Field notes

Notes from the front line of applied AI.

What we learn deploying custom AI inside healthcare systems, law firms, and sports marketing teams.

Deployment — Aug 2026

What forward deployed engineering actually looks like

Embedding with a client team changes what you build. Here's the operating model we use to get AI into production in weeks, not quarters.

Dara Welk — 8 min read

Healthcare

PHI-safe RAG: an architecture that survives audit

Retrieval over clinical archives without leaking protected data across boundaries.

Priya Raman6 min read

Legal

Privilege-aware access control for legal AI

Matter scoping is a permissions problem before it is a model problem.

Marcus Feld5 min read

Evaluation

Evals your stakeholders will actually read

Turning model quality into a metric an operations lead can act on.

Priya Raman7 min read

Sports Marketing

Brand voice tuning for a 162-game season

Keeping generated creative on-brand when volume goes up ten times.

Jules Ortega4 min read

Automation

Human-in-the-loop is a design decision, not a fallback

Where to put the review step so the workflow still saves time.

Marcus Feld6 min read

Engineering

Shipping AI inside the client's perimeter

Private hosting patterns that keep security teams comfortable and velocity high.

Dara Welk9 min read

Want this applied to your business?

Our engineers embed with your team and ship working AI in weeks.

Start a project