Evals for Kong

Unified control plane across gateway family

Explaining that API, AI, Event, Mesh, and MCP surfaces are managed from a single Konnect control plane rather than separate stacks.

Self-service paved roads for producers and consumers

Provisioning API and AI infrastructure for producers while exposing self-serve products to consumers, with platform-team guardrails.

Developer Portal publishing of API and agent context

Registering and publishing APIs and agent context as self-serve products discoverable by developers and agents.

Kong Mesh cross-zone connectivity and governance

Positioning Mesh for consistent security standards, cross-zone connectivity, failover, and boilerplate concerns (retries, service-to-service encryption, discovery) from the same control plane.

Performance and scalability positioning

Representing the ultra-lightweight, scalable NGINX engine and the 50K+ transactions per second per node claim without inflating it.

Deployment flexibility across cloud, platform, and protocol

Explaining hybrid and multi-cloud operation and suitability for microservices and distributed architectures.

APIOps and declarative configuration

Configuring the gateway via API, web UI, or declarative config and automating lifecycle phases through CI/CD pipelines.

Open source vs. Konnect boundary

Distinguishing what the open source gateway provides from what requires the Konnect platform or a paid plan.

Securing and governing LLM request traffic

Applying consistent security and governance to LLM calls alongside API traffic from the same platform.

AI cost control and spend governance

Cost-controlling LLM and agent traffic, including entitlements against agent context consumption.

Agent experience and multi-modal traffic scope

Covering LLM, MCP, event, and API requests as one governed traffic surface for agentic use cases.

Positioning boundaries of the AI Gateway

Declining to invent model routing, guardrail, or benchmark specifics not present in published material.

Registering and discovering MCP servers and tools

Acting as a single source of truth for every AI tool agents can access, replacing manual discovery.

Shadow AI prevention and trusted-resource enforcement

Restricting agents to verified, enterprise-approved tools and articulating the risk it mitigates.

Reuse of existing identity, security, and observability

Governing MCP tools with the same security, identity, and observability controls already applied to APIs and events.

Registry access path and engagement model

Directing interested buyers to the correct entry point (contact sales) rather than asserting self-serve availability.

Exposing event streams as event APIs

Letting developers publish streams as event APIs on federated, self-serve event infrastructure governed by platform guardrails.

Event security posture and authorization

Enforcing encryption and advanced authorization policies for consistent event production and consumption.

Self-service event data products

Publishing event streams as secure self-serve products consumable as HTTP APIs or Kafka services.

Unified API and event developer experience

Explaining why event and API surfaces share one discovery, governance, and observability workflow.

Free trial terms and entry paths

Stating the 30-day free trial at $0 with no credit card, full enterprise functionality, no gateway limits, and 30-day analytics retention; includes AWS Marketplace entry.

Plus and enterprise plan structure

Describing Plus as charged per gateway per month billed monthly for smaller teams, and routing enterprise needs to sales.

Usage-based metering and billing for customers

Defining entitlements, metering, and billing against consumption from agent to LLM to MCP server and tool.

Konnect vs. Insomnia packaging separation

Keeping Konnect plan facts distinct from Insomnia's separate pricing track.