Skip to main content

AI & Workflow Automation

Production AI integration with observability and governance.

Step 4

How it works

I have been shipping Claude and OpenAI into production systems since 2023. This phase is about wiring AI into specific workflows that benefit from it, with cost tracking and human review where the output matters.

Most AI work I take on starts 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, and prompt testing infrastructure. I stay away from speculative autonomy and focus on systems that can be shipped, measured, and improved.

What this covers

Key activities

The specific areas I focus on during this phase.

Workflow audit
A structured audit of where language-model work will be useful and where it will not, with specific integration points identified.
Model selection and orchestration
Per-step model selection with cost projections and fallback behavior — not a single model for every task.
Cost and quality tracking
Per-token cost ledger and output quality metrics wired into the integration from day one.

Outcomes

What gets produced

Concrete artifacts from this phase.

Working AI integration
A functioning integration running against the real workflow, built inside existing platforms where possible.
Cost ledger
Per-token cost tracking and monthly reporting so the AI spend is measurable and controllable.
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.
Extensibility notes
Documentation of how the integration can be extended to adjacent workflows without rewriting the core.

See how this fits together

The process steps sit inside the larger shape of a consulting engagement. Read about the areas I take on outside work, or browse the portfolio.