Quick start

1. Start with a task

Don’t start with a full portfolio. Select a surface that will directly solves a user problem:

Problem	Entry point
Route calls to models	IAM.Router
Check AI traffic and apply DLP	IAM.Secure
Create and launch agents	IAM Agent or IAM Marketplace
Work with documents and sources	IAM.Docs
Add a consultant to the site	IAM.Bot
Handle calls and speech	IAM.Voice + IAM.Speech
Describe architecture as code	IAM.Solution / AGL
Save personal memory locally	IAM.Mind

2. Fix the perimeter

For each thread, determine where the data, model, and decision log reside:

local — execution and memory on the owner’s device;
on-prem / perimeter — services within the selected contour;
dedicated — separate installation or tenant boundary;
cloud — managed public surface of the product.

The production status does not mean that any deployment profile is available without preparation. The product page lists dependencies and operational separately. borders.

3. Create a chain of command

The minimum managed chain looks like this:

User / application
  → IAM.Identity context
  → IAM.Secure inspection
  → IAM.Router decision
  → approved model or tool
  → audit receipt

IAM.Core becomes necessary when multiple products need to detect capabilities of each other, transfer scoped context and coordinate agent actions via MCP/API.

4. Open the desired surface

There are four types of links on each product page:

Landing — public description and connection request;
Application - a working interface, sometimes requiring an account;
API — external endpoint, if it is part of a public contract;
Portfolio - return to the general card iamgroup.ru.

5. Check the integration

Before production use, confirm:

tenant and identity scope;
model route and fallback;
DLP/retention policy;
timeouts, retry and idempotency;
audit/observability;
fault tolerance of downstream services;
staging regression for the target scenario.