From the State Government edition of September 27, 2026
Missouri AWS Champions profile quantifies DOLIR UI AI at 93% review-time cut and ~79,000 hours recaptured
Missouri Office of Administration — IT Services Division (OA-ITSD); Department of Labor and Industrial Relations; Department of Revenue · State shared services / unemployment insurance adjudication / revenue virtual assistant / developer productivity · Missouri, United States
- Publisher
- Government Technology — AWS Champions Program 2026: State of Missouri OA-ITSD
- Original publication
- 2026-09-04
- Source retrieved
- 2026-09-28
- Event date
- 2026-09-04
What happened
Why now: after the 2026-09-26 Digital States Missouri interview archived high-level UI/grant claims without hard baselines, this Sep 4 AWS Champions profile (still unarchived) supplies quantified operating metrics for the same enterprise AI stack. It states the DOLIR Adjudication Assistant (Amazon Bedrock/RAG) cut initial unemployment case review from about 30 minutes to two minutes (93%), recapturing about 79,000 staff hours annually; DORA (Revenue virtual assistant upgraded to GenAI) has >11 million inquiries / 2.6 million sessions with GenAI upgrade estimated to return 400–500 staff hours annually; Kiro expanding from 40 to 400 developers. Quotes Tim Marczewski on human-in-the-loop trustworthy AI.
Why it matters
Fills measurement gaps beside Missouri’s Digital States narrative—concrete UI adjudication cycle-time and hours-saved claims plus developer-assist scale for benefits and shared-services engagements.
Evidence and measured results
GovTech AWS Champions sponsored profile attributing metrics to Missouri OA-ITSD/DOLIR/DOR; Bedrock/RAG named for adjudication assistant; no independent audit or methodology appendix in opened page.
Limitations and uncertainty
Sponsored AWS Champions content; 79,000-hour and 93% figures are state-attributed without methodology, time window, or independent verification in the opened page. Do not treat as audited ROI.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-09-27; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Benefits and shared-services leaders need credible, human-supervised UI GenAI cycle-time and hours-saved evidence before scaling adjudication assistants.
- Stakeholders
- State CIO / OA-ITSD
- DOLIR adjudicators
- Department of Revenue (DORA)
- AI/Innovation director
- Privacy/security
- Discovery questions
- What case volume and workforce formula produced the 79,000-hour figure?
- Which adjudication steps remain human-mandatory?
- How were 30-minute and 2-minute baselines timed?
- What appeal/error rates changed after Bedrock assist?
- Value hypothesis
- A measurement engagement can convert Missouri’s Champions metrics into decision-grade baselines reusable across UI and revenue assistants.
- Potential engagement
- 6–8 week DOLIR/DORA instrumentation and Kiro expansion readiness pilot.
- Unsupported claims
- Audited statewide FTE reduction
- Guaranteed 93% cycle-time transferable to other states
- Unsupervised benefit auto-adjudication
Pre-sales engineering
Role takeaway
- Fit
- Strong fit for human-supervised UI decision-support and GenAI virtual-assistant upgrades with developer-assist scale.
- Architecture and integration
- Amazon Bedrock/RAG in DOLIR workflows; DORA GenAI upgrade; Kiro for AppDev; enterprise productivity/IDP layer.
- Prerequisites
- UI case taxonomy and appeal rules
- Adjudicator override UX
- DORA content/corpus governance
- Secure SDLC for Kiro
- Constraints
- Sponsored metrics only
- Missouri policy/EO context
- UI federal program rules
- Unpublished full control matrix
- Security
- UI/tax PII requires disposition logs, least privilege, and prohibition on unsupervised benefit decisions.
- Proof of value
- Replicate 30→2 minute timing on sample
- Appeal/error monitoring
- DORA maintenance-hour ledger
- Kiro PR quality/security gates
Delivery
Role takeaway
- Work
- Measurement dashboards, override UX, DORA content ops, and Kiro enablement with agency owners.
- Dependencies
- OA-ITSD platform
- DOLIR adjudicators
- Revenue DORA ops
- Security review
- Ownership
- OA-ITSD enterprise AI with DOLIR/Revenue operational ownership of outcomes.
- Skills and adoption
- Adjudicators need AI-assist QA literacy; developers need secure-codegen review skills.
- Governance checkpoints
- Pre-scale UI accuracy gate
- Monthly appeal review
- DORA hallucination sampling
- Kiro access/class reviews
- Acceptance criteria
- Published 79k methodology
- Documented human-mandatory steps
- No unsupervised benefit auto-decision
- Kiro expansion security sign-off
- Risks
- Silent adjudication errors
- Inflated sponsored ROI marketing
- Shadow Kiro use on restricted repos
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Bedrock + RAG adjudication assist; DORA rules→GenAI virtual assistant; IDP and productivity tools; Kiro developer assistant seat expansion.
Governance
Who approves, reviews and stays accountable for outcomes?
Human-in-the-loop and trustworthy-AI framing from AI/Innovation director; no published EO checkpoint detail in this profile (EO 26-02 covered separately in archive).
Security and privacy
What data, permissions and controls need testing?
UI and tax/revenue workflows imply sensitive PII; profile asserts secure, governed solutions but does not publish DLP, retention, or model-logging controls.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Hours-recaptured claims imply workforce reallocation capacity; accessibility impacts for claimants using DORA not quantified here.
Procurement
What should contracts, pricing and exit terms secure?
AWS Bedrock/RAG and Kiro named; confirm contract vehicles, data-processing terms, and Champions marketing vs contractual SLAs before scale.
Operating model
Which teams own the service once it runs?
OA-ITSD-centered enterprise AI with DOLIR and Revenue as operational owners; developer-assist as parallel AppDev track.
Publication history
- 2026-09-27State Government · Issue 225 resources
Stable resource ID: missouri-aws-champions-dolir-dora-kiro-metrics-20260904