Features

Everything you need to build intelligent, deterministic applications.

Intent Resolution

Natural language → structured tool calls with semantic caching. Results are cached automatically, and you can correct mistakes to improve accuracy over time.

  • • Cached responses for repeat queries
  • • Vector similarity matching
  • • Learning from corrections via /correct
  • • Flat per-request pricing
Request
{
  "query": "What's the weather in Tokyo?",
  "toolsets": ["weather-v1"]
}
Response
{
  "resolved": {
    "tool": "get_weather",
    "parameters": { "location": "Tokyo, Japan" }
  },
  "metadata": {
    "source": "cache",
    "latency_ms": 12,
    "requests_used": 1
  }
}
Request
{
  "items": [
    {
      "query": "book a table",
      "correct_tool": "reserve_restaurant"
    }
  ]
}

Memory Banks

Portable, domain-specific context libraries. Create banks for different domains (e.g., travel, finance, support) and assign them to apps.

  • • Reusable across apps
  • • Bulk import via API
  • • Public or private visibility
  • • Automatic learning via /correct

Text Classification

Single and batch classification using state-of-the-art models. Simple, predictable results for any text.

Semantic Cache Hit
Cached response via Classification Sets
Batch Processing
Classify up to 100 items in a single request
Request
{
  "data": "I need to cancel my order",
  "classes": [
    { "label": "support", "description": "Customer support" },
    { "label": "sales" },
    { "label": "billing" }
  ]
}
Response
{
  "results": [
    {
      "input": "I need to cancel my order",
      "classification": "support",
      "confidence": 0.94
    }
  ],
  "metadata": {
    "requests_used": 1
  }
}
Request
{
  "data": "turn on kitchen and turn off bedroom",
  "classification_set": "area-router-v1"
}
Response
{
  "results": [{
    "classification": "",
    "extraction_needed": true,
    "extracted": [
      {
        "query": "turn on kitchen lights",
        "classification": "kitchen",
        "confidence": 0.97
      },
      {
        "query": "turn off bedroom lights",
        "classification": "bedroom",
        "confidence": 0.96
      }
    ]
  }],
  "metadata": {
    "requests_used": 2,
    "extraction_performed": true,
    "sub_classifications_performed": 2
  }
}
⚡ New Feature

Classification Extraction

Automatically extract and classify compound queries. Split multi-intent queries into individual sub-queries in a single API call.

  • • LLM-powered multi-intent detection
  • • All sub-queries classified in one request
  • • Saves HTTP round trips
  • • Transparent cost tracking (2 requests)
Enable per classification set
Toggle extraction on/off for each classification set in your dashboard.

Conversational Responses

Get human-readable responses alongside structured tool calls. Define personas to control tone and style — from friendly assistants to gruff NPCs.

  • • Tool resolution + natural language in one call
  • • Per-app personas for tone control
  • • Response templates cached automatically
  • • Parameter substitution in responses
Request
{
  "query": "Set living room lights to 50%",
  "toolsets": ["smart-home-v1"],
  "persona": "friendly-assistant"
}
Response
{
  "resolved": {
    "tool": "set_lights",
    "parameters": { "room": "living room", "level": 50 }
  },
  "response": {
    "text": "Living room lights set to 50%!"
  },
  "metadata": {
    "source": "cache",
    "latency_ms": 15,
    "requests_used": 2
  }
}
Request
{
  "query": "reserve a table",
  "correct_tool": "book_restaurant",
  "correct_params": { "type": "reservation" },
  "target_bank": "restaurant-intents"
}
Response
{
  "success": true,
  "message": "Correction stored successfully"
}

Correct & Learn

Self-improving system. When the API gets something wrong, correct it once and it learns for all future requests.

  • • One-shot learning
  • • Automatic bank updates
  • • Improves cache accuracy
  • • No retraining required

Developer Console

Full management UI for apps, memory banks, API keys, and usage analytics. Monitor performance, track request consumption, and manage your entire Intentgine deployment.

App Management

Create apps, generate API keys, configure toolsets, assign memory banks.

Usage Analytics

Track request volume, cache hit rates, latency, and endpoint usage.

Memory Bank Editor

Bulk import/export, item management, visibility controls.

Billing Dashboard

Monitor request balance, view invoices, manage subscriptions.

Ready to build?

Start using Intentgine today.