AGI, MCP Explained: 2026 AI Glossary for Founders
07 Jul 2026
The terms shaping AI conversations in 2026
As AI terminology continues to evolve, TechCrunch has published a glossary aimed at helping founders and industry watchers keep pace with the language driving the sector. Two terms stand out for early-stage founders: AGI (artificial general intelligence) and MCP (Model Context Protocol).
How OpenAI's Altman defines AGI
Rather than relying on a purely technical benchmark, OpenAI CEO Sam Altman has described AGI as the "equivalent of a median human that you could hire as a co-worker." This framing shifts the conversation from abstract capability thresholds to something more relatable — a comparison to average human job performance.
The report notes some ambiguity here: definitions like Altman's median-human framing may create inconsistent expectations among stakeholders, since "median human co-worker" is open to interpretation depending on the role, industry, or task in question.
MCP: from Anthropic to industry standard
The glossary also highlights the Model Context Protocol, a term with a clearer origin story:
- 2024: Anthropic introduces MCP.
- After 2024: Anthropic hands MCP over to the Linux Foundation.
- Following adoption: OpenAI, Google, and Microsoft adopt MCP.
This progression — from a single company's creation to an open-foundation-governed standard adopted across major AI labs — suggests MCP may be emerging as a common protocol for interoperability among AI systems.
That said, several details remain unspecified in the current reporting. There's no exact date for when Anthropic transferred MCP to the Linux Foundation, no specifics on how or when each of OpenAI, Google, and Microsoft adopted it, and no technical explanation of what MCP actually does beyond its name and adoption history.
Why founders should care
For founders building AI-powered products, these terms carry practical weight:
- MCP's broad adoption by OpenAI, Google, and Microsoft may indicate it's becoming a de facto standard. Founders whose products need to interoperate with these platforms could benefit from understanding MCP early, though the lack of technical detail in current reporting makes it hard to assess implementation specifics right now.
- Anthropic's move to hand MCP to the Linux Foundation may signal a broader trend toward open governance of AI infrastructure standards — a dynamic founders may want to monitor for future compliance or integration considerations.
- Altman's AGI framing could shape how leading labs — and by extension, customers, investors, and partners — set expectations for AI capability benchmarks. Founders pitching AI products may find it useful to understand this framing, since it likely influences how stakeholders judge "how smart" a given AI system needs to be.
A caveat on sourcing
It's worth noting this glossary draws on a single source for its definitions, which could limit the completeness or nuance of the terms as founders apply them in their own communications. Founders relying on these definitions in pitches, documentation, or product positioning may want to cross-reference with primary sources — such as Anthropic's MCP documentation or OpenAI's public statements — before treating these as settled industry consensus.
The bottom line
MCP appears to be gaining real traction as a cross-platform standard, given its adoption by three of the industry's largest AI labs. AGI, by contrast, remains more conceptually fluid — useful as a communication shorthand, but not yet a fixed technical benchmark. Founders navigating both fundraising conversations and product architecture decisions may find it worthwhile to track how these terms solidify — or continue to shift — over the coming year.