Overview
How I approach this work
I have been shipping Claude and OpenAI into production systems since 2023. Content Mint runs a multi-brand content pipeline orchestrating Haiku, Sonnet, and Opus with per-token cost tracking. Brand KB does RAG over pgvector with an embeddable SDK. Harness Lab is a self-improving meta-harness that analyzes historical runs and proposes prompt optimizations. The common thread across all three: AI as a governed system with cost tracking, review gates, and explicit failure modes.
Most AI engagements I take on start with a workflow audit: what repeats, what depends on judgment, and where a language model can do useful work without generating liability. From there I design implementations that use the smallest viable model for each step — Haiku for classification and scoring, Sonnet for drafting, Opus for evaluation and judgment — with a per-token cost ledger and human review gates where the output matters. I have built this pattern across content pipelines, retrieval systems, prompt testing infrastructure, and customer-facing RAG chat. I stay away from speculative autonomy and focus on systems that can be shipped, measured, and improved.
Deliverables
What a typical engagement produces
Concrete artifacts from this kind of work.
- Implementation plan
- A written integration design for the specific workflow, including model selection per step, cost projections, fallback behavior, and human-in-the-loop checkpoints.
- Working pilot
- A functioning prototype running against the real workflow, built inside existing platforms where possible, with observability and cost tracking wired in from day one.
- Governance and review process
- Written guidelines for when the team trusts model output, when it requires review, and how to handle model drift and new releases without rewriting the integration.
- Production AI integrations
- 4+
- Models in production orchestration
- Haiku / Sonnet / Opus
- Related tooling
- pgvector, Pinecone, MCP
Related areas
Other parts of my practice that overlap with this one.
Performance Consulting
Infrastructure and web performance that shows up in the numbers.
Analytics Consulting
Measurement systems built to drive decisions, not reports.
Content Consulting
Editorial systems, automated production, and programmatic SEO at scale.