Lighthouse AdvisorySLED AI Adoption Intelligence

Public Sector & Government · Issue 01 ·

State Government

Three unarchived sources connect state AI purchasing, a bounded benefits workflow and independent scrutiny of outcome measurement. California reports easier Claude access without quantified service gains; Pennsylvania reports document-quality improvements with human eligibility authority; Code for America finds impact reporting limited. Historical evidence is explicitly dated. Two cross-source interpretations guide local validation, not guaranteed outcomes. International audit access failed and no fresh September effectiveness study was established.

Evidence records
3
Cross-source patterns
2
Evidence classes
1 standards or public-body guidance1 independent research1 government evaluation
Outcomes
1 emerging1 mixed1 effective
Source freshness
1 recent1 undated1 older, newly relevant
Research completed
2026-09-07

Choose a role to see its takeaway beside every record in the ledger.

Synthesis · Lighthouse Advisory interpretation

Patterns across the evidence

2 patterns, each supported by at least two sources

  1. Purchasing access needs a separate outcome test

    California's access announcement and Code for America's measurement gap support separating procurement milestones from demonstrated service value. Neither establishes a causal return on adoption.

    Operating questionWhich baseline, cost and quality measures must improve before the agency expands licenses?

    Supporting evidenceCalifornia Department of TechnologyCode for America

  2. Bounded workflows make operational feedback concrete

    Pennsylvania's upload-quality tool illustrates a specific assistance boundary; the national assessment calls attention to recurring evaluation. Combine narrow scope with ongoing error and access monitoring, without generalizing the pilot's aggregate results.

    Operating questionCan the service owner detect harmful flags and recover an applicant's workflow without delegating eligibility to the model?

    Supporting evidencePennsylvania Office of Administration and Department of Human ServicesCode for America

Full record · every source keeps its link and limitations

Evidence ledger

3 records
  1. Standards or public-body guidanceEmergingRecent

    California centralizes Claude purchasing; service benefits remain unquantified

    CDT announces centralized Claude purchasing through SITeS, a 50% discount, and training/support. It describes existing agency uses but supplies no measured service improvement.

    California Department of TechnologyCalifornia, United StatesJuly 8, 2026

    Why it matters, evidence and limitations
    Why it matters
    Direct state shared-service evidence. Classified as guidance because the permitted taxonomy has no government-announcement category; this is not an evaluation.
    Evidence and measured results
    The announcement names software security, DMV customer service and Medicaid administration uses. Discount comparator, realized total costs, evaluation baseline, sample and outcome methods are absent.
    Limitations and uncertainty
    Operator announcement with vendor quotations, not independent effectiveness evidence. No precise agreement execution date established. No claim that a discount equals realized savings.
    Government procurement and deployment announcement Role takeaways & implementation
  2. Independent researchMixedUndated source

    State AI assessment finds impact reporting trails experimentation

    The assessment finds widespread experimentation but limited impact reporting and continuous learning. It distinguishes readiness, piloting, implementation and impact.

    Code for AmericaUnited StatesMay 2026; exact day not established on inspected assessment

    Why it matters, evidence and limitations
    Why it matters
    Direct state-government portfolio evidence with a benefits-access lens. Use it to frame evaluation questions, not to infer any state's present performance.
    Evidence and measured results
    Research ended in March 2026. Methods combine public-document desk research, state feedback opportunities and advisory rubric review. This is a maturity assessment, not a controlled test of AI effectiveness.
    Limitations and uncertainty
    Public reporting can miss internal work. Rendered state totals were unavailable; no counts or rankings are asserted. Full PDF was email-gated and not accessed. Findings are not a September status census.
    Independent nonprofit desk-research assessment Role takeaways & implementation
  3. Government evaluationEffectiveNewly relevant · Dec 2025

    Pennsylvania reports document-quality gains while keeping eligibility decisions with staff

    Pennsylvania reports an 80% reduction in illegible or incorrect documents and over 700 staff hours saved in its COMPASS document-processing pilot.

    Pennsylvania Office of Administration and Department of Human ServicesPennsylvania, United StatesDecember 15, 2025

    Why it matters, evidence and limitations
    Why it matters
    A state benefits workflow with a bounded assistive role. Effective refers only to operator-reported document-quality results, not verified eligibility outcomes.
    Evidence and measured results
    The release describes approximately 12,000 documents screened and concerns identified in 25% of cases. Piloting began in October 2025; December 15 is the announced launch. The tool screens image quality and relevance, not eligibility.
    Limitations and uncertainty
    Operator-reported evaluation in a launch release. No control group, baseline counts, error intervals, time-accounting method or subgroup analysis. Document totals are not participant counts. No independent replication.
    Government pilot results and launch announcement Role takeaways & implementation

How to read this edition

Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.

Standards or public-body guidance
Normative or advisory guidance from a standards body or public institution.
Independent research
Research conducted outside the implementing organization.
Government evaluation
A public body’s measured evaluation or documented pilot.