Lexery - Unknowns Queue

Known unknowns: questions that block confident planning until someone pulls ground truth (dashboards, runbooks, stakeholders). Status here is inferred from repo + wiki ingest — not audited financial or customer data.

Pair with Lexery - Drift Radar (contradictions we can already see) and Lexery - Open Questions and Drift (long-form discussion).

Queue (investigation backlog)

1. Production deploy process

  • Question: What is the authoritative production deploy path (who triggers, from which branch, with which checks)?
  • Why it matters: ORCH rollouts, hotfixes, and trace fixes all need the same answer.
  • Clues in wiki: CI visible for portal and LLDBI brain-admin; full Brain deploy story is unclear.

After answer, update: Lexery - Deployment and Infra, Lexery - API and Control Plane.

2. Customer / user base size

  • Question: How many active users, seats, or orgs — and over what window?
  • Why it matters: Roadmap weighting (MM vs retrieval vs portal) and support staffing.

After answer, update: Lexery - Business Model, Lexery - Current State.

3. Monthly AI spend (aggregate)

  • Question: What is actual monthly LLM + embedding + reranker cost across environments?
  • Why it matters: Unit economics and whether “cheap path” work is ROI-positive.
  • Clues: Token usage appears in some reports; no single aggregate ledger in-repo.

After answer, update: Lexery - Cost Ledger, Lexery - Provider Topology (if split by vendor).

4. Staging environment

  • Question: Is there a staging Brain + Portal + LLDBI slice distinct from prod?
  • Why it matters: Safe ORCH shadowing and verifier changes before customer traffic.

After answer, update: Lexery - Deployment and Infra.

5. Import proposal → actual import

  • Question: When an import_proposal is approved, what systems run, who approves, and where is idempotency enforced end-to-end?
  • Why it matters: Prevents “proposal theater” without corpus movement.

After answer, update: Lexery - Import Proposal Loop, Lexery - LLDBI Surface.

6. seredyuk2077/Ukrainan-Lawyer-LLM vs public beta

  • Question: Is the private repo the same lineage as public beta, a fork, or abandoned parallel tree?
  • Why it matters: IP boundaries, licensing, and what can be open-sourced later.

After answer, update: Lexery - Repo Constellation, Lexery - Legacy Beta App.

7. Paying customers / revenue

  • Question: Are there paying customers today? If yes, pricing motion and churn?
  • Why it matters: Separates “beta traction” from “business validation.”

After answer, update: Lexery - Business Model, Lexery - Team and Operating Model.

8. Supreme Court case-law RAG integration status

  • Question: What is live in prod vs branch-only for case-law RAG?
  • Why it matters: Product promises vs retrieval reality.

After answer, update: Lexery - Branch Supreme Court Case Law RAG, Lexery - Corpus Evolution.

9. Memory pipeline production readiness

  • Question: Is LEX-160 class recall issue resolved in prod or only in dev experiments?
  • Why it matters: MM is user-visible; failures look like “the AI forgot.”

After answer, update: Lexery - Memory and Documents, Lexery - Linear Roadmap.

10. Team communication stack

  • Question: Beyond Linear, what are Slack / Telegram / email norms and escalation paths?
  • Why it matters: Incident response and external partner coordination.

After answer, update: Lexery - Team and Operating Model.

Working the queue

StepAction
TriageTag each unknown with owner and evidence type (dashboard URL, runbook path, stakeholder).
ResolveMove finding into a canonical wiki page; strikethrough or archive the row here.
Re-openIf infra changes, link from Lexery - Log so history is traceable.

See Also