Gridscloud
Sign inGet started
What’s new

The engine keeps growing.

Recent additions to the Grid engine, newest first. Everything below is live in the hosted runtime and documented in the language docs.

  1. Aug 2026

    Symbolic mathematics

    Declare symbol x and compose it with ordinary operators into exact expressions. Canonicalize, simplify, expand, factor, differentiate; solve exactly with certified root intervals; approximate to any bit precision. Assumptions gate conditional rewrites — x/x only cancels when x ≠ 0 is known — and every transformation returns a replayable certificate.

  2. Jul 2026

    Live models

    Authored relational pipelines — filter, enrich, group, aggregate — are now maintained incrementally and exactly, across restarts and failures. Changed rows flow through to cells, rules, and dashboard tiles without recomputing the world.

  3. Jul 2026

    Native SQL with warehouse pushdown

    SELECT is source syntax, lowered to a typed relational plan — not a SQL string. Joins, CTEs, window functions, set operations, and DISTINCT, with pushdown to BigQuery, Snowflake, Redshift, and Databricks and the residual computed locally.

  4. Jul 2026

    Predicates, inference, and WHY

    Assert facts about model symbols, derive predicates from rules — predicate releasable means reviewed AND NOT blocked AND EVERY dependency IS :releasable — and reason over time, topology, and conceptual sets. MUST / MAY / CANNOT express modal constraints, and WHY(cell) explains any inferred answer with its exact evidence.

  5. Jul 2026

    Grammars and parse trees

    Define CFG, PEG, or regular grammars in source, generate parsers (LALR, Earley, packrat, DFA), and parse text into trees that are native values — ready for graph queries, predicate reasoning, pattern matching, and transformation.

  6. Jul 2026

    Games and mechanisms

    Declare players, strategies, and payoffs; get certified equilibria — pure Nash, zero-sum LP duals, correlated equilibria — plus dominance, Pareto, and welfare analysis, auction and mechanism checks with concrete counterexamples.

  7. Jul 2026

    Algebraic data types and inferred functions

    Payload-bearing choices — CHOICE Delivery = Pending | Delivered(at: date) — with exhaustive constructor matching, and source-authored named functions with full type inference: the compiler works out the types, no annotations required.

  8. Jul 2026

    Rules, hardened

    Admission control (SERIAL, bounded CONCURRENT), retention policies (LATEST, bounded QUEUE, durable LEDGER), DEBOUNCE / THROTTLE rate limiting, and required missed-run policies (SKIP MISSED / BACKFILL) on every schedule.

  9. Jul 2026

    APIs as imports

    Import an OpenAPI, AsyncAPI, GraphQL, WSDL, or gRPC schema as a versioned, digest-pinned package: USE "api:example/catalog@1.0.0" AS catalog. Operations become namespaced functions with declared pagination, retries, and webhook support.

  10. Jul 2026

    A 2,816-function catalog

    The built-in catalog now spans 51 categories — from math, statistics, and finance to logistics, geospatial, computer science, science, game theory, and computational biology — auto-generated from the runtime, so the docs always match the binary.

  11. Jul 2026

    An MCP server for AI agents

    Any MCP-capable agent can search the function catalog, read the docs and canonical examples, explain features — and validate and evaluate candidate models against the real runtime before returning them.

See it in contextCapabilities