AI engineering firm Solvd announced this week it has officially joined the Agentic AI Foundation (AAIF), a Linux Foundation project established to develop open, interoperable standards for agentic AI systems.
The membership puts Solvd alongside founding members Anthropic, AWS, Microsoft, Google and OpenAI in shaping the protocols that will govern how AI agents operate across enterprise environments – and comes at a moment when Europe is paying particularly close attention to who controls the infrastructure underlying AI.
AAIF was launched in December 2025 as enterprise AI deployments began accelerating rapidly. Just 23% of enterprises currently report using agentic AI at least moderately, but that figure is expected to rise to 74% within two years, with the global agentic AI market expanding from $7.6 billion in 2025 to a projected $10.8 billion this year.
As deployments scale, the absence of shared standards risks creating a fragmented landscape in which agents built on different platforms cannot reliably communicate or interoperate – a compounding problem that grows harder and more expensive to fix the further adoption advances without agreed infrastructure.
Solvd’s focus: MCP and open agent tooling
As a member, Solvd’s engineering leaders will contribute to AAIF working groups focused on the Model Context Protocol (MCP) – the open standard that defines how AI agents connect to external tools and data sources.
Originally developed by Anthropic and donated to AAIF in December 2025, MCP has become the default integration layer for enterprise AI with remarkable speed. The protocol reached 97 million monthly SDK downloads within 16 months of launch – a scale that took React approximately three years to reach – and now supports an ecosystem of over 10,000 active servers, with adoption from every major AI platform.
Solvd will also contribute to AGENTS.md, a markdown-based standard for giving agents project-specific context across repositories, and to open-source agent tooling more broadly.
“The technologies we use every day were all made possible by open, shared protocols – TCP/IP, HTTP, REST,” said Solvd CTO Skylar Roebuck.
“AI is the next technology revolution and the standards we set now will define what’s possible for years to come. We’re joining AAIF to help build that foundation alongside the industry’s leaders.”
A European dimension
For European enterprises and policymakers, the question of who shapes foundational AI protocols is not merely technical, but also a sovereign question.
The European Commission has made clear that dependence on non-EU technology providers in the digital sphere “reduces users’ choice, hampers EU companies’ competitiveness, and can raise supply chain security issues.”
That concern applies directly to the infrastructure layer of agentic AI: if the protocols governing how agents connect to enterprise systems are designed in closed, proprietary environments controlled by U.S. hyperscalers, European firms building on top of them inherit the same dependency risks the EU has been working to reduce – and outright avoid – across its digital economy.
AAIF’s governance model – a neutral Linux Foundation project with open working groups and multi-stakeholder membership – offers a structurally different answer to that risk.
It is, in fact, the same logic that underpins European enthusiasm for open-source software as a path to digital sovereignty: shared infrastructure, openly governed, which no single vendor can close off or monetize unilaterally.
The Amsterdam stop matters in this context. Solvd engineers will be on the ground at the MCP Dev Summit Europe in the Dutch capital on September 17-18, bringing the standards conversation directly into the European ecosystem at a moment when the region is simultaneously navigating AI Act implementation and trying to reduce structural dependence on foreign technology infrastructure.
How agentic AI protocols are designed will have direct implications for compliance, interoperability, and the degree of control European organizations retain as they adopt these systems.
Bridging research and production
For the standards-development process itself, Solvd’s value proposition rests on a combination that is relatively rare: deep academic credentials alongside direct production experience at enterprise scale.
The company has published over 100 research papers, employs more than 15 PhDs, and has contributed to NeurIPS, ICML, and CVPR – including a NeurIPS 2025 Best Paper Award winner. Its 700+ engineers simultaneously serve more than 100 enterprise clients across technology, fintech, e-commerce, hospitality, and banking.
That dual profile matters in a standards environment where the gap between theoretical design and real-world deployment can be significant. Gartner, for one, has warned that over 40% of agentic AI projects are at risk of cancellation by 2027 if governance, observability, and practical interoperability are not established.
Such are exactly the failure modes that robust standards are designed to prevent.
“We are excited to offer our experiences and research to this important effort to ensure robust and safe AI solutions are deployed,” said Solvd CEO Mike Hulbert.
Solvd was built through the acquisition of EastBanc Technologies and Tooploox, and describes its focus as the “last mile” of enterprise AI – the engineering work required to turn a working model into a reliable production system. Beyond Amsterdam, the company will also participate at the AGNTCon + MCPCon in San Jose on October 22-23.
Regardless, the foundational protocols being written in AAIF working groups today will shape the answer to a question Europe is already asking: who gets to define infrastructure that enterprise AI runs on? And will it be open enough for everyone to build on equally?
Featured image: Solvd via LinkedIn
Disclosure: This article mentions clients of an Espacio portfolio company.
