Handbook

Forge — meeting model (operational)

Purpose: Preserve the minimal memorable set of Forge meetings, how they are delivered (human-only, hybrid, Versona-assisted), and what accountability requires—without inventing a parallel taxonomy. Meeting is the primary…

Scrum as memory aid: Forge uses a similar shape (short daily sync, time-boxed planning, review, retro, gate) but does not copy Scrum mechanics verbatim—see iteration language (Ore → Ingot → Spark → Charge) in Forge — major processes & flow maps.

Charge: Meetings are not modeled as items inside Charge. Charge is today’s selected Sparks; the daily sync confirms Charge—see Daily operations.

Artifacts & decisions: How durable artifacts, owners, and meetings connect across PDLC/SDLC — Forge — artifact and decision model.

1. Meeting delivery modes

Meetings use delivery modes, not separate object types:

Mode Meaning
Human-only Facilitation and accountable decisions are human-led in the room or on a live call.
Hybrid Humans in the loop; Versonas prepare drafts, checklists, or §5-shaped discipline passes that humans review and sign.
Versona-only Session used to prepare, simulate, synthesize, or draft (e.g. async package before a human sign-off). Not a substitute for human accountability on binding decisions—see §6.
Meeting In default cadence?
Refinement (Ore → Ingot) Yes
Planning (Ingot → Sparks) Yes (iteration start)
Daily sync Yes (every working day; short)
Review (inspect increment + evidence posture) Yes (iteration end)
Assay Gate (release decision) Yes when a release candidate exists
Retrospective Yes (regular, iteration end)
Ad hoc decision meeting Only when real decision pressure exists—not a standing weekly slot

Ore intake remains continuous, not a scheduled meeting.

3. Refinement vs Planning

Question Answer
Default Refinement is its own meeting. Different primary question: shape what is plannable (Ore → Ingot) vs commit how much and which Sparks (Planning).
Conditional merge Very small teams or very low Ore throughput may use one combined calendar block (“Planning + refinement”) only if Ingots still satisfy Definition of Ready before Sparks are committed—you merged the schedule, not the gate.
Never Treating refinement as always part of Planning—this collapses backlog shaping into scope commitment and invites thrash.

4. Meeting matrix (delivery modes)

Meeting Trigger / cadence (lean) Timebox (typical) Human-only Hybrid Versona-only
Refinement 1× weekly (2× if Ore flow is high) ≤ 1 h ✓ ✓ ✓ (draft; Ingot promotion human-signed)
Planning Start of 1-week iteration ≤ 2 h ✓ ✓ ✓ (decomposition drafts; scope human-signed)
Daily sync Every working day 15 min ✓ ✓ ⚠ prefer assistant, not sole chair
Review End of iteration ≤ 1.5 h ✓ ✓ ✓ (prep; readiness human)
Assay Gate When candidate + evidence package ready ≤ 30 min ✓ ✓ ✓ (checklist; release human)
Retrospective End of iteration ≤ 1 h ✓ ✓ ✓ (themes; experiments human-owned)
Ad hoc decision Decision pressure only 30–90 min ✓ ✓ ✓ (options; decision human)

Daily sync: Versona-only is a poor fit as the team event; use Versona to support a human-led standup, or stay human-only when presence is required.

5. Per-meeting specification

Each subsection uses the same field names. Inputs means tracking-spine material (Ore, Ingots, Charge, logs)—not a separate “artifact object” taxonomy.

5.1 Refinement (Ore → Ingot)

Field Detail
Name Refinement
For what Turn selected Ore into Ingots plannable with clear acceptance and evidence expectations.
Why Without shaping, Planning commits guesses; rework rises.
Primary decision question Which Ore becomes an Ingot (ready), merged, Banked, or rejected—and on what terms?
Trigger / cadence Standing: 1× weekly (2× if intake is heavy).
Suggested timebox ≤ 1 h; ~10–15% iteration capacity on refinement over the week.
Allowed delivery modes All three; Versona-only may draft Ingots; human signs promote/Bank/reject.
Required participants Product hat (R), Engineering hat (R), Challenge hat (O — Versonas at decision points).
Human owner Product (value/scope framing); Engineering (feasibility band).
Versona owner/support Discipline Versonas by risk (architecture, security, compliance).
Required inputs Ore backlog, product context, constraints, discipline knowledge links.
Recommended pre-work Ranked Ore list; rough sizing; known dependencies.
Agenda / flow (1) Top Ore (5–10m) → (2) Problem/value/constraints + feasibility (15–20m) → (3) Versona time-box on high-risk items (10–15m) → (4) Decisions + Ember Log (5–10m).
Created or updated Ingots; Ember Log; optional forge-logs/ notes.
Decisions expected Promote / merge / Bank / reject; recorded assumptions.
Exit criteria Each promoted Ingot meets Definition of Ready (problem, acceptance, constraints, effort band, phase affinity).
Anti-patterns Detailed iteration commitment here; skipping Challenge on risky items; “refinement” that is secretly Planning.
What must never happen Binding iteration commitment or Spark pull in this meeting (belongs in Planning).

5.2 Planning (Ingot → Sparks)

Field Detail
Name Planning
For what Decompose Ingots into Sparks, sequence for risk reduction, fit scope to capacity for the iteration.
Why Aligns execution with capacity; makes trade-offs explicit before the week runs.
Primary decision question What is in scope for this iteration—and in what order—given capacity and risk?
Trigger / cadence Start of each iteration (default 1 week).
Suggested timebox ≤ 2 h per 1-week iteration.
Allowed delivery modes All three; Versona may propose Spark breakdown; humans commit scope.
Required participants Product (R), Engineering (R), Governance (O).
Human owner Product + Engineering (scope content); Governance (policy/evidence fit).
Versona owner/support PM/architecture/compliance Versonas as needed.
Required inputs Ordered Ingot backlog, capacity, past metrics, Definition of Done.
Recommended pre-work Ingots ready; optional Spark candidates drafted.
Agenda / flow Focus proposal → decompose → sequence by risk/learning → capacity + margin → Versona needs before risky Sparks.
Created or updated Sparks with phase prefixes; iteration scope; Charge seed for day one.
Decisions expected In/out list; ordering; explicit “not this week.”
Exit criteria Spark quality check passes; scope fits with slack; Versona needs flagged.
Anti-patterns Ore→Ingot work here; 100% capacity fill; hiding dependencies.
What must never happen Treating Charge as the planning container (Charge is daily execution, confirmed at sync).

5.3 Daily sync

Field Detail
Name Daily sync
For what Execute and unblock: confirm today’s Charge, hats, blockers, Banking.
Why Short loop prevents silent drift.
Primary decision question What is today’s Charge—and what must change (Banking, help, follow-up)?
Trigger / cadence Every working day.
Suggested timebox 15 minutes (strict).
Allowed delivery modes Human-only or hybrid preferred; Versona as support (draft Charge, blocker summary), not accountable chair.
Required participants Engineering (R); Product (O); others if blocked.
Human owner Engineering (flow); Product (priority/Banking prompts).
Versona owner/support Optional: prep of blocker list—does not replace facilitation accountability.
Required inputs Current Charge, iteration Sparks, blockers, Ember Log.
Recommended pre-work Spark states updated; blockers listed.
Agenda / flow Yesterday → today’s Charge → blockers/Banking → hat → Versona needs → schedule time-boxed follow-ups (do not expand this meeting).
Created or updated Charge for the day; follow-up invites; Ember Log if decisions.
Decisions expected Banking/unbanking; who pulls which Spark; whether a follow-up is needed.
Exit criteria Everyone knows Charge and next touch on blockers.
Anti-patterns Status theater; problem-solving in the 15m; extra standing meetings that duplicate without new decisions.
What must never happen Modelling the meeting as an item inside Charge; letting this become a 45m project meeting.

5.4 Review

Field Detail
Name Review
For what Inspect the increment, assess evidence quality, gather feedback, judge Assay readiness (not the final ship verdict unless policy merges sessions).
Why Connects building to expectations before release pressure peaks.
Primary decision question Does the increment meet intent—and are we ready to pursue release evidence / Assay?
Trigger / cadence End of iteration.
Suggested timebox ≤ 1.5 h.
Allowed delivery modes All three for prep; stakeholder-facing demo usually human or hybrid.
Required participants Product (R), Engineering (R), Challenge (R), Governance (R), stakeholders (O).
Human owner Product (acceptance narrative); Engineering (demo truth).
Versona owner/support Discipline Versonas; evidence summaries.
Required inputs Completed Sparks, increment, Assay criteria, Ember Log.
Recommended pre-work Demo script; evidence index; known gaps.
Agenda / flow Demo → acceptance → discipline review → evidence vs Assay criteria → readiness (distinct from Assay pass/fail).
Created or updated Feedback; backlog updates; Assay readiness notes.
Decisions expected More evidence vs scope adjust vs proceed toward Assay.
Exit criteria Clear readiness position; gaps with owners.
Anti-patterns Slide-only review; skipping evidence; lowering Assay standards to match demo mood.
What must never happen Confusing readiness with Assay Gate without an explicit evidence package and governance path.

5.5 Assay Gate (release decision)

Field Detail
Name Assay Gate
For what Evidence-based release decision—strict question, separate from Review in Forge — ceremonies & events (prescriptive).
Why Keeps “looks good” separate from “releasable under policy.”
Primary decision question Is this candidate releasable now—with defined evidence—yes or no?
Trigger / cadence When release candidate + evidence package exist (often same week as Review).
Suggested timebox ≤ 30 min.
Allowed delivery modes All three for pre-checks; pass/fail and exceptions are human (Governance + Product + Engineering per prescriptive doc).
Required participants Governance (R), Product (R), Engineering (R).
Human owner Governance (policy bar); Product (user/value risk); Engineering (technical truth).
Versona owner/support Compliance/security Versonas may pre-audit; output feeds humans.
Required inputs Tests, acceptance, risk checks, performance as configured, Ember Log, rollback/support readiness.
Recommended pre-work Evidence index complete; gaps explicit.
Agenda / flow Evidence walkthrough → gap decision → pass/fail with remediations → release notes handoff.
Created or updated Release decision; release notes draft; Ember Log for risk acceptance.
Decisions expected Ship, hold, or ship with explicit conditions.
Exit criteria Unambiguous outcome; owners for follow-up Sparks if any.
Anti-patterns “Ship anyway”; Assay as second demo; rubber-stamping Versona text.
What must never happen Versona-only “release approval” with no human sign-off on accountability.

5.6 Retrospective

Field Detail
Name Retrospective
For what Improve the system of work: metrics, Ember Log, Versona effectiveness, experiments.
Why Without it, friction repeats; learning does not become Ore.
Primary decision question What will we experiment with next iteration—and who owns it?
Trigger / cadence End of iteration; ≤ 1 h.
Suggested timebox ≤ 1 h.
Allowed delivery modes All three; themes may be Versona-synthesized; committed experiments need human owners.
Required participants All hats (R).
Human owner Rotating facilitator; Governance ensures follow-through to directives when needed.
Versona owner/support May cluster themes from logs; does not replace psychological safety and ownership.
Required inputs Iteration metrics, Ember Log, Versona trends, team experience.
Recommended pre-work Data pulled; anonymized themes optional.
Agenda / flow Metrics → Ember Log → Versona value/noise → insights → 1–3 experiments with owners/dates → new Ore where useful.
Created or updated Experiments; new Ore; optional directive updates (see retro section in Forge — ceremonies & events (prescriptive)).
Decisions expected Which experiments; what to log as debt.
Exit criteria At least one owned experiment or explicit “no change” with rationale.
Anti-patterns Blame; infinite venting; items with no owner.
What must never happen Retro outcomes that never connect to Ember Log or directives when process change is claimed.

5.7 Ad hoc decision meeting

Field Detail
Name Ad hoc decision meeting
For what Resolve one contested decision under pressure (scope fork, architecture choice, incident response, policy exception).
Why Prevents gridlock without inventing a standing committee.
Primary decision question What do we decide now—with owners and log entry—and what is explicitly deferred?
Trigger / cadence Only when real decision pressure exists.
Suggested timebox 30–90 min (single topic).
Allowed delivery modes All three; Versona may branch-analysis; decision human.
Required participants Minimum: accountable hat owners + affected implementers; Challenge/Governance as needed.
Human owner Named domain owner (in invite).
Versona owner/support Scenario analysis; draft ADR text.
Required inputs Problem statement, options, constraints, blast radius; links to Ore/Ingot/Spark when applicable.
Recommended pre-work One-pager or Versona digest before the meeting.
Agenda / flow Frame → options → criteria → decision → Ember Log → actions.
Created or updated Ember Log; ADR; Spark updates.
Decisions expected One primary decision or explicit escalation path.
Exit criteria Logged outcome with owners; no vague “later” without trigger date.
Anti-patterns Standing weekly “decision meeting”; revisiting with no new information.
What must never happen Replacing Refinement, Planning, or Retro with perpetual ad hoc chaos.

6. Decisions kept (accountability)

Decision type Minimum human sign-off
Ore → Ingot promotion / Bank / reject Product + Engineering (Challenge engaged at risk)
Iteration scope & Spark commitment Product + Engineering; Governance when policy-bound
Banking / major priority shift Product (with Engineering per team policy)
Assay pass / fail / ship with conditions Governance + Product + Engineering
Risk acceptance / policy exception Governance + domain owner; Ember Log
Retro experiments Named human owners; directive changes per governance

Versona-only tracks may prepare and recommend; they do not own these accountabilities.

7. Drift signals

Drift Symptom
Planning eats Refinement Ingots vague; mid-week scope churn.
Daily sync becomes mini-planning 15m → 45m; Charge thrashes daily.
Review ≡ Assay Ship without evidence-package discipline.
Versona-only substitute for accountability Log shows “AI approved” with no human owner.
Charge as object model Meetings nested under Charge instead of confirming Charge at sync.

8. Inconsistencies to resolve locally

Topic Tension Resolution
Versona sessions vs meetings Forge & planning — naming reference: Versona sessions are not a substitute for team meetings Treat delivery mode as how the same meeting type is run; raw forge-logs/versona/ work without a meeting record is still not a team meeting.
Review vs Assay same day Convenience vs strict separation Same day allowed; distinct decision questions and minutes.
Daily sync + Versona-only Weak fit for team event Prefer hybrid; Versona supports, does not chair.

9. Open questions (tailor per team)

  1. Governance always in Planning/Assay vs on-call async sign-off?
  2. Combined Review + Assay time block: when is evidence maturity high enough?
  3. 2-week iterations: scale timeboxes linearly or keep caps?
  4. Minimum stakeholder attendance at Review for internal increments?
  5. Regulated contexts: does Assay require specific Versona §5 outputs in the evidence package every time, or only per work type?