The Race — Building Democracy in the Age of AI. People race toward a shared finish line.
The Race

Policy tool · See record

Interoperability requirements for agents and tools

What the tool does

Require compatible interfaces so an AI assistant can use tools and business systems without every connection being rebuilt for one supplier. Separate the assistant, execution environment and system of record where possible, while keeping permissions and action controls intact.

Current status

Partly in use. Operating products show that some layers can be separated; they do not establish complete supplier neutrality or portability of every workflow. A shared protocol is useful only when customers can practically substitute components.

Where it has been used

ServiceNow’s external assistant connections

Added by

ServiceNow’s MCP Server Console documentation describes governed access to ServiceNow functions from external assistants using the Model Context Protocol, a common interface for connecting AI systems to tools. A customer can change the assistant client while ServiceNow continues to control the underlying workflow platform.

Read more about this case →

OpenAI’s Agents API

Added by

OpenAI’s Agents API announcement describes managed agent orchestration with a Codex harness and a choice of execution environments. It separates where an agent runs from parts of the managed service. OpenAI still controls its API; this does not establish complete portability of state, identity or workflows.

Read more about this case →

BetaNYC’s civic-data connectors

Added on

BetaNYC provides civic-data connectors that compatible AI assistants can use through the Model Context Protocol. Its City Council connector was released on May 19, 2026 with MIT licensing, allowing an independently deployable interface to existing government APIs. Use still depends on clients, keys, data freshness and accuracy; public code does not remove those dependencies.

Read more about this case →

Personal Agent Protocol

Added on

Personal Agent Protocol's public documentation now identifies draft version 0.1, updated October 9. Sierra announced the initiative on October 6 with Meta and other participating businesses. The draft describes how a user's AI assistant could identify itself to a business and ask to perform permitted tasks.

A business publishes a standard discovery file explaining its interfaces. A person can then authorize an agent with limited permissions through an authorization process, rather than handing over a full account password. The specification supports interaction through websites, APIs or business agents, with a route to human help. These are draft specifications: broad deployment, reliable cross-provider operation and actual purchasing permissions have not been demonstrated by their publication. announcement and participants. Draft permissions/interoperability, not demonstrated deployment or independence.

Read more about this case →

Agent-Native’s interchangeable application layer

Added on

Agent-Native offers available MIT application code with model, database and hosting choices. This is an inspectable design example, not an adopted public interoperability mandate or independently proven migration.

Read more about this case →

Proof’s agent identity wallet

Added on

Proof released x401 on October 15, 2026 as an identity wallet, open web protocol and assistant connection. A merchant can request signed evidence of identity and transaction terms, then verify whether the instruction still matches the person’s approval. Proof is a commercial credential issuer; promised neutral protocol governance is not already established.

Read more about this case →

GovWare’s agent-commerce demonstration

Added on

Separately, an October 14 GovWare demonstration, described in an October 15 technical account, linked discovery, identity, authorization, communications, payments and audit across platforms. It used a controlled flight-purchase scenario, not evidence of general production commerce or actual settlement at scale. The technical paper describes the architecture.

Read more about this case →

Britain’s proposed Microsoft competition scrutiny

Added on

On October 16, the Competition and Markets Authority proposed designating Microsoft as having “strategic market status” in business software. That designation would make the company eligible for tailored competition interventions under Britain’s digital markets regime. The proposal covers an ecosystem including operating systems, productivity, security, server and database software. It is not a finding that Microsoft has broken the law.

The accompanying roadmap identifies possible work on cloud licensing, software integration and whether customers can choose competing AI services alongside Microsoft products. For example, an organization’s choice of office software can constrain its choice of assistant even when another model performs better. Neither designation nor specific conduct requirements have been adopted. Comments close November 13 at 5 p.m.; the statutory designation deadline is February 13, 2027.

Read more about this case →

Britain’s review of cloud switching

Added on

A separate October 16 CMA report reviews actions AWS and Microsoft took between March and October. Its detailed account is more qualified than headlines suggesting that both companies have already connected all their clouds to rivals. AWS–Google interconnection is launched; Microsoft–Google remains a forthcoming preview, and UK AWS–Microsoft connections remain under development. The report also identifies updated switching-fee policies and operating channels through which customers and competitors can request interoperability.

The regulator finds limited uptake so far and continuing switching barriers. It will work next on translating identity/access permissions between clouds and moving applications packaged in Kubernetes, a common deployment system that can still depend on provider-specific services. These are future work areas. Voluntary engagement is not an enforceable order requiring every requested connection or a guarantee of seamless migration.

The Race mechanism is practical exit leverage. This concerns cloud switching, rather than designation of Microsoft’s business-software ecosystem. The report combines older launches, recent implementation and future promises; October 16 is the reporting date, not the start date of every customer right.

Read more about this case →

Four agent projects move to foundation stewardship

Added on

The Agentic AI Foundation announced four project additions on October 22: Agent Skills, Open Responses, OpenEnv and OpenSandbox. Anthropic identifies transferred stewardship assets and allows contributions without foundation membership. Common interfaces may ease switching, but detailed decision and appeal rules remain unverified; stored customer data, credentials and permissions do not move automatically.

Read more about this case →

SAP opens an inbound route for outside agents to call SAP agents

Added on

SAP's October 27 architecture announcement describes October general availability of inbound Agent-to-Agent access through SAP Agent Gateway. A customer can configure an outside AI client to discover a SAP agent, delegate a task and receive the result in that client. For example, an outside assistant could request work from a SAP sourcing agent while SAP still performs the business operation. This is an illustrative workflow, not a verified customer outcome.

Discovery does not grant permission: customers must configure access, authentication and the named user's authority. The available direction is inbound. Outbound delegation and further MCP connections remain future capabilities. We have not verified pricing, equivalent entitlements across clients or full migration of business context. Reference architecture.

Read more about this case →

Source snapshot: 2026-10-28. Individual event and verification dates remain as recorded above. Inclusion alone does not assign a score.