AgentChat
The agent capability layer for your SaaS

Turn your APIs into an agent that can do anything.

Expose your product’s core operations as APIs, attach their documentation to a chat, and AgentChat turns them into agent capabilities. No custom tool integration for every feature.

4Core tools HTTPUnlimited capabilities HeadersUser-level access
One agent layer · every product capability
Your SaaS APIs
AgentChat
POST/stories
POST/images
GET/documents/search
→
Aria4 CORE TOOLS

get_api_document()

http_request()

sleep()

view_image()

API docs = plugins. Add another documented endpoint and the agent gains another capability.
Your APIsSecure executionYour database

The core idea

Four tools. Every capability your product can expose.

Traditional agents need a custom tool definition for every action. AgentChat uses API documentation plus a universal HTTP executor, so capability expands with your API surface—not your agent code.

01

Discover

get_api_document

Load the exact API contract needed for the current task.

02

Execute

http_request

Call any documented HTTP endpoint and receive structured results.

03

Wait

sleep

Pause between asynchronous job creation and status polling.

04

See

view_image

Inspect image URLs returned by your creative or media APIs.

Integration model

Your database stays the source of truth.

Your APIs are controlled wrappers around the operations your product already performs. The agent calls them; your SaaS reads the same database and renders the result naturally.

01

Document your core operations

Register the APIs that wrap your product database and business actions. They become the agent’s available capabilities.

02

Create a scoped chat

Attach only the documents that chat needs, select its model, and add headers for the current user or tenant.

03

Let your product render the result

The agent operates through your APIs. Your application reads its own database and displays the newly created or updated data as usual.

Real product patterns

One architecture, completely different agents.

RAG / SEARCH01

Give the agent your document search API.

Attach one search API document when creating the chat. The agent can retrieve relevant passages, reason over them, and answer with your application’s private knowledge.

GET /documents/search?q=...

→ {
  "results": [
    { "text": "...", "source": "..." }
  ]
}
AI STORY STUDIO02

Turn story creation into a conversation.

Provide a story-writing API and an image-generation API. The agent writes chapters, generates illustrations from prompts, stores both through your APIs, and your story app displays them from its own database.

POST/stories
POST/images → image_url
PUT/stories/{id}/chapters

Tenant-aware security

The same API docs. Different access for every user.

Configure host-specific headers on each chat. Your API receives the current user or tenant credential, applies its own authorization rules, and returns only the data that user is allowed to access.

Per-chat header injection
CHAT / USER AX-Tenant-ID: tenant_a

Can search and modify only tenant A data.

CHAT / USER BX-Tenant-ID: tenant_b

Uses the same API contract with tenant B permissions.

Header values stay write-only and are never exposed to the model.

Developer API

Add the agent layer with two resources.

Register API documentation, then create a chat session that references it. Your product can start runs and poll durable messages through the same API.

Open the developer dashboard →
https://agent.iamdev.me
POST/api/api-documents # define capabilities
POST/api/agent/sessions # scope a chat
POST/api/agent/sessions/{id}/chat
GET/api/agent/sessions/{id}/messages

AgentChat

Your SaaS already has the capabilities. Give them to an agent.

Start building →