The last decade bet that value would concentrate in a single global ledger. We bet the opposite: that software should be agent-centric, deterministic, and content-addressed — reproducible rather than agreed-upon.

Agent Elmer is that bet, built: an engine for software that is provable by construction.

AI can now write software — and quietly get it wrong. Every serious organisation faces the same gap: the capability is real, the guarantees are not. The market needs a way to use AI without surrendering determinism, ownership, or control.

That is the gap Agent Elmer fills, at the exact moment it became urgent.

• A deterministic substrate. Content-addressed, replayable, fail-closed — not a wrapper over a model.
• Agent-centric architecture. No global singleton to centralise, censor, or collapse.
• A decade of conviction. The idea predates the hype cycle it is now squarely positioned for.
• Governance as topology. Security is the fractal shape, not a feature: rogue agents are structurally inert, every parent governs its children, and components hot-swap with zero downtime.
• Compounding components. Every piece created makes the whole more valuable, and harder to leave.

• Agent Elmer — the engine. Free core, Pro for teams, Enterprise for organisations. Open at the core, monetised at the edge.
• Creative Development — the platform on top: a deterministic software factory for everyone, with its own tiered subscription.
• The creator economy — later, components live peer-to-peer, and their makers hold rights and earn a share of reuse.

Recurring software revenue now; a network effect behind it.

• Now — Agent Elmer free core, public.
• Next — Pro tier; the Creative Development private beta.
• Then — the component network: peer-to-peer, with creator rights.
• Eventually — the open marketplace.