Say you work on a platform team, and someone has asked you to shortlist a Multi-Cloud API. Two names rise to the top of the Multi-Cloud tag on APIs.io: Azure Arc and DoiT. You check the Kin Score, which measures the public API surface, not the product and not the company. Azure Arc scores 72.5. DoiT scores 72.3. Both land in the exemplar band, and they are 0.2 apart. As far as a ranking can tell you anything, you could flip a coin.
But you know the two are not the same purchase, so you keep reading. Azure Arc is Microsoft's hybrid and multi-cloud management service. It projects servers and Kubernetes clusters running outside Azure into Azure Resource Manager, so you can govern, patch, extend and monitor them with Azure tooling. DoiT sells DoiT Cloud Intelligence, a FinOps platform that unifies cost, usage and savings data across AWS, Google Cloud, Azure, Kubernetes and more than forty other clouds and SaaS providers. One is in the tag to control infrastructure, the other to account for it. Both records show a freemium, self-serve model with a try-now path and no trial.
So you look at the APIs. Azure Arc publishes 18 API pages carrying 79 operations, all of them on management.azure.com. Among them are the Machines API, the Connected Clusters API, the Extensions API and the Machine Run Commands API, plus 51 JSON Schemas and three Postman collections. DoiT publishes 49 API pages and 250 operations on api.doit.com, which are resource groups of one Platform API contract. They include Anomalies, Budgets, Allocations and Ava, its AI assistant. DoiT also runs a first-party remote MCP server at mcp.doit.com listing 90 tools.
There is one caveat you should hold onto here. The Azure Arc contracts were not derived by the catalog, but its MCP, skills and agentic-access artifacts were. Its agentic-access profile classifies 7 operations from one source file, against 79 on its pages. So the catalog says plainly that it is not setting Azure Arc's acting count next to DoiT's.
Next you go facet by facet. Azure Arc wins three and DoiT wins two, with discoverability tied at 75.0. The widest gap is operational transparency, 86.8 to 61.8 for Azure Arc. That covers status, changelog, rate limits and the signals a team needs to run against an API in production. Azure Arc also takes access clarity, 81.6 to 76.3, and contract governance, 22.0 to 18.2. You notice that governance at 22.0 and 18.2 is a weakness they share, not a win for anyone. DoiT takes contract quality, 63.5 to 56.4, which is the heaviest facet at 0.25, and developer ergonomics, 73.8 to 73.2.
On the six weighted facets alone, Azure Arc leads 66.5 to 63.6. Then the conditional layers move DoiT by 8.7 and Azure Arc by 6.0, mostly through the upsert check, where DoiT scores 61.1 to Azure Arc's 22.2. Both sit under the same Horizontal regulatory regime, so the comparison runs over the same facet set. Azure Arc's 0.2 lead is a 2.9-point base lead that DoiT nearly erases on that one conditional check.
Then you open the Agent Readiness score, and the tie is gone. DoiT is agent-native at 74.1. Azure Arc is agent-aware at 25.5. That is 48.6 points apart, in opposite bands. You read the dimensions and find the gap is not one thing but nine. DoiT has a verified MCP server, verified idempotency, documented reversibility, a dry-run mode, a described event surface, served delegated identity, verified protected resource metadata, dynamic client registration, and partial examples in its OpenAPI. Azure Arc has none of those on its record. What it does have is a bearer-auth contract, a verified rate-limit signal and error semantics the catalog derived.
The agent-native band is gated on idempotency and a stable error envelope, not just the score, and DoiT clears both gates verified. You also learn this is not an accident of being a FinOps company. The rest of the Multi-Cloud tag lights idempotency at 8.9%.
Now you can answer the person who asked. If the job is operating servers and clusters that live outside Azure from inside Azure, Azure Arc is the API, and its surface is the more transparent one to run against. If the job is putting an agent in front of cloud spend across many providers, DoiT is the clearer choice by a wide margin. It ships the MCP server, the identity plumbing and the safety signals an agent needs, from the company itself.
The post also turns the lens on both vendors' gaps. Azure Arc's agent number would move mostly with things that already exist elsewhere in Azure: a first-party MCP surface, idempotency and reversibility documented on the operations that create and delete machines and extensions, and protected resource metadata for the OAuth it already uses. DoiT's Kin Score would pass Azure Arc's if it fixed the facet both are weakest on. Contract governance, at 18.2, is a versioning and lifecycle story it has not yet told.
What you take away is simple. A 0.2 gap between two scores tells you almost nothing until you ask which job you are hiring the API for, and a second score can split what the first one joined.
Read the original: https://apis.io/2026/11/17/azure-arc-vs-doit-two-tenths-apart-and-forty-nine-points-of-agent-readiness/