Nine example formats show not just what can be built, but what good AI work contains: goal → AI role → human check → verifiable artifact. These are not claims about specific clients or learners.
All content remains available without JavaScript. The filter runs locally in the browser and sends nothing out.
Topic → 3+ sources → claim table → short presentation of what was verified and what was not.
01GoalSeparate a verifiable fact from a convincing-looking answer.
02AI roleHelps form search queries, compare versions and organize findings.
03Human checkOpens the sources, marks contradictions and explains why a conclusion is trustworthy.
04ArtifactSource table + claim map + mini presentation.
Character bible, visual rules, storyboard and a short final story with clear separation between the learner’s idea and AI assistance.
01GoalTurn an idea into a coherent system of world and character rules.
02AI roleExplores alternatives, reference directions and consistency checks.
03Human checkOwns the creative choices and can explain what came from them versus AI assistance.
04ArtifactCharacter bible + storyboard + finished story.
Rules, states, simple logic and a prototype the learner can explain.
01GoalBuild a small working system instead of a one-shot generation.
02AI roleHelps decompose mechanics into rules, states and test scenarios.
03Human checkChecks causality, repairs broken rules and defends the project logic.
04ArtifactGame rules + state map + playable prototype.
A reusable research template: question, sources, claims, verification, memo and presentation.
01GoalMake research a repeatable process instead of a pile of browser tabs.
02AI roleExpands search, groups evidence and surfaces conflicting versions.
03Human checkEvaluates source quality and owns the argued conclusion.
04ArtifactResearch template + evidence map + memo + presentation.
A simple assistant or workflow for a real task: input, rules, output, failure cases and README.
01GoalTurn a task into a specification that can be tested.
02AI roleHelps with decomposition, prompt/system rules, glue code and test cases.
03Human checkDefines boundaries, checks failures and decides where automation must stop.
04ArtifactPrototype + test cases + README + failure map.
User problem, prototype, test scenarios, limitations and a short demo.
01GoalTest a product hypothesis with a working artifact.
02AI roleExplores directions, helps assemble the interface and accelerates iteration.
03Human checkCuts scope, talks to users and owns product decisions.
04ArtifactWorking prototype + QA checklist + demo + known limits.
A repeatable system for search, source comparison, synthesis and verification.
01GoalShorten the path from question to decision without losing source discipline.
02AI roleCollects alternatives, compares evidence and structures a decision brief.
03Human checkChecks primary sources, assumptions and owns the final decision.
04ArtifactReusable workflow + source map + decision brief template.
An assistant or automation serving one repeatable process, with ownership and fallback.
01GoalAutomate one process while keeping an explicit owner and rollback path.
02AI roleRuns repeatable transforms, routing and first-pass processing.
03Human checkDefines permissions, failure modes, acceptance criteria and fallback.
04ArtifactWorkflow + acceptance tests + handoff + rollback path.
Task templates, verification criteria, data boundaries and repeatable instructions.
01GoalTurn scattered successful prompts into a governed operating system.
02AI roleHelps formalize repeatable templates, checks and execution variants.
03Human checkDefines permissions, data boundaries, accountability and escalation rules.
04ArtifactPlaybook + task templates + verification rules + governance notes.
We choose the level, task and depth, then define the success criteria before starting.