Ayantram
Demo

How Ayantram Agents Work

Agents do more than chat. They plan, call tools, and produce audit-ready outcomes—securely and autonomously.

Step 1

Ingest

Bring your knowledge—docs, forms, and APIs—into the agent. Index PDFs & docs, register tools, and define policies.

Step 2

Reason

VidyutAI runtime + AnuvaX memory provide context, planning, and routing. Short/long/decay memory keeps agents on-task.

Step 3

Execute

OrchestrateX runs the plan. TantuAPI calls external systems with auth, rate limits, and approvals; ExecuteX handles retries.

Step 4

Respond

InsightX summarizes results, formats responses, updates memory, and hands off to the next step or channel (Slack, email, API).

No-code

Drag-and-drop Visual Editor

Design multi-agent workflows by connecting blocks—Agent, API, Function, Condition, Loop, Router, Response. Versioned deploys, approvals, and rollback are built-in.

Agent
Drag from palette · configure · connect
API
Drag from palette · configure · connect
Function
Drag from palette · configure · connect
Condition
Drag from palette · configure · connect
Loop
Drag from palette · configure · connect
Router
Drag from palette · configure · connect
Parallel
Drag from palette · configure · connect
Response
Drag from palette · configure · connect

Perfect for ops & support teams: ship in hours, not weeks.

For engineers

SDKs & APIs

Everything in the editor is available in code. Use our TypeScript or Python SDKs to define tools, memory, and policies—then deploy.

TypeScript
import { MandalaStudioClient } from 'mandalastudio-ts-sdk';

// Initialize the client
const client = new MandalaStudioClient({
  apiKey: 'your-api-key-here',
  baseUrl: 'https://app.ayantram.com' // optional, defaults to https://app.ayantram.com
});

// Execute a workflow
try {
  const result = await client.executeWorkflow('workflow-id');
  console.log('Workflow executed successfully:', result);
} catch (error) {
  console.error('Workflow execution failed:', error);
}
Python
from mandalastudio import MandalaStudioClient

# Initialize the client
client = MandalaStudioClient(
    api_key="your-api-key-here",
    base_url="https://app.ayantram.com"  # optional, defaults to https://app.ayantram.com
)

# Execute a workflow
try:
    result = client.execute_workflow("workflow-id")
    print("Workflow executed successfully:", result)
except Exception as error:
    print("Workflow execution failed:", error)
Explore the SDK docs

Runtime architecture

VidyutAI
LLM runtime & tool-use
AnuvaX
Short/long/decay memory
InsightX
Docs, OCR, contract parsing
OrchestrateX
Multi-agent workflows
TantuAPI
API gateway & tracing
SentinelX
RBAC, SSO, audit & limits

Security & governance

  • RBAC, SSO/SAML, API keys, and signed tool calls
  • Rate limits, approvals, & policy checks via SentinelX
  • PII redaction & data residency controls
  • Audit logs for every action and prompt

Observability

  • Traces across steps, tools, and external APIs
  • Retries with exponential backoff, dead-letter queues
  • Prompt/response diffs & versioned deploys
  • Metrics: success rate, latency, token usage

Example: Password reset workflow

  1. Step 1
    User requests reset in Slack
  2. Step 2
    Agent validates identity (policy check)
  3. Step 3
    TantuAPI calls IdP with approval
  4. Step 4
    ExecuteX confirms and updates memory
  5. Step 5
    InsightX posts result + audit link

Build visually or with SDKs.

Drag-and-drop editor for speed. TypeScript/Python SDKs for full control.

Platform — VidyutAI, AnuvaX, OrchestrateX, TantuAPI, SentinelX · Ayantram