Cossette Consulting LLC

Capability Statement · Public-Sector AI Modernization

Loren Cossette

Principal · AI Program Architect

lorencossette.com

Firm Profile

Cossette Consulting LLC is a Texas-domiciled professional services firm specializing in embedded AI program leadership for state, county, and city government agencies. The firm's principal, Loren Cossette, currently serves as the embedded AI lead for Multnomah County, Oregon, running five concurrent production AI initiatives across accessibility, site intelligence, legacy modernization, and QA automation. Unlike traditional consulting engagements that staff multiple vendors against a single program, Cossette Consulting delivers full-lifecycle AI work — strategy, architecture, build, knowledge capture, stakeholder enablement, and documented handoff — under one accountable principal.

Core Competencies

  • Embedded AI program leadership (fractional CAIO / AI lead)
  • Multi-agent AI system architecture & implementation
  • WCAG / Section 508 / ADA Title II accessibility automation
  • Drupal-aware CMS auditing & site intelligence platforms
  • Legacy-system modernization with behavioral test capture
  • Knowledge corpus & SME-confirmation workflows
  • RAG over agency data (pgvector, embeddings, retrieval)
  • Multi-tenant architecture (RLS, IAP, Workload Identity Fed.)
  • AI governance frameworks & change management
  • Executive AI training & staff enablement programs
  • Independent technical review & validation
  • AI ROI modeling & cost-benefit analysis

Differentiators

  1. 1 · Currently embedded as an AI lead in a real public agency. Not a slide-deck consultant. Five production AI initiatives running simultaneously inside Multnomah County, with a dated public delivery record (multco-presentations.web.app) any procurement officer can verify.
  2. 2 · Independent third-party validation. File Intelligence Platform validated against Xingwu PHP audit at 92.3% agreement on 33,738 UUIDs. A11yReady validated by axe-core at 95%+ first-pass WCAG 2.1 AA compliance. No vendor-supplied testimonials; verifiable metrics.
  3. 3 · Documented handoff is the engagement outcome. The objective is to leave the agency stronger, not dependent. Every engagement produces leadership explainers, operations security handbooks, reviewer onboarding guides, validation pipeline infographics, and technical turnover binders.
  4. 4 · Agency-controlled cloud, zero data exfiltration. All AI processing runs inside the agency's own cloud tenant. Model calls flow through enterprise APIs (Vertex AI) that contractually exclude customer data from foundation-model training.
  5. 5 · 20-year military leadership background. U.S. Air Force E-9 (top 1% of enlisted leadership), Distinguished Graduate of the Senior NCO Academy, multiple NCO-of-the-year awards. Disciplined operations, documentation, and accountability are habitual, not aspirational.

Past Performance · Multnomah County AI Program

A11yReady · Production AI document accessibility platform. 30-second PDF→HTML conversion (vs. 1–4 hours manual). 95%+ first-pass WCAG 2.1 AA. 34 document clusters. 734 automated tests. Built to clear the County's public-PDF backlog before the ADA Title II April 2027 deadline. Live at a11y-ready.multco.us.

File Intelligence Platform · Nightly Drupal-aware audit of 56,500+ files and 9,900+ pages across 30 departments. Six-bucket deterministic file classification, 10-module audit pipeline. Validated against independent Xingwu PHP audit at 92.3% agreement on 33,738 UUIDs.

AI-Powered Testing Platform · Six-agent system for legacy-system modernization. 9,420 indexed code symbols. Multi-tenant from day one (Row-Level Security, per-agent service accounts, Identity-Aware Proxy, Workload Identity Federation). Designed to onboard additional agencies after Multnomah County's UCR rebuild.

UCR Modernization · Behavioral test capture from retiring Subject Matter Experts. 704 knowledge sessions ingested with full provenance. 208 behavioral Playwright tests at 98.1% compile-clean. Behavioral acceptance contract: the modernized Universal Client Registry must pass 208/208 before it ships.

Content Automation (Linkit) · Earlier engagement. Four-agent PHP pipeline translation-ready 20,000+ government pages. 633 hours manual effort reduced to 24 hours deployed (26× faster). $34,946 saved (89% cost reduction). Protected 32K internal links from translation corruption.

Project descriptions reflect Loren Cossette's professional work and do not imply endorsement by Multnomah County or any other public agency.

NAICS Codes

541511Custom Computer Programming Services
541512Computer Systems Design Services
541618Other Management Consulting Services (incl. AI Strategy)
541690Other Scientific and Technical Consulting Services
611430Professional & Management Development Training

Engagement Types

  • Discovery Engagement — 2–3 weeks, $8K–$15K fixed-price. AI program assessment, opportunity mapping, recommended roadmap.
  • Embedded AI Program Lead (Flagship) — 12–24 weeks minimum, $25K–$40K/month retainer. End-to-end program leadership: architecture, build, knowledge capture, stakeholder enablement, team handoff.
  • Architecture & Build Sprint — 4–10 weeks, $30K–$80K fixed-price. Specific system architected, built, deployed, and documented.
  • Executive Workshop & Training — 1 day to multi-cohort curricula, $5K–$40K depending on scope.
  • Technical Review / Validation — Independent expert review of in-progress or completed AI systems. Fixed scope.

Contracting & Procurement

Legal EntityCossette Consulting LLC
Entity TypeTexas Domestic Limited Liability Company
Date FormedDecember 29, 2022
TX Secretary of State File #0804858459
TX Taxpayer ID32087716570
Federal EIN92-1502035
Registered AgentRegistered Agents Inc., 5900 Balcones Drive STE 100, Austin, TX 78731
Principal OfficeSan Antonio, TX 78261
InsuranceGeneral Liability · Professional Liability (E&O) · Cyber Liability — Certificates of Insurance available upon request
BankingEstablished U.S. business banking relationship
Engagement OptionsDirect PO · Fixed-price · Monthly retainer · Sole-source justification · Cooperative purchasing subcontract
Cooperative PurchasingAvailable via Sourcewell, OMNIA Partners, TIPS, and NASPO ValuePoint primes — vehicle identification during Discovery
Geographic Service AreaUnited States · Remote primary · On-site as needed
Conflict-of-InterestReviewed during Discovery prior to any binding scope

Principal · Loren Cossette

Currently the embedded AI lead for Multnomah County, Oregon. PhD candidate in Leadership & Program Evaluation. M.S. Organizational Development. M.A. Psychology. Prosci-certified Change Practitioner. SHRM-SCP. SAFe Agilist. Twenty-year U.S. Air Force career culminating at E-9 (top 1% of enlisted leadership), including Distinguished Graduate of the Senior Noncommissioned Officer Academy, 19th Air Force First Sergeant of the Year (2016), and Air Force Materiel Command Noncommissioned Officer of the Year (2013). Five peer-reviewed AI publications. Executive AI training engagements at UT Austin and Johns Hopkins.

Contact

loren.cossette@evolviqtx.com
210.836.4789
lorencossette.com
linkedin.com/in/loren-cossette
San Antonio, Texas, United States
Book a working session via the website
© 2026 Cossette Consulting LLC · Texas LLCCapability Statement · Revised June 2026

Save as PDF using your browser's print dialog → Destination: Save as PDF