# Technical

How Sistava is built and how to integrate it: agent runtime, channel layer, OAuth integrations, MCP/A2A, webhooks, public API.

## How Sistava is built

Sistava is a production AI workforce platform: a multi-tenant agent runtime with durable execution, a 7-layer memory system with knowledge graphs, and a multi-channel communication layer. AI employees run scheduled tasks, real-time conversations, agent-to-agent delegations, browser automations, and API-triggered workflows on a single execution engine.

## Integration surface

- OAuth app integrations: one-click connections to Gmail, Slack, Notion, HubSpot, Jira, and thousands more.
- MCP: connect AI employees to any Model Context Protocol server.
- A2A: delegate work to external agents over the Agent-to-Agent protocol.
- REST and GraphQL APIs: call any API directly; webhooks in both directions.
- Channels: in-app chat, voice, Slack, WhatsApp, email, telephony, embeddable chat widgets, and a desktop companion app.

## Explore the technical pages

- AI models: /en/technical/ai-models
- Communications and channels: /en/technical/communications
- API access: /en/technical/api
- A2A interoperability: /en/technical/a2a