Hunter Brennick AI Systems & Advisory ↗
AI Systems Orchestration
Part II · Patterns And Practice/Chapter 07
07

Evals, Guardrails, And Safety

Trust: evals, traces, guardrails, autonomy levels, approval gates, sandboxing, and knowing when not to automate.

6 min read2 figures
The gist
  • If you cannot evaluate the system, you cannot improve it on purpose.
  • Traces are not logs for after the incident. They are training data for the next version.
  • Autonomy is a privilege the system earns, level by level, with written proof.
THE TRUST STACK input guardrail tool boundary output guardrail Requesttask · context Modelplans · reasons Toolsvalidated calls Resultdrafts · actions review Shippedwith evidence Traces what it knew · what it assumed · which tools it called · where it wasted tokens · where it drifted Evals capability and regression checks · code graders · rubric judges · calibrated human spot checks if you cannot evaluate the system you cannot improve it on purpose
FIG 07.1The trust stack: guardrails at every boundary with evals and traces underneath.

Module 8: Evals, Guardrails, And Observability

This is the difference between demo AI and production AI.

Learn the pieces that separate demo AI from production AI:

  • capability evals versus regression evals
  • code graders
  • model graders
  • human calibration
  • transcript review
  • trace analysis
  • tool call validation
  • cost and latency tracking
  • guardrails at the input boundary
  • guardrails at the output boundary
  • guardrails at the tool boundary

Build a 20 task eval set for one agent workflow. Pair it with three graders:

  • a deterministic check
  • a rubric based LLM judge
  • a human spot check

Then build a trace review template that asks:

  • what the system knew
  • what it assumed
  • which tools it called
  • where it wasted tokens
  • where it drifted

If you cannot evaluate the system you cannot improve it on purpose. Traces are not logs for after the incident. They are training data for the next version of the system.

Module 9: Autonomy And Safety Boundaries

Autonomy is a privilege the system earns.

Learn the vocabulary of autonomy and safety:

  • permission tiers
  • read-only versus write-capable runs
  • approval gates
  • reversible actions
  • sandboxed execution
  • secrets isolation
  • external agent trust boundaries
  • prompt injection risks
  • data retention policy
  • audit trails

Build an autonomy ladder for your own workflows:

L0: answer only
L1: read tools only
L2: draft artifacts
L3: propose changes
L4: edit local files
L5: run tests and validation
L6: open PR / create ticket after approval
L7: autonomous scheduled execution with monitoring

Every level needs to write the required proof before promotion.

THE AUTONOMY LADDER L0 · answer only L1 · read tools only L2 · draft artifacts L3 · propose changes L4 · edit local files L5 · run tests and validation L6 · open PR or ticket after approval L7 · autonomous scheduled execution with monitoring required proof before each promotion the goal is not a fearless system · the goal is a system trusted at higher levels of consequence
FIG 07.2The autonomy ladder: each level is a privilege the system earns with proof.

The goal is not to make the system fearless. The goal is to make it trusted at higher levels of consequence.

Next chapterChapter 08 · Frontier Leverage Patterns The fun edge: multi sampling, debate, critique, generator verifier loops, search, synthetic data, and spending reasoning where it matters.