Skip to content
← AI-native training by LockedIn Labs

Pilot preparation / Two audiences

Find out what
people can use.

Prepare a small learning pilot around real decisions. Keep an opening task, guided work and a later changed case separate, then let another person attempt the handoff.

Choose a pilot path ↓
  1. 01
    Attempt before instructionIndividual response, sources and declared help.
  2. 02
    Practise with feedbackRunnable labs and useful corrections.
  3. 03
    Try a later changed taskActual interval, exposure and receiving action.

Materials ready / Participants to be confirmed

Prepare both groups.
Keep individual evidence.

These protocols are available for planning and rehearsal. They do not announce a staffed cohort, collect participants or enroll a team.

01 / AI at work

Enterprise employee pilot

Check an AI draft against its sources, correct an unsupported commitment and give the next owner an answerable action.

A narrow test of source-based judgment and handoff. It does not measure general AI fluency, productivity or live-model quality.

Prepare this pilot →

02 / Engineering delivery

FDE cohort pilot

Hand over an identifiable engineering candidate, challenge its evidence and lead a bounded receiving decision when scope changes.

A narrow test of engineering handoff and delivery judgment. This written probe does not establish coding independence, production authority or client placement.

Prepare this pilot →

Use the pilot to decide
what to improve next.

Agree a narrow question and the evidence you will retain before recruiting. A useful first group may be 10–15 consenting learners, but that planning range is not a power calculation or a success threshold.

  • Keep each person’s opening, practice and later task records, with actual attendance and missing follow-ups visible. State the denominator for every reported count; do not quietly drop unfinished or unsupported work.
  • Have two reviewers independently judge the three fictional examples on each pilot page before discussion, then use an agreed shared sample before reviewing learner work. Preserve disagreements and specific evidence; a polished sample answer is not a calibrated instrument.
  • Compare source-based decisions, required help, receiving success and remaining errors descriptively. The two public tasks can differ in difficulty; practice effects, prior exposure and missing follow-ups prevent a causal training-effect claim.
  • Use findings to revise a specific lesson or support arrangement, then test that revision. Independent qualification, client work and production authority require their own evidence and decisions.

Before launch, confirm participants, consent and handling arrangements, a facilitator, reviewers, an accountable receiving team, approved tools and a later follow-up opportunity. Leave unconfirmed details pending.