Non-profit technology advocacy group The MACH Alliance has forged new connections for agent-to-agent AI services. Standing up for the proposed benefits of Microservices-based, API-first, Cloud-native SaaS and Headless technologies, alliance members include firms from PayPal to MongoDB to Google Cloud. The group wants to “clarify the how” in agent-to-agent ecosystem interconnections through connected composable architectures.
Certifications and governance
The Alliance is refreshing its positioning to drive interoperability through new certifications and governance standards. This is agent-to-agent AI, so the usual suspects are all here (in a good way) including Model Context Protocol (MCP), Google’s gift to agentic interconnection agent-2-agent (A2A), Open Data Model (the latter being technology that maps, cross references and indexes all legacy system data to a “semantic hub” in the form of an authoritative data model) as well as AI-connected certification.
According to OpenDataModel.com, “The use of the data governance functions of the Open Data Model results in the creation of a multi-dimensional metadata matrix – technically a ‘data metadata repository’ i.e. a database in which all the chaos of legacy systems data is replaced by a rich, meticulously organised knowledge base that may be interrogated on-line to rapidly satisfy any data archaeology need”… so, in other words, just the kind of fuel needed to forge agentic connections diligently.
The group is launching programs, including the MACH AI Exchange and FutureMACH. It is publishing what it promises are practical tools, reference architectures and business transformation guides. Why? Because it is aiming to make agent-to-agent ecosystems a trusted, open and scalable thing.
Isolation equals desolation
“AI is already reshaping how enterprises operate, from decision-making to customer interaction and its impact is accelerating. But its true value will only emerge through connected, open and interoperable ecosystems, not isolated tools,” said the alliance, in a press statement.
MACH members have further commented on their work in this space, saying that they are focused on “data and systems agility” because this is the crux of making generative AI and agents able to function across silos.
What is agentic negotiation?
Real-time interaction patterns are at play here because agentic systems require real-time trust, governance and “negotiation”, which (the group claims) can be enabled by MACH standards.
When the industry talks about agent-to-agent negotiation, it is a reference to the point where two (or potentially) systems of automation come together and interact with the aim of coordinating their individual knowledge bases and data functionalities to coalesce and form a higher level AI function that its greater than the sum of its parts… this can be especially challenging when agent negotiation has to straddle and manage inter-agent dependencies in complex scenarios like supply chains or e-commerce.
The MACH team says it is focused on composable AI ecosystems, where no single player can necessarily build the future of AI ecosystems i.e. this is basically insurance against some form of AI hegemony in the future. Finally, there is historical validation here i.e. we are told that enterprises that are “well along in their MACH journey” are twice as likely to successfully deploy AI, with 77% achieving success compared to just 36% for those new to MACH.
Image Credit: Google Gemini