The Map Is the App: a workflow map language for building software

The Symbol Grounding Framework (SGF) introduced WML, a workflow map language that lets developers declare AI software as deterministic maps composed of fifteen versioned primitives, inverting the typical AI-coding approach by treating the model as one component rather than the substrate. The language aims to eliminate nondeterministic memory, non-restartable execution, and non-auditable reasoning by enforcing grammar-based control flow.

An engineering-grade grammar for declaring AI software workflows as maps. WML lets you declare AI software as a workflow map the "what" instead of prompting an agent to generate it the "how" . The map is the application; source code is a derived, ephemeral artifact. A workflow map composes fifteen versioned primitives call llm , read file , loop , if , assert , … connected by frozen contracts. The runner executes the map deterministically. If the map parses and the primitives are wired correctly, the generation behavior follows from the grammar — not from the mood of the model. Most AI-coding tools treat the model as the substrate: the model is asked to generate code, reason about code, and decide what to do next, all in free-form character space. This conflates model with infrastructure and produces three structural failure modes: nondeterministic memory, non-restartable execution, non-auditable reasoning. WML inverts this. The model becomes one primitive call llm among fifteen. Control flow lives in the map, not in prompts. Every action is versioned, timed out, and audited. The runner enforces the grammar, not hope. Start here: CLAIMS.md /SymbolGroundingFramework/SGF-manifest/blob/main/wml-spec/CLAIMS.md — the argument in 2 minutes Read the article: the-map-is-the-app.md /SymbolGroundingFramework/SGF-manifest/blob/main/wml-spec/the-map-is-the-app.md See also: What Makes WML a Formal Language /SymbolGroundingFramework/SGF-manifest/blob/main/wml-spec/what-makes-wml-a-formal-language.md WML is one component of its broader parent project, the Symbol Grounding Framework SGF . SGF is a multi-layer architecture for grounded machine meaning. It defines: Core Lexicon of sense-disambiguated concepts grounded in ~65 semantic primes- Synapses: hub-and-spoke event structures with 15 fixed semantic roles Wire protocol – a versioned, machine‑to‑machine message format that lets services exchange grounded semantics without prior integration contracts.- HFF: a wire format for sealed, signed, provenance-tracked meaning - AFP: an act protocol INFORM, QUERY, COMMAND, etc. with receiver sovereignty Omega : a formal governance grammar with 13 primitives and a deterministic Safety Kernel WML is a workflow-map language for composing AI software from primitives. See also: