AI agents are joining the workforce, and they're becoming part of the teams we serve.
At Middesk, we've always believed in meeting our customers where they are and anticipate where they’re going. For years, that meant building APIs, dashboards, and workflows that fit seamlessly into how compliance, product, and operations teams verify businesses, manage risk, and drive growth. But now those teams include AI agents.
To meet the moment, we’re building foundational infrastructure for this shift. Today we're releasing Middesk MCP and our GraphQL API to give agents the same direct access to business verification and risk intelligence that human teams have always had.
Middesk MCP: Business verification without code
MCP (Model Context Protocol) is an open standard that lets AI applications securely connect to external tools and data sources. With Middesk MCP, AI agents can now verify businesses and assess legitimacy and risk without writing a single line of integration code.
Before MCP, your engineering team had to build and maintain authentication flows, HTTP requests, retries, and error handling. Now you install the remote server, and agents instantly access Middesk's business intelligence and risk decisioning capabilities. No custom plumbing, no stitching logic. Your underwriting assistant, compliance workflow, or fraud detection system simply requests the verification it needs, and Middesk handles the rest.
GraphQL API: Efficient data for complex workflows
Business verification data is inherently graph-structured. Entity relationships form complex, multi-level hierarchies: businesses to principals, principals to addresses, entities to subsidiaries. Traditional API patterns don't reflect this well.
Our GraphQL API exposes business identity as a traversable graph. Define your data requirements declaratively and resolve entire relationship trees in a single request. What used to take 3-5 sequential API calls with client-side orchestration now executes as one optimized server-side query.
This is particularly valuable if you're managing your own orchestration logic or working with constrained engineering resources. Instead of building sequential call chains and data assembly, you write one query that describes exactly what you need. The API handles the complexity of fetching and assembling nested relationships.
LLM-based agents operate under strict token budgets, and over-fetching compounds rapidly across hundreds of API calls. GraphQL's field-level granularity eliminates payload bloat. Agents can introspect the schema, construct queries programmatically based on workflow requirements, and request only the fields they need. This reduces context window consumption by 60-80%, lowering inference costs and improving latency.
The strongly typed schema provides reliable guarantees and enables more sophisticated tooling. Agents can validate queries before execution, leverage type hints for intelligent field selection, and adapt to schema changes without brittle parsing logic.
Start building
We're launching with core business attributes: name, address, registration details, and key entity relationships. The schema will evolve as we incorporate feedback from early customers and expand the data available through GraphQL.
These releases are the foundation for how we're building going forward. We've always built infrastructure for our customers and their teams. Now we're building the infrastructure their agents need too.
GraphQL API and Middesk MCP are available now for early access. Get in touch to explore how they fit your workflow: info@middesk.com
Stay updated on new releases
We’re shipping foundational infrastructure for AI-native onboarding, fraud, and risk. Follow our product updates to get early access, demos, and release notes. See what we're building as we build it.
