For developers

Create AI apps, agents, and voice interfaces on the Yandex AI Studio platform.

Fast and stable inference, easy-to-use APIs, and built-in tools that speed up development and reduce manual integration.

Minimal latency and stable performance

Stable inference, streaming responses, and agent status management make applications fast and reliable.

Full compatibility with OpenAI API and OSS frameworks

The Completions API, Responses API, and Realtime API support the industry standard. It works with LangChain, LangGraph, LlamaIndex, n8n, Dify, and others.

Everything you need for a quick start and scaling

Ready-made APIs and SDKs, tools, batch pipelines, and infrastructure for real loads.

Integration with a company’s internal systems

Connecting CRM, ERP, and internal databases via MCP and API. Support for RAG, vector storage, and enterprise data sources.

What the platform consists of

Responses API for creating text agents

Features for agents:

Conduct dialogs based on history

Search for information on the internet

Calling tools

Access a company’s internal and external systems via MCP servers

Use RAG and vector databases, document analysis

Work with all AI Studio models

Examples of text agents

Compliance agent

Goal: Automatically check counterparties before signing an agreement, call external databases (the Federal Tax Service, inspection lists), and generate reports.

Creation methods:

Agent Atelier

Create a text agent for legal or accounting staff using the Responses API

MCP Tool

API integration Federal Tax Service, Federal Financial Monitoring Service, and commercial services

Workflows

Set up a scenario: enter a TIN → request to databases → generate a report → send to employee

Internal employee assistant

Goal: Help employees quickly find instructions, regulations, and answers in documentation and internal systems.

Creation methods:

Agent Atelier

Set up a text agent for a chat based on the Responses API

File Search Tool and Vector Store API

Index a company's knowledge base Search documents and articles

Document analysis and summarization agent

Goal: Analyze voluminous documents (reports, contracts, minutes of meetings), extract facts and conclusions, and form a brief summary.

Creation methods:

Agent Atelier

Create a text agent based on the Responses API

File Search Tool

Search documents to extract relevant snippets

Workflows

Set up a scenario: upload a document → analyze content → extract points → generate brief report → send result to corporate system

Realtime API for creating voice agents

WebSocket streaming interface for voice interaction

A ready-made pipeline: speech recognition → LLM + File Search → speech synthesis with minimal latency

Work with audio and text at the same time

Support for a long dialogs in real time

Generating text and audio in stream

Calling tools right in the process of communication

Support for agent interruption

Examples of voice agents

Customer support agent

Goal: принимать входящие обращения клиентов и помогать с вопросами о продуктах, услугах или сервисах компании (например, тарифы, парковка, условия обслуживания, статусы заявок).

Creation methods:

Agent Atelier

Low-latency voice dialogue using the Realtime API

File Search Tool

Search for answers in a company's knowledge base

MCP Tool

Access to CRM, billing, orders, and other systems

Outgoing voice agent

Goal: To make outgoing calls to customers for service notifications, register for services or confirm actions.

Creation methods:

Agent Atelier

Set up a voice agent using the Realtime API

MCP Tool

Work with schedules, CRM, requests, bookings

Brand Voice (опционально)

A unique voice for a company

Website voice assistant

Goal: To act as the site’s voice interface and help users find information, choose products, and perform actions.

Creation methods:

Agent Atelier

Set up a voice interface based on the Realtime API

Web Search Tool

Search sites and public sources

File Search Tool (опционально)

Responses based on internal materials

Additional features for developing AI solutions

Vector Store API

Ready-made RAG for searching corporate data.

Yandex Search API

Getting relevant information from the internet.

Tuning API

Fine-tuning of models using the LoRA method on your data.

Batch API

Asynchronous sending of large sets of requests.

Developer workspace

Testing, integrating, and configuring agent logic in the platform’s tools, without unnecessary infrastructure.

Playground

An interface for testing models, prompts, dialogs, and agent chains, without needing to write code. It allows you to quickly experiment and test hypotheses.

SDK and examples

Python™ and JavaScript libraries, ready-made requests, tokenization, convenient tools for working with the API.

Interface for creating agents

Connect tools, customize search strategies, and set agent logic in a low-code environment.

Start working with Yandex AI Studio

Build AI apps, agents, and voice interfaces. Automate business processes. The platform already has everything you need to get started.

*The activities of Meta Platforms Inc. are recognized as extremist and are prohibited on the territory of the Russian Federation.