Product

Extend what your assistant can do

Plug in order tracking, customer profiles, delivery status, and custom workflows without rebuilding your assistant.

How Tools Work

Every tool is a function the agent can call during conversation. Some run on your server, some run in the customer's browser, and you can build your own.

Server Tools

Search products, check inventory, retrieve order history, and look up customer profiles. All executed server-side for speed and security.

Client Tools

Add to cart, highlight products, apply filters, and capture preferences. Executed in the browser for instant UI updates.

B2B Processing

Validate incoming orders against your catalog, match products by name or code, and auto-complete purchase flows.

Custom Workflows

Build your own tools: connect ERPs, trigger webhooks, query external APIs, or add domain-specific logic to the agent loop.

Built-In Tools

These ship with the platform. Enable them per project from the management dashboard.

Outcome-Aware Search

Core

Searches your catalog by intent, not keywords. Categorizes results as essential, recommended, or complementary.

  • Intent inference from natural language
  • Multi-category bundling
  • Popularity + relevance scoring
See How Search Works

Order History & Reorder

Popular

Loads past orders into context so the agent can suggest reorders, track delivery status, and answer 'where is my package?'

  • Past order lookup
  • Delivery status tracking
  • One-message reorder
See the Assistant

Customer Profile

Persists dietary restrictions, brand preferences, sizes, and allergies across sessions so the agent never asks twice.

  • Persistent preference storage
  • Session-spanning memory
  • Profile-aware filtering
Request a Demo

Define Custom Tools

Each tool is a function the agent can call during conversation. Define the name, parameters, and handler. The agentic loop decides when to use it.

Server or Client Execution

Server tools run on your backend; client tools execute in the browser for instant UI updates

Typed Parameters

JSON Schema validation ensures the agent passes correct data to your tool

Simulation Gym

Test tools with LLM-simulated users before deploying to production

Developer Documentation
// Custom server tool definition
{
"name": "check_loyalty_points",
"type": "server",
"parameters": {
"customer_id": "string"
},
"handler": "/api/loyalty/points"
}
// Agent uses it automatically:
"You have 2,340 points, enough for
free shipping on this order."

Use ours or build yours

Enable built-in tools from the dashboard, or define your own with a JSON schema and an API endpoint.

IsarTech | Conversational AI That Sells