Use AI with a process, not a guess

Goal: Create a short briefing that separates facts, uncertainty, and next questions.

Audience: Students, workers, managers, lifelong learners, and small teams researching public topics.

Inputs: A public question, 2 to 5 trusted source links or excerpts, audience, desired length, and decision context. Do not include private records.

Preparation: Write the question in one sentence. Gather direct sources before asking AI to summarize. Decide what claims need verification.

Human checkpoints: Check all dates, names, numbers, legal or policy claims, quotes, and source relevance. A person decides what to use.

Risks: Fake citations, outdated facts, one-sided sources, missing caveats, and overconfident conclusions.

Success criteria: The briefing names what is known, what is uncertain, what sources support the summary, and what still needs checking.

Reusable template: Create a briefing on [question] for [audience]. Use only the source notes I provide. Separate summary, key facts, uncertainty, opposing evidence, and next questions. Do not invent sources. List claims I should verify.

Do not use when: The question needs professional legal, medical, tax, financial, safety, or emergency advice; the sources contain private data; or the decision is high stakes.

Goal: Produce a clear follow-up note with decisions, action items, owners, dates, and open questions.

Audience: Team members, volunteers, students, project groups, or small-business staff.

Inputs: Low-risk notes, meeting purpose, attendee roles, known decisions, and deadlines. Use fictional notes for practice.

Preparation: Confirm the notes do not include confidential personnel, legal, financial, medical, security, or customer-sensitive information. Decide who approves the final follow-up.

Human checkpoints: Verify attendees, decisions, commitments, deadlines, sensitive details, and tone before sending.

Risks: Invented action items, wrong owners, false deadlines, exposure of private content, or a tone that sounds final before decisions are approved.

Success criteria: People can tell what was decided, who owns each task, when it is due, and what still needs discussion.

Reusable template: Turn these safe meeting notes into a draft follow-up. Include decisions, action items, owners, due dates, open questions, and items needing confirmation. Do not invent commitments. Mark anything uncertain.

Do not use when: The meeting involves HR, discipline, legal advice, medical details, confidential finance, security incidents, layoffs, children, investigations, or sensitive customer records.

Goal: Build a 30-day learning plan with practice, evidence, reflection, and safety checks.

Audience: Beginners, students, workers, retirees, job seekers, and lifelong learners.

Inputs: Skill goal, current level, weekly time range, preferred learning style, public resources, and safe practice topics.

Preparation: Keep the goal narrow. Use public or fictional information. Check school, workplace, or program rules if the plan affects assignments or employment.

Human checkpoints: Review whether the plan is realistic, accessible, and safe. Ask a teacher, mentor, peer, or supervisor for feedback when possible.

Risks: Overly broad goals, unsafe data use, unrealistic time demands, skipped practice, and relying on AI instead of learning.

Success criteria: The plan has weekly tasks, one artifact, reflection questions, and a review date.

Reusable template: Create a 30-day learning plan for [skill] for a [beginner/intermediate] learner with [hours per week]. Include weekly goals, practice tasks using public information, one evidence artifact, reflection questions, and safety checks. Do not promise outcomes.

Do not use when: The plan involves diagnosis, therapy, legal deadlines, professional licensing decisions, medical treatment, financial commitments, or urgent safety needs without qualified guidance.

Goal: Map a small workflow and find safe places where AI might assist, not blindly automate.

Audience: Small-business owners, managers, team leads, freelancers, and nonprofit staff.

Inputs: A low-risk workflow name, steps, roles, handoffs, public or fictional examples, and current pain points.

Preparation: Do not include customer lists, payment data, employee records, contracts, tax records, medical information, passwords, or private messages. Start with a workflow that can be improved without high-stakes decisions.

Human checkpoints: Review privacy, policy, customer promises, accessibility, bias, and the final decision owner.

Risks: Automating a broken process, exposing customer data, inventing policies, reducing service quality, or removing human escalation.

Success criteria: The output includes a workflow map, bottlenecks, risks, human checkpoints, success measures, and a clear decision about whether to test AI.

Reusable template: Review this low-risk workflow: [workflow]. Steps are [steps]. Pain points are [pain points]. Suggest where AI might assist, where human review is needed, what data must stay out, what risks to watch, and how to measure success. Do not recommend full automation unless risk is low and review is clear.

Do not use when: The workflow affects hiring, firing, credit, housing, health, safety, legal rights, taxes, regulated records, children, employee discipline, or confidential customer decisions.

Goal: Turn real experience into a truthful skill inventory for resumes, interviews, or learning plans.

Audience: Job seekers, students, career changers, entry-level workers, and employed learners.

Inputs: Non-sensitive summary of experience, target role, public job posting language, and examples you can explain honestly.

Preparation: Remove private employer, customer, student, patient, coworker, salary, ID, and reference information. Use a role summary rather than a full personal history.

Human checkpoints: Check that every skill, tool, credential, result, and project is real and explainable. Remove unsupported metrics and hidden keywords.

Risks: Inflated expertise, fake projects, copied posting language, private data exposure, and AI-written bullets the applicant cannot defend.

Success criteria: The inventory lists real tasks, tools, review steps, outcomes, evidence artifacts, and gaps to learn next.

Reusable template: Using this safe summary of my experience and this target role, help me organize a truthful skill inventory. Separate proven skills, developing skills, examples I can explain, evidence artifacts, and gaps. Do not invent experience, metrics, tools, credentials, employers, or outcomes.

Do not use when: The user wants fake experience, hidden keyword stuffing, false credentials, misleading references, or unauthorized use of confidential work samples.