ck@46: ~ — tmux: consult

ck@46:~$ cat /consult/index.md

/consult

Focused advisory and implementation support for teams building, evaluating, or hardening AI systems.

ls services/ · 3 ways to work

▸LLM Safety & Reliability Sprint

2–3 weeks

Evaluate an agentic system, build a focused test harness, and define quality gates that catch regressions before release.

  • [OK]Custom eval dataset & metric definition
  • [OK]Automated regression suite
  • [OK]Risk report & mitigation plan

▸Agentic Risk Review

1 week

A structured review of an agent's architecture, tool use, and failure modes.

  • [OK]Threat model document
  • [OK]Failure mode taxonomy
  • [OK]Immediate fix recommendations

▸Build-with-you Advisory

Monthly

Ongoing architecture review, evaluation design, and implementation guidance for teams moving quickly.

  • [OK]Weekly architecture reviews
  • [OK]Code-level guidance
  • [OK]Hiring support for safety roles
./book_discovery_call.sh(15–20 min)

typical deliverables

Eval harness + regression suiteRisk model + failure taxonomyQuality gates + guardrail recommendations

ls case_snapshots/ · 3 engagements

▸ case_1.log

Context

Financial analysis agent

Work

Evaluation harness and guardrail review

Result

Details available on request

▸ case_2.log

Context

Clinical guideline assistant

Work

Safety gate design and multi-turn evaluation

Result

Details available on request

▸ case_3.log

Context

Code-generation workflow

Work

Regression testing and failure taxonomy

Result

Details available on request

[ck46]0:home1:research2:projects3:consult4:contact