'%20fill='%23FB923C'/%3e%3c/svg%3e)
Functions/MCP
Model Context Protocol for property data
MCP makes Kliva's Nordic property data available as secure context in MCP clients like Claude, ChatGPT, and Microsoft Copilot. No copies, no insecure exports, and access control on every call.
Vision · MCP
The hub for future AI development.
AI does not replace real estate systems; it enhances their value. AI quality depends on the connected data it uses. Kliva MCP is the neutral data and agent layer where industry AI connects:Verified data in, trusted agents out.
01 — Connected data
Verified real estate data in one hub
Ownership, rents, transactions, climate and company data — always with a stated confidence level and source.
02 — The CRM hub
The place where all information converges
Pipelines, contacts, property relations, and processes — integrated with data and AI flows.
03 — AI / MCP gateway
Agent layer on top of the hub
A neutral protocol layer allowing agents and apps to build on Kliva — live today.
"The more AI, the more critical a central, connected platform becomes."
Each connection is individually secured. Auth0 binds every call to a verified identity with assigned rights — no shared tokens, no anonymous sessions.
What MCP enables
Fem datadomäner — i naturligt språk.
Transaction & market
Comparable deals, yield trends, and benchmarks across the Nordics.
The built environment
Properties, buildings, ownership, technical, and spatial attributes.
Early stages
Zoning plans, building permits, and municipal decisions – via Byggfakta.
Tenant intelligence
Demographics, visitor flows, and establishment analysis.
Climate, risk & environment
Climate benchmarks, exposure, and taxonomy-aligned analysis.
Why graph + AI belong together
A traditional relational database forces properties, owners, detailed development plans, and transactions into separate tables that must be joined for every query. The more hops — from property to owner to group to other holdings — the slower and more fragile the result.
This is also why the model is the right match for AI. Large language models reason in entities and relations — the same structure as the graph. When an agent queries via our MCP server, it receives nodes and edges rather than table rows it must piece together. The context becomes smaller, answers more traceable, and the risk of hallucinations lower.
Silver & gold — never raw data
All data served by MCP has passed Kliva's quality pipeline: cleansed, validated, enriched.
Auth0 per user
No data leaves Kliva without verified identity and assigned rights.
Data in Sweden · ISO 27001
The entire data flow is within Swedish territory and enterprise security frameworks.
MCP Security
Controlled access, not open data exposure
Kliva's MCP support is designed around established security patterns: OAuth-based authorization, strong authentication with MFA, and clear limits on the context an AI client may use.
Verified Identity
Each connection is initiated via an OAuth flow, Auth0, and MFA before an AI client can request context from Kliva.
Least Possible Access
Access is limited per user, client, and purpose. The client receives only the context needed for the current query.
Traceable Usage
Calls can be logged, linked to the correct user, and audited afterwards for compliance and incident management.
MCP Compatibility
Kliva data for your AI assistants
Model Context Protocol allows Claude, ChatGPT, Copilot, and other MCP-compatible clients to retrieve the correct property context from Kliva — with permissions, logging, and control for each call.
Security Across the Protocol
Auth0, MFA, and Policy-Driven Access
OAuth-based authentication for MCP clients.
MFA and permissions before data leaves Kliva.
Traceable calls, scopes, and data integrity per user.
Confidence
Verified for machines
Varje svar från Kliva MCP bär samma källhänvisning och konfidensnivå — Guld, Silver, Brons — som i plattformen. Era agenter gissar inte; de citerar riktig data.
Gold
Directly from the primary source — registry data, title deed transfers, current zoning plans. Verified and referenced.
Silver
Derived from multiple sources or with a known delay — clearly marked with the underlying data.
Bronze
Modeled or estimated — with stated assumptions, never presented as absolute fact.
FAQ
Common Questions About MCP
What you get
All key information, gathered in one feed
MCP kopplar ihop data och analys så att teamet kan gå från insikt till beslut med mindre handpåläggning.
Works with Claude, ChatGPT, and Copilot
Standardized Model Context Protocol
Correct property context directly in the AI flow
Auth0 and MFA over the MCP protocol
Permission-controlled scopes per user
Logging and data integrity in each call
'%20fill='%23FB923C'/%3e%3c/svg%3e)