Solution · AI-Native Development

Agents as language primitives

Mainstream language abstractions predate AI instructions, tool-calling and agent loops. So all three arrive as libraries — each with its own conventions, versioning and orchestration model.

The structural argument

This is a bet, and we state it as one

Every significant shift in what software does has eventually changed the programming model rather than being absorbed permanently into libraries. Concurrency did. Memory safety did. Our position is that agentic execution follows the same path.

We could be wrong. If agent frameworks mature into a stable, universal layer, the library approach wins and this bet does not pay off. We think the abstraction belongs lower down, and we have built accordingly.

Concretely

What a language-level primitive looks like

// tools and agents are keywords, not imports

@tool "Search the web"
function search_web(query)
    return http get "https://example.org/search?q=" + query
end

agent researcher
    tools: search_web, summarize
end

run agent researcher with "Research NC and summarize"
The tool registry, the reasoning loop and the model call are part of the runtime. There is no orchestration library to install, version, or keep in step with a provider SDK. You can verify this by reading the language reference — these are keywords in the grammar.
Business effect

Why this matters commercially

Design decisionOperational effectBusiness effect
Agents in the grammarAgent code is readable by anyone who reads the languageReview and handover stop depending on framework expertise
Provider-independent AI instructionSwitching model provider is configuration, not a rewriteReduced lock-in to any single AI vendor
Runtime owns the loopOne execution model instead of severalFewer failure modes to understand and operate
Limitations

The honest counter-argument

A new language is a real cost

Tooling, hiring, ecosystem and institutional knowledge all start from zero. For most organisations that is the deciding factor, and it should be weighed honestly.

Frameworks are catching up

Agent frameworks improve quickly. The gap we are addressing may narrow. We think the abstraction level still matters; we do not claim the question is settled.

Judge it by reading the code

The agent example above is about fifteen lines. Compare it with the same thing in your current stack.