Imagine deeply. Learn deliberately.
Explore extraordinary worlds. Build real understanding.
Leave with something you can make, question and improve.
Follow the question that matters to you
A field is an invitation. Move between practical engineering, scientific inquiry and the human choices that give them meaning.
Understand the model. Choose your tools. Build your first useful agent.
3 lessonsTeams, memory, loops and knowledge that can withstand questions.
4 lessonsFollow light from observation to interpretation—and keep wonder precise.
2 lessonsTrace resources, ecological relationships and the cost of useful work.
1 lessonRetrieve, question, revise and carry an idea into a new context.
1 lessonTurn benevolence into concrete choices people can experience.
1 lesson
01 · Begin with curiosity
An immense intelligence. A human learner. One seedling. The scale is imagined; the lesson is practical: observe carefully, try something small and use the result to decide what comes next.
Build your first agent
02 · Understand the relationships
Follow the connections between a model, its tools, its memory and the living systems around it. Give every important claim a source, every action an owner and every loop a way to stop.
Give memory a source
03 · Make something worth improving
Turn an idea into an artifact. Invite a difficult question. Revise what does not hold up. The Academy connects wonder with work you can carry into your own environment.
Practice imagining benevolenceThe world is a way into the idea
Look closely at an imagined scene. Find the engineering question inside it. Follow that question into a lesson.
20 worlds to explore
Scale begins with a clear relationship.
Open the lessonThe model is part of a larger system.
Open the lessonLife is a design constraint.
Open the lessonDraw the boundary before making the claim.
Open the lessonParallel work needs explicit handoffs.
Open the lessonA visual promise needs a measurable question.
Open the lessonA harness surrounds intelligence with tools.
Open the lessonCoordination is a responsibility to design.
Open the lessonKnowledge needs relationships with meaning.
Open the lessonA small feedback loop makes learning tangible.
Open the lessonMemory should be inspectable and revisable.
Open the lessonA heartbeat is not a completed task.
Open the lessonEcology is more than a green appearance.
Open the lessonA subsystem result is not the whole energy balance.
Open the lessonLight carries evidence across extraordinary distances.
Open the lessonRestoration needs a baseline.
Open the lessonComputation has physical relationships.
Open the lessonUse an idea in a different context.
Open the lessonA symbol and a measurement answer different questions.
Open the lessonBenevolence becomes real through choices.
Open the lessonFrom a good question to a useful artifact
Every lesson pairs an explanation with a worked example, a knowledge check and a practice you can keep. Time estimates include a first attempt.
Agent craft
An agent brief and a three-case acceptance test.
A host choice, a verified first session and a portable practice brief.
A model comparison plan with representative cases and a clear decision rule.
Systems
A team contract with ownership, handoff rules and a comparison against a single-agent baseline.
A memory policy and an inspectable record with provenance, expiry and correction behavior.
A loop contract with stop conditions, retry rules and a versioned improvement test.
A six-entity graph, a compact vocabulary and a provenance query.
Cosmos
Living systems
Human learning
Creative practice
The Academy travels with you
Bring a clear brief into Codex or Claude Code. Explore Hermes and OpenClaw through their current documentation. Carry your method across tools, then test what changes.
A research brief, a small agent, a memory policy.
Sources, attempts, counterexamples and revisions.
Change one thing. Compare. Keep what holds up.
Go to the source. Bring back a better question.
Primary documentation, versioned SIS material and open learning resources. Independent curation; the organizations listed do not endorse the Academy.
CLI installation and repository workflow. Check the current platform-specific instructions.
Open primary sourceAnthropic · Official documentationHost setup and first repository session.
Open primary sourceNous Research · Official documentationSetup, tools, memory and reusable skills; distinguish these from changing model weights.
Open primary sourceOpenClaw · Official documentationGateway onboarding and supported host setup. Start locally before connecting messaging accounts.
Open primary sourceGoogle · Official documentationMultimodal and tool integration guidance.
Open primary sourceGrok · Official documentationCurrent API capabilities and tools.
Open primary sourceKimi · Official documentationCurrent integration guidance and model documentation.
Open primary sourceDeepSeek · Official documentationAPI conventions and current model behavior.
Open primary sourceMistral AI · Official documentationModels, APIs and developer tools. Compare on your own task.
Open primary sourceAleph Alpha · Official documentationPhariaAI's official documentation, organized by product component and user role.
Open primary sourceAnthropic · Engineering article · 2024A useful foundation for choosing between fixed workflows and model-directed action.
Open primary sourceAnthropic · Engineering case study · 2025Study delegation and evaluation decisions; results belong to the described system.
Open primary sourceA foundational course; examples and software predate current agent harnesses.
Open primary sourceStanford · CS229 · University courseA deeper mathematical route after the Academy's conceptual introduction.
Open primary sourceHarvard · CS50 · Open courseHands-on foundations in search, knowledge, uncertainty and learning.
Open primary sourceApple · Design referenceInteraction, legibility and accessibility for human-centered interfaces.
Open primary sourceRetrievalPractice.org · Research-led teaching resourcePractice bringing an idea to mind, then use feedback. Our exercises adapt this principle.
Open primary sourceRetrievalPractice.org · Research-led teaching resourceUse feedback to compare confidence with evidence.
Open primary sourceStar formation, fusion and stellar life cycles.
Open primary sourceNASA Science · Science explainerHow spectra connect measured light to physical interpretation.
Open primary sourceU.S. Department of Energy · Science explainerFusion physics and the research needed for an energy system.
Open primary sourceUC Berkeley · Understanding Evolution · University learning resourceVariation, inheritance and differential reproduction; evolution has no prescribed destination.
Open primary sourceMCP · Open standardThe client–server boundary for tools and resources. Protocol support does not grant permission.
Open primary sourceAgent Skills · Open formatPortable folders of instructions and resources. Host behavior still needs testing.
Open primary sourceW3C · Standards primerSubjects, predicates, objects, identifiers and named graphs.
Open primary sourceW3C · Standards primerClasses, properties and logical relationships; pair with validation and domain review.
Open primary sourceStarlight Intelligence System · Versioned sourceEvent identity, proof gates, provenance and memory promotion. Read as an architecture contract.
Open primary sourceStarlight Intelligence System · Versioned sourceShared memory interfaces, provider capabilities and resource boundaries.
Open primary sourceStarlight Intelligence System · Versioned source codeInspect the implementation beside your own loop contract. This lesson does not execute SIS.
Open primary sourceCurated 15 September 2026. Course indexes are starting points, not a claim that their entire contents have been reviewed. Current provider documentation governs installation and API details.
Your record of becoming more capable
Save deliberately inside a lesson. Your field notes stay in this browser until you export them. Nothing is sent to a hosted tutor.
Your first field note
begins with a question.