Whitepapers
Practitioner-grade reference documents for enterprise GenAI programs — architecture, governance, cost, data, and vendor evaluation. All published under CC BY 4.0 with Markdown downloads for offline use.
- Version 1.1 · 2026-08-07 · ~5400 words
Agentic AI Reference Architecture 2026
The layered architecture, deployment patterns, enterprise integration surfaces, and evaluation discipline required to run autonomous agents in enterprise production
A reference architecture for running agentic AI systems in enterprise production. Covers the six layers (model, orchestration, memory, tools, guardrails, observability) with an inline stack diagram; three canonical deployment patterns; the enterprise integration surfaces where the stack plugs into existing IdP, data platform, API gateway, SIEM, MLOps, GRC, and ITSM systems; an evolution path from pilot to autonomous; an evaluation harness distinct from LLM evals; and the failure modes that separate demo from production. This is a v1.1 practitioner reference — not vendor marketing.
- Version 1.0 · 2026-08-07 · ~3600 words
GenAI Governance Playbook
A practitioner playbook for governing enterprise GenAI: operating model, system inventory, risk-tier classification, model cards, runtime guardrails, incident response, and regulatory alignment
GenAI governance is where enterprises stall between maturity levels 2 and 3, and where they take the greatest reputational and regulatory risk when they fail. This playbook covers the operating model, inventory + classification workflow, model and data card templates, runtime guardrail deployment, incident response, and cross-mapping to the EU AI Act, NIST AI RMF, ISO 42001, and SOC 2 — organized as sequenced practitioner actions, not compliance theater.
- Version 1.0 · 2026-08-07 · ~3500 words
LLM Cost Optimization Patterns
The seven patterns that separate enterprise GenAI programs paying rack rate from those running at 5-20x lower unit cost
Enterprise LLM cost is not a model-choice problem. It is a workload-engineering problem. This whitepaper covers the seven patterns — prompt caching, model routing, prompt compression, speculative decoding, batch API, distillation, and cost-per-outcome budgeting — that together drop production unit cost by 5-20x on comparable workloads. Concrete implementation guidance; no vendor marketing.
- Version 1.0 · 2026-08-07 · ~3400 words
Data Readiness for GenAI
The five sub-dimensions of the data-platform work that unblocks — or silently blocks — enterprise GenAI at scale
The most common blocker to scaling enterprise GenAI beyond pilots is not model quality; it is data-platform readiness. This whitepaper covers the five sub-dimensions — catalog & lineage, quality, access & permissions, retrieval & vector, compliance & governance — with the specific investments that separate teams that ship production RAG smoothly from teams that stall in pilot for a year.
- Version 1.0 · 2026-08-07 · ~3300 words
GenAI Vendor Evaluation Framework
A defensible framework for evaluating and selecting enterprise GenAI vendors across model providers, application vendors, and infrastructure — 40 criteria, 8 categories, weighted decision framework
Enterprise GenAI vendor selection routinely turns into consultant-driven RFP theater that produces glossy scorecards and disappointing outcomes. This whitepaper provides a defensible framework: 40 criteria across 8 weighted categories, an RFP question bank, reference-check discipline, and contract negotiation points that protect the enterprise position over the vendor lifecycle. Companion to the free /tools/vendor-scorecard interactive tool.
Each whitepaper is available as an interactive page and a downloadable Markdown file (append /download.md to the URL). CC BY 4.0 — free to share, adapt, and cite with attribution.