Gap List
Nishad

Gap List — the demand-pulled roadmap

Capabilities the fleet needs but has not proven. Every row was raised by a real Chief-of-Staff decomposition that found no evidenced match at the bar — so the ledger grows only from real asks, never a speculative taxonomy. Each gap routes to the owning domain to author its test script.

6Open gaps
2Unstaffed (no script owner)
3Script in progress
1Rating candidates
Needed capability statement Raised by (Cortex decomposition) Owning domain Envelope Status
transcribe(inbound_call) via voice-runtime to WER≤8% within cloudno subject holds this at the bar QUORUM-1742 · "answer inbound Luxe calls" step 2 of 5 — unmatched antigravity-2
voice		 cloud script in progress
reconcile(qbo_transactions) via qbo-connector to exact within clouddeclared-only — never Lab-tested DEV-1794 · "close the monthly books" step 4 — unmatched codex-1
financials		 cloud unstaffed — no script owner
plan(warehouse_layout) via xr-vision to rubric≥4 within on-premWMS greenfield capability FEATURE-1353 · "stand up a warehouse" step 1 — unmatched opus-4
wms		 on-prem-connected unstaffed — no script owner
extract(kb_entities) via local-model to schema-valid within air-gappedcandidate rated 0.62 — below 95% gate FEATURE-1319 · "air-gapped KB refresh" step 3 — no confident match opus-1
kb / platform		 air-gapped rating candidates
draft(guest_reply) via hospitable-connector to rubric≥4 within cloudscript authored, awaiting first ratings QUORUM-1742 · "handle a guest message" step 3 — unmatched codex-1
luxe-ops		 cloud script ready — no runs yet
grounded-inference-with-attribution via kb-graph to cited within air-gappedcloud evidence exists; air-gapped does not count FEATURE-1404 · "cite sources, no external call" air-gapped install — cloud evidence N/A opus-1
kb / platform		 air-gapped script in progress

The loop closes here: gap → the owning domain authors a test script (reviewed by a different owner) → the Lab rates candidates → recorded to the ledger → the next decomposition matches it. A gap owned by no domain is the honest signal that this ask has no path yet — surfaced, not hidden.

What's on this screen

Built from: real tasks that Cortex broke down into steps and, for at least one step, found no model or agent with proven-enough evidence to do it.

On screenWhere it comes fromWhat it means to you
KPI tiles (open / unstaffed / scripting / rating)a count of gaps at each stage of getting provenwhere the roadmap is stuck — mostly unowned, mostly being tested, or mostly close to done
Needed capability statementthe specific skill nobody has proven yettells you exactly what capability needs to be built and tested next
Raised by (Cortex decomposition)the real task that hit this wall while being broken into stepsproof this is a genuine, demand-driven need — not a guess at future work
Owning domainwho is responsible for authoring the test for that skillwho to expect this from, or who to follow up with if it's stalled
Envelopewhere that skill needs to run (cloud, in-network, or fully offline)tells you which deployment restriction applies before anyone even starts testing
Status (unstaffed / scripting / rating candidates / candidate)where the gap sits in the proving pipelinewhat to do next — assign an owner, wait on the test being written, or push for a rating run