Skip to content
← AI-native training by LockedIn Labs

PRACTITIONER INSTRUCTORS / OBSERVATION PRACTICE

Observe the teaching.
Keep the evidence.

A demonstration becomes useful teaching when another person can attempt the task, explain the result and act on a correction. Keep those observations separate from a polished presentation.

Start an observation ↓
  1. 01 / CANDIDATEPredict and demonstrateExact version, command and observed result.
  2. 02 / LEARNERAttempt the receiving taskOwn action, located evidence and assistance.
  3. 03 / OBSERVERChallenge the explanationChanged condition, reasoning and unresolved gap.
  4. 04 / CORRECTIONObserve the next attemptRevised explanation and bounded next role.

An observation record,
with the missing parts visible.

Use three people where possible: candidate instructor, learner and observer. Record combined roles and solo self-review. Leave findings unobserved until you have evidence; a selected label creates no verified result.

Use this alongside the sixty-minute teach-back agenda. Prepare the lab and reading beforehand, then retain each person’s own practice record and the observer’s account separately. No account is needed and nothing is submitted.

Review the prepared teach-back agenda →Download the instructor practice kit ↓

LOCAL OBSERVATION / HUMAN NOTES

Keep the teaching evidence.

Use a separate record for each observed attempt. Entries stay in this tab; save an editable draft before leaving. Use fictional aliases and permitted synthetic artifacts.

01Set the observation context

Name one bounded topic the candidate is teaching.

What should a learner already know or have prepared?

Exact file/version; date and time zone only if actually observed.

Use a fictional alias; note combined roles.

Who actually attempts the receiving task?

Name combined roles or solo self-review explicitly.

What may be shown, and what did the candidate actually contribute?

What action should the learner attempt, and what would you inspect?

02Record what actually happened

Runtime/version; actual execution, reading or fallback. Preserve any setup gap.

Keep the prediction, exact command, actual output and evidence location distinct.

Which input changed? Record the prediction before the new result.

What did the learner do themselves? Record hints, candidate takeover and remaining confusion.

Record the missing source, conflicting result or setup fault and the candidate's response.

03Attach evidence to each finding

Choose a label only for what you observed. These are your notes, not independently verified findings or a score. Leave missing observations visible.

Technical accuracy

Locate the exact code, contract or output supporting the explanation.

Reproducibility

Could another person prepare, run and locate the stated result?

Failure reasoning

Compare the prediction with actual refusal/recovery and preserve uncertainty.

Learner coaching

Name the learner's own action, assistance and any candidate takeover.

Safe practice

Record permitted data, access boundaries and the fallback actually used.

Receiving transfer

What did the receiver attempt, what evidence did they find and what remains untested?

04Leave a useful receiving action

Quote the correction and the learner's next attempted action; do not infer transfer.

Evidence still needed, responsible role and next observed attempt or proposed co-teaching scope.

Keep or resume this record

The Markdown record is for reading. JSON preserves editable notes and labels. Opening a file previews it; replacing your current work requires a separate action.

This record creates no instructor approval, faculty appointment or delivery authority. The accountable teaching owner reviews actual evidence and agrees any repair or co-teaching scope separately.