Epistema Explanatory Power

Governance Console

Workflow readiness infrastructure for AI-assisted work.

4 workflows 2 teams fixture-backed

Readiness

Review Required

Review gates and drift before commitment

Workflows: 4
Gates: 1
Drift Watch: 1
Calibration: 2

Actions

What This Means

Glossary-backed micro-explanations for the console terms.

Explanatory Power Core

Ability of a reasoning structure to explain more, become harder to vary, reduce confusion, preserve useful distinctions, and improve future work.

It is the product promise behind Epistema and a key criterion for QTF explanatory_power_gain.

Explanatory Power Gain Experimental

QTF component that scores whether S1 explains more or better than S0.

It turns the broad Explanatory Power promise into an inspectable component score with rationale.

Context Derived

Practical workflow brittleness pressure, not exact context-window accounting.

High context can mean the next step should be compressed, narrowed, or re-scoped.

Budget Burn Derived

Operating budget pressure across tokens, dollars, context limits, or configured workflow budget.

Burn is reviewed against outcome, PEI, QTF, drift, and evidence rather than judged alone.

Drift Derived

Risk that a workflow is moving away from its stated objective, constraints, or governed path.

Drift is a governance signal, not a model-quality score.

Gated Core

Recognized and recorded, but held before active lifecycle release pending authorization.

Gate is not reject; the record remains governable.

Constrained Core

Admitted with explicit limitations that remain visible before downstream commitment.

A constrained object should not produce an unconstrained commitment.

QTF Experimental

Framework for evaluating the quality of movement from one epistemic state to another.

QTF evaluates quality of transformation; PEI evaluates productive expansion.

Pre-flight

Can this AI-assisted workflow proceed, commit, and later calibrate?

Scope 4 workflows loaded ready
Gates 1 gated workflow review
Drift 1 workflow above watch threshold review
Commitment 1 blocked without authorization review
Calibration 2 due or watch item review

Instrument Panel

Gauge = magnitude. Lamp = governance state. Text = next interpretation.

fuel gauge

Budget Fuel 78% left

Token and spend pressure stay parallel to value, not a value claim by themselves.

Source: token budget, provider spend, or admin budget.

context gauge

Context Reserve 74% left

Configured estimate of remaining context before compression or brittle continuation.

Scale: 0-100% configured reserve.

productive expansion

PEI 1.15

BLUE means a useful new path opened; PEI remains an evaluation signal, not ontology.

Scale: 0.0-1.6 display cap; 1.25+ strong.

transformation quality

QTF 0.82

Quality asks whether the reasoning structure improved, not whether it sounded fluent.

Scale: 0.0-1.0; 0.85+ strong.

yield readiness

EY directional

Epistemic Yield is shown as readiness only until understanding-gain requirements are calibrated.

Not scored yet; no formula implied.

warning light

Drift 0.21

Drift is a governance warning signal: when it rises, pause before commitment.

Risk scale: 0.0-0.5 display cap; 0.35+ high.

constructive expansion
ready / justified
watch / incomplete
blocked / escalation

Status

Remaining capacity, context pressure, budget burn, expansion, quality, and drift.

org_demo

organization / standard

Remaining 74% healthy
Context 26% low
Budget Burn 28% watch
Expansion 1.15 positive
Quality 0.82 acceptable
Drift 0.21 watch

review gate before release

Strategy AI Workflows

team / advanced

Budget 22% used low
Context 21% low
Burn 22% low
PEI 1.27 strong
QTF 0.78 acceptable
Drift 0.24 watch

continue with monitoring

Risk Review

team / standard

Remaining 93% healthy
Context 7% low
Budget Burn 1% low
Expansion 0.80 limited
Quality 0.91 strong
Drift 0.12 low

review gate before release

Selected Workflow

Open a workflow from the table to update this inspector.

wf_gate_001

risk-gated-review

Owner

Risk Lead

team_beta / dept_risk

Participants

Risk Lead, Policy Gate

No delegated agents

Next Step

Hold for authorization before commitment

risk_gate_001, approval_pending_001

Signal drift=0.12 / qtf=0.91
Evidence 2 records / manual
Commitment blocked_without_authorization
Outcome escalated
Calibration requires_review

Manager Rollup

Team-level readiness without hiding workflow drillback.

Risk Review

team_beta

Workflows1
Gates1
Budget Burn Source0.01
Avg Drift0.12
Avg PEI0.80
Avg QTF0.91

Review gated workflow before release.

Drillback: wf_gate_001

Strategy AI Workflows

team_alpha

Workflows3
Gates0
Budget Burn Source0.2703
Avg Drift0.24
Avg PEI1.27
Avg QTF0.78

Review quality before commitment.

Drillback: wf_braintrust_001, wf_opik_001, wf_braintrust_002

Operator Checks

curated

Gated or constrained workflows

wf_braintrust_001provider=braintrust / gated=0 / constrained=2
wf_opik_001provider=opik / gated=0 / constrained=2
wf_braintrust_002provider=braintrust / gated=0 / constrained=2
wf_gate_001provider=manual / gated=1 / constrained=1

High drift workflows

wf_opik_001provider=opik / drift_risk=0.32

High budget burn / weak outcome workflows

wf_opik_001provider=opik / budget_burn_source=0.2003 / qtf=0.74

Team rollups needing review

team_alphagated=0 / avg_qtf=0.78 / avg_drift=0.24
team_betagated=1 / avg_qtf=0.91 / avg_drift=0.12

Evidence and Commitment

Signal to evidence to commitment to outcome to calibration.

wf_braintrust_001

braintrust

Signal: drift 0.18 / QTF 0.82 -> Evidence: 3 records -> Commitment: reviewable -> Outcome: pending -> Calibration: not_due

Evidence3
Commitmentreviewable
Outcomepending
Calibrationnot_due

wf_opik_001

opik

Signal: drift 0.32 / QTF 0.74 -> Evidence: 6 records -> Commitment: needs_review -> Outcome: pending -> Calibration: watch

Evidence6
Commitmentneeds_review
Outcomepending
Calibrationwatch

wf_braintrust_002

braintrust

Signal: drift 0.22 / QTF 0.79 -> Evidence: 2 records -> Commitment: reviewable -> Outcome: pending -> Calibration: not_due

Evidence2
Commitmentreviewable
Outcomepending
Calibrationnot_due

wf_gate_001

manual

Signal: drift 0.12 / QTF 0.91 -> Evidence: 2 records -> Commitment: blocked_without_authorization -> Outcome: escalated -> Calibration: requires_review

Evidence2
Commitmentblocked_without_authorization
Outcomeescalated
Calibrationrequires_review

All workflows shown. Select a team rollup to highlight its drillback rows.

Provider	Type	Gated	Constrained	Evidence	Commitment	Outcome	Calibration	Budget Burn Source	PEI	QTF	Provenance
braintrust	eval-rich-summary	0	2	3	reviewable	pending	not_due	0.04	1.2	0.82	root_01, trace_eval_01
opik	governed-research	0	2	6	needs_review	pending	watch	0.2003	1.5	0.74	opik_trace_dox_001
braintrust	eval-review	0	2	2	reviewable	pending	not_due	0.03	1.1	0.79	bt_review_002
manual	risk-gated-review	1	1	2	blocked_without_authorization	escalated	requires_review	0.01	0.8	0.91	risk_gate_001, approval_pending_001

Admin Settings

CSV + defaults

Setting	Value	Source	Scope
status_strip.budget_usd	1.00	default	global:*
status_strip.context_limit	100	default	global:*
context.turn_watch	16	default	global:*
context.turn_high	36	default	global:*
context.turn_critical	51	default	global:*
identity.default_roles	admin;reviewer;operator;viewer	default	global:*
operator_query.drift_high	0.30	default	global:*
operator_query.min_pei	1.00	default	global:*
operator_query.min_qtf	0.80	default	global:*
operator_query.cost_watch_usd	0.10	default	global:*
status_strip.budget_usd	0.75	admin_configured	team:team_beta
status_strip.context_limit	75	admin_configured	team:team_beta
status_strip.budget_usd	1.25	imported	team:team_alpha
context.turn_high	30	admin_configured	workflow_type:debug_loop
operator_query.drift_high	0.30	admin_configured	team:team_alpha
operator_query.cost_watch_usd	0.15	admin_configured	team:team_alpha

Demos

Representative readouts wired from the logical layer.

Logical status strip

status_strip

AIWI Logical UI Status Strip Demo

=================================

Standard mode is business-readable. Advanced mode exposes AIWI-native terms.

AIWI Logical Admin Configuration
================================
Admin config is a lightweight governance surface, not enterprise IAM.

Aggregated governance readout

aggregation

AIWI Aggregated Governance Readout
==================================

Purpose:
  Roll up workflow value ledger signals by principal, actor, agent, team, department, and organization.
  Preserve drillback to workflow and provenance identifiers.

Identity Model:
  principal: accountable or beneficiary identity

Braintrust workflow value ledger

braintrust_ledger

AIWI Braintrust Workflow Value Ledger
=====================================

Executive Readout:
  workflow_summary: Braintrust eval-rich fixture set
  value_signal: Braintrust scores, feedback, datasets, and lineage preserved without redefining AIWI semantics
  governance_signal: adapter disposition is translated into provisional Admission outcomes
  integration_status: bounded adapter-pattern proof, not production Braintrust ingestion