conversations: Different consultation modes per persona
  • Anonymous & Authenticated: Works for both user types
  • Knowledge Creation: Special create_knowledge action for persona owners (requires authentication)

​
Thread Management

First Message (New Thread):
  • Don’t include thread_id in request
  • API creates new thread automatically
  • Thread ID returned in X-Thread-Id header
  • Store this ID for continuing the conversation
Follow-up Messages (Existing Thread):
  • Include thread_id from previous response
  • Continues the conversation in same thread
  • Maintains full conversation context

​
Authentication

  • For Users: No authentication required for basic chat
  • For Persona Owners: Bearer token required when using action: "create_knowledge"
    • Include Authorization: Bearer <jwt_token> header
    • Must be authenticated as the persona owner specified in coachId

​
Use Cases

  • 1-on-1 Conversations: Real-time persona consultations powered by real human expertise
  • Course Learning: Contextual help within structured courses built on human knowledge
  • Document Review: Upload files for persona feedback based on real expertise
  • Knowledge Creation: Persona owners can create training content via chat
  • Global Multilingual Support: Personas respond in user’s native language while maintaining authentic human voice

​
Supported Languages (65) 🧪 Experimental

The Chat API supports 65 languages for AI responses. Simply include the language parameter with the ISO 639-1 code.
Experimental Feature: Multilingual support is currently in beta. While the AI can respond in all listed languages, quality and accuracy may vary. We recommend testing thoroughly for production use cases.

​
Language Codes Reference

CodeLanguageNative NameFlagNotes
enEnglishEnglish🌍Default language
sqAlbanianShqip🇦🇱
amAmharicአማርኛ🇪🇹
arArabicالعربية🇸🇦RTL
hyArmenianՀայերեն🇦🇲
bnBengaliবাংলা🇧🇩
bsBosnianBosanski🇧🇦
bgBulgarianБългарски🇧🇬
myBurmeseမြန်မာ🇲🇲
caCatalanCatalà🇪🇸
zhChinese中文🇨🇳
hrCroatianHrvatski🇭🇷
csCzechČeština🇨🇿
daDanishDansk🇩🇰
nlDutchNederlands🇳🇱
etEstonianEesti🇪🇪
fiFinnishSuomi🇫🇮
frFrenchFrançais🇫🇷
kaGeorgianქართული🇬🇪
deGermanDeutsch🇩🇪
elGreekΕλληνικά🇬🇷
guGujaratiગુજરાતી🇮🇳
heHebrewעברית🇮🇱RTL
hiHindiहिन्दी🇮🇳
huHungarianMagyar🇭🇺
isIcelandicÍslenska🇮🇸
idIndonesianBahasa Indonesia🇮🇩
itItalianItaliano🇮🇹
jaJapanese日本語🇯🇵
knKannadaಕನ್ನಡ🇮🇳
kkKazakhҚазақша🇰🇿
koKorean한국어🇰🇷
lvLatvianLatviešu🇱🇻
ltLithuanianLietuvių🇱🇹
mkMacedonianМакедонски🇲🇰
msMalayBahasa Melayu🇲🇾
mlMalayalamമലയാളം🇮🇳
mrMarathiमराठी🇮🇳
mnMongolianМонгол🇲🇳
noNorwegianNorsk🇳🇴
faPersianفارسی🇮🇷RTL
plPolishPolski🇵🇱
ptPortuguesePortuguês🇵🇹
paPunjabiਪੰਜਾਬੀ🇮🇳
roRomanianRomână🇷🇴
ruRussianРусский🇷🇺
srSerbianСрпски🇷🇸
skSlovakSlovenčina🇸🇰
slSlovenianSlovenščina🇸🇮
soSomaliSoomaali🇸🇴
esSpanishEspañol🇪🇸
swSwahiliKiswahili🇰🇪
svSwedishSvenska🇸🇪
tlTagalogTagalog🇵🇭
taTamilதமிழ்🇮🇳
teTeluguతెలుగు🇮🇳
thThaiไทย🇹🇭
trTurkishTürkçe🇹🇷
ukUkrainianУкраїнська🇺🇦
urUrduاردو🇵🇰RTL
viVietnameseTiếng Việt🇻🇳

​
Right-to-Left (RTL) Languages

Four languages automatically render right-to-left:
  • Arabic (ar) - 🇸🇦 العربية
  • Hebrew (he) - 🇮🇱 עברית
  • Persian (fa) - 🇮🇷 فارسی
  • Urdu (ur) - 🇵🇰 اردو

​
Usage Notes

  • Defaults to English: If language parameter is omitted or invalid, defaults to en
  • Case Sensitive: Use lowercase codes (es, not ES)
  • Natural Responses: AI maintains coaching style while responding in selected language
  • No Translation Needed: User can send messages in any language; specify response language with language parameter