"Vox bell-curve strategy"

Vox bell-curve strategy

Program status

  • status: in_progress
  • scope: center-of-bell-curve app software
  • design_center: common app software first, with strong AI-generation ergonomics and explicit escape hatches

Target software categories

Vox is optimizing for:

  1. CRUD and line-of-business web apps
  2. internal tools and operator consoles
  3. content, admin, and research workflow apps
  4. API-backed dashboards and portals
  5. automation and background job systems
  6. AI-assisted application scaffolding, repair, and orchestration

Non-goals

Vox is not currently trying to become:

  • a universal systems language
  • a framework-neutral frontend platform
  • a first-class host for arbitrary Rust or JS APIs
  • a scientific-computing language
  • a multi-frontend-target language before WebIR owns the current web path

Product lanes

Use these lane ids in contracts, docs, command metadata, examples, and future dashboards:

product_laneMeaningTypical surfaces
apptyped web app constructionbuild, run, island, WebIR, AppContract
workflowbackground work, automation, durable-ish task flowsscript, populi, workflow runtime
aimodel generation, eval, review, orchestration, speechmens, review, dei, oratio
interopapproved integration surfaces and escape hatchesopenclaw, skill, bindings, wrappers
datadatabase and publication workflowsdb, codex, scientia
platformpackaging, install, compliance, diagnostics, secretspm, ci, clavis, doctor

Ranking model

Every bell-curve addition should score against the same dimensions:

DimensionWeightQuestion
bellCurveReach30How many common app tasks does this unlock?
llmLeverage25How much prompt/repair burden does it remove?
surfaceStability20Does it fit current IR, registry, and runtime boundaries cleanly?
implementationRisk15What compiler/runtime/docs migration risk does it introduce?
driftReduction10Does it eliminate duplicate semantics or conflicting docs/code?

Proposal template

Use this checklist for stdlib, interop, workflow, and measurement proposals:

FieldRequired content
laneone product_lane from the table above
user_problemnarrow statement of the common task being improved
preferred_boundaryWebIR, AppContract, RuntimeProjection, builtin registry, approved binding, or docs-only
fallback_escape_hatchhow uncommon cases work without broadening the main surface
rankingscore all five ranking dimensions
semantics_stateimplemented, partially_implemented, planned, or docs_only
drift_riskwhat could diverge if the proposal lands incompletely
acceptancetests, docs, and contract gates needed before release

Promise language

All docs in this program should explicitly label one of these states when a surface is easy to over-claim:

  • implemented semantics
  • planned semantics
  • language intent
  • escape hatch

This is especially important for workflows, frontend emission ownership, and interop claims.