PISMOPISMO
IK

PISMO Platform · Internal

PISMO Output QA Agent

Runs rubric-based scoring, factuality checks and policy screens across all production PISMO surfaces, routing borderline outputs into the human review queue.

AdoptedHealthy
GSJ
Back to portfolio

Business owner

Daniel Whitfield

Executive sponsor

Renee Alvarado, VP Quality

Start date

Apr 22, 2025

Target rollout

Feb 2, 2026

Completion

100%

ROI

Finance-reviewed, annualized where noted

Hours saved / mo

380

Hours saved / yr

4,560

Cost avoided

$54K

Margin impact

$196K

Confidence level: High — based on baseline cycle-time study and realized utilization to date.

Health & Risk

Steady state. Operating as the org-wide quality gate.

Blockers

No open blockers.

Risk flags

None raised this cycle.

Mitigation notes

Quarterly rubric recalibration with practice leads.

Activity feed

Synced from connected delivery systems

  1. G

    PR merged: add factuality probe for tabular outputs

    3h ago

  2. S

    #qa-alerts: nightly run 98.6% pass (1,450 samples)

    9h ago

  3. J

    QA-812 closed — rubric v4 rollout

    2 days ago

PISMO Output QA

Last validation run Aug 19, 2026 · 09:05 UTC · 1450 samples

98.6%

Reviewer notes · Daniel Whitfield

Golden set expanded to 1,450 cases. Regression suite green for 22 consecutive runs.

Governance posture

NIST AI RMF-aligned control checks

4.9 / 5
  • Risk AssessmentPassed

    Platform-tier assessment current.

  • Bias MitigationPassed

    Rubric includes demographic parity probes.

  • Data PrivacyPassed

    Evaluation corpus fully de-identified.

  • TransparencyPassed

    Scores published to initiative owners.

  • SecurityPassed

    Read-only scoped service accounts.

  • Human OversightPassed

    Escalation path staffed 5x8.

Next initiative: Agentic Portfolio Dashboard