Integrations · Platform
API / MCP
Connect systems without an off-the-shelf connector to the AI layer via API or MCP.
Overview
Every organisation runs something bespoke or niche. The API / MCP layer registers those systems as tools for the AI.
MCP (Model Context Protocol) gives models a standard interface to tools, so each integration is not written from scratch.
How it works
- Connection and scope: a system account is created and the readable fields and permission scope are defined.
- Data mapping: the relevant record types are registered as tools for the AI layer.
- Usage: teams ask questions in natural language within their own permissions and get answers with source references.
Benefits
- Shortens the report request chain
- Provides access without copying data
- Deploys without replacing the existing system
Features
- REST/GraphQL connectivity
- MCP server definitions
- Tool-level authorisation
Use cases
- Connecting in-house software
- Integrating niche industry applications
- Granting tool permissions to agents
Data security is not a feature — it is a deployment decision.
In every project, access rights, data boundaries and logging are defined from the very beginning.
- With on-premise deployment data never leaves the corporate network
- Role-based authorisation: users only see data they are entitled to
- KVKK-compliant data processing approach
- Per-request logging and traceability
Other integrations
SalesforceAI-assisted querying across customer records and sales processes.
LogoNatural-language reporting on accounting and operations data.
MikroAI access to stock, account and invoice data.
ParaşütSummarising and reporting pre-accounting data.
ZirveAI-assisted analysis of commercial software data.
WhatsAppEmployee and customer access to the AI assistant through WhatsApp.