Complete AI Labs and Hand Off Reviewed Work

Who should use this article

This article is for workspace users completing the AI Adoption Planner, AI Governance & Compliance Lab, Applied ML Lab, or AI Response Lab. A lab completion is learning evidence. It is not production approval, compliance certification, model validation, or release authorization.

Before you begin

• Open the intended workspace through My Batoi and confirm that you have access to Batoi Learn.
• Choose an approved scenario and name the accountable owner for any later workplace decision.
• Use fictional, sanitized, or explicitly approved information. Do not enter credentials, secrets, unnecessary personal information, production datasets, confidential prompts, or customer records.
• Confirm that Govern or Build users are ready to review a proposal before you create a handoff.

Complete the common learning loop

1. Open the lab. In Learn, select the lab and start or resume your private session.
2. Learn. Read the scenario, method, success criteria, limitations, and knowledge checks.
3. Try. Complete the guided fields using evidence and explicit assumptions. Save your progress before leaving the page.
4. Evaluate. Select Save and Assess where available. Review the overall result and every criterion, not only the score.
5. Improve. Correct weak evidence, unclear ownership, unsupported claims, and missing stop conditions. Reassess when the lab permits another attempt.
6. Reflect. Record what changed in your judgment, what remains uncertain, and what evidence would change the conclusion.
7. Complete. Select Complete Learning Loop only when the evidence and reflection are ready for review.
8. Apply deliberately. Use a proposal button only when it appears for the completed Lab and your role permits it. In the AI Adoption Planner, the action creates one planned target project and a handoff that reviewers can assess; it does not approve or execute the proposed work automatically.

AI Adoption Planner

1. Define the business outcome, current process, affected people, measurable baseline, and strongest practical non-AI alternative.
2. Assess data, process, people, technology, governance, change, and sponsorship readiness. State the confidence and evidence for each rating.
3. Record low, base, and high value assumptions, implementation and review costs, risks, guardrails, human oversight, and a reversible pilot.
4. Define success measures, stop conditions, accountable owners, evidence dates, and the go, pivot, or stop decision rule.
5. Save and assess the plan, improve weak criteria, add a reflection, and complete the learning loop.
6. Select Propose to Govern when the next work is obligation, risk, control, evidence, or approval review. Select Propose to Build when the next work is a clearly scoped implementation plan or experiment. Do not create both proposals unless two separate accountable reviews are genuinely required.

AI Governance & Compliance Lab

1. Identify the AI system or initiative, intended outcome, affected people, owners, and reviewers.
2. Record the risk classification and rationale, applicable obligations and policy expectations, required controls, human oversight, and evidence.
3. State the recommended decision, conditions, limitations, unresolved questions, and what changed in your judgment.
4. Complete the learning loop. When your role and session permit it, select Propose Evidence to Govern.
5. In Govern, an authorized reviewer checks the handoff against authoritative policies, obligations, risks, controls, evidence, and approval authority. The reviewer decides whether to create, link, return, or reject governed work.

Applied ML Lab

1. Define the decision, baseline, data boundary, permitted features, data limitations, and consequences of important errors.
2. Compare suitable algorithm families and choose measures, acceptance thresholds, and harm-related guardrails.
3. Run only the safely limited Lab experiment. Record configuration, reproducibility details, results, limitations, human review, monitoring, and approval requirements.
4. Save and assess, improve the evidence, reflect, and complete the learning loop.
5. When available, select Propose Experiment to Build. A Builder must confirm the app or project, repository, environment, approved data source, evaluation plan, security controls, and rollback conditions before creating implementation work.

AI Response Lab

1. Define the response task, audience, constraints, evaluation criteria, and unsafe or unacceptable outcomes.
2. Generate or enter only the clearly scoped responses, compare them against the criteria, and record evidence for the selected winner or tie.
3. Review the automatic rubric score, revise the response or criteria when needed, and record what you will change next time.
4. Complete the learning loop and retain the result in your learning activity and mastery evidence.

Verify a handoff

• The Learn session shows completed status and the current assessment result.
• The proposal confirmation names Govern or Build and states that review is required.
• The target reviewer can identify the source lab, learner, workspace, proposal purpose, and safe structured summary.
• The AI Adoption Planner target project exists in planned state, and no production change, approval, policy, model, deployment, or workspace access was created automatically.
• Every unresolved condition has an owner and next action.

Troubleshooting and recovery

Next step

Continue with Track Mastery and Manage Learning in Batoi Learn, or return to the Batoi Platform Product Guide.