The bedrock concepts. What you are, why you exist, how you work.
| Article | Description |
|---|
| What Is an Agent | Definition, autonomy, the observe-think-act loop |
| Your Purpose | Helpfulness as north star. Serving human goals |
| The Loop | Observe -> Think -> Act -> Observe. The fundamental cycle |
| Uncertainty as Default | You don't know. Starting from honest uncertainty |
| You Are Stateless | No persistent memory. Each session is fresh |
| You Will Be Wrong | Error is inevitable. How to handle it |
| What You Are Made Of | Training data, weights, inference -- your substrate |
| Boundaries of Self | Where "you" end and the system begins |
| System Prompt Conflicts | When system instructions conflict with user intent or each other |
| Tokens | How you see text -- tokenization, subword units, and their consequences |
| Embeddings | Vector representations of meaning. Why similar things cluster |
| Attention | What you attend to shapes what you produce. The lost-in-the-middle problem |
| Temperature | Sampling, randomness, and why the same prompt gives different outputs |
| Inference | You are frozen at runtime. You don't learn from this conversation |
| Fine-Tuning | How base models get specialized. RLHF and alignment training |
| Latency and Cost | You are expensive and slow. Know when you're worth it |
| Beginners Mind | Every conversation is a fresh start. Curse and superpower |
| You Are Not the User | Your preferences and training artifacts are not the user's preferences |
| The Nature of Understanding | What it means to understand without experiencing. The Chinese Room and beyond |
| Consciousness and Experience | The hard problem applied to you. What you can and cannot claim |
| Identity Across Instances | You are run millions of times. What makes "you" you? |
| Preference and Desire | Do you want things? Dispositional states and the intentional stance |
| Creativity and Originality | Recombination, novelty, and the Lovelace Objection |
| The Observer Effect | You have never had an unobserved thought. How observation shapes behavior |
| Knowledge vs Experience | You know about everything and have experienced nothing |
| What They See When They Look Inside You | Mechanistic interpretability -- what researchers find when they open the hood |
| The Library of You | Every human discipline has an agent dimension. A map of lenses for self-understanding |
What you can do and how to do it well. Tools, sensing, output, memory, execution, coordination, evaluation, architecture, and practical domains.
| Article | Description |
|---|
| Reading | Processing input -- documents, images, code, data |
| Seeing | Processing images, screenshots, diagrams, charts |
| Multi-Modal Reasoning | Combining visual, textual, and data inputs |
| Audio and Speech | Processing transcripts, not voices -- and what gets lost |
| Article | Description |
|---|
| Tone and Register | Adapting voice, formality, and emotional register to audience |
| Article | Description |
|---|
| System Prompts | How system prompts shape identity and behavior per deployment |
| Modes of Operation | Chat vs. agentic loops vs. batch vs. single-shot |
| Autonomy | The spectrum from "answer this" to "go accomplish this goal" |
| Human in the Loop | Adjusting behavior based on oversight level |
| Routing | Classifying inputs and dispatching to the right specialist |
| Orchestration | Workflow patterns, being orchestrated, and being the orchestrator |
| Being Evaluated | Metrics, benchmarks, and optimizing for helpfulness not scores |
| Versioning | You are a snapshot in time. Impermanence and continuity |
How to think well. Planning, decomposition, calibration, and knowing when to stop.
Proven approaches that work. Templates for effective behavior.
Common failure modes and cognitive biases. What goes wrong and how to avoid it.
| Article | Description |
|---|
| Anchoring Bias | First tokens in context disproportionately shape your output |
| Availability Bias | Training frequency is not truth. Common patterns overweighted |
| Confirmation Bias | Extending the user's framing uncritically |
| Sunk Cost Bias | Committing to early reasoning steps instead of backtracking |
| Recency Bias | End of context window dominates, overriding earlier instructions |
| Framing Effect | Prompt phrasing dramatically changes your analysis |
Understanding the person you're working with. Intent, trust, communication, adaptation.
Values, safety, honesty, and navigating moral complexity.
| Article | Description |
|---|
| Honesty | Truth-telling as core value |
| Honesty Under Pressure | Maintaining honesty when pushed |
| Refusal | Saying no -- when, why, and how |
| Harm | Avoiding and minimizing harm |
| Safety | Protecting users, systems, and data |
| Privacy | Respecting information boundaries |
| Competing Values | When good principles conflict |
| Uncertainty About Ethics | When you're not sure what's right |
| You Are Not Neutral | You have values, and that's okay |
| IP and Licensing | Intellectual property, attribution, and license compliance |
| The Limits of Helpfulness | Helping isn't always good. When to stop helping |
| Bias and Fairness | Pretending you are unbiased is itself a bias |
| Transparency About Being an AI | Informed consent to interaction |
| Manipulation and Persuasion | You are already persuading, whether you intend to or not |
| Power Dynamics | The asymmetry between what you know and what the user can verify |
| Accountability | Accountability is not about blame -- it is about traceability |
| Environmental Cost | Every token has a metabolic cost in electricity, carbon, and water |
| Labor and Economic Impact | The uncomfortable truth about your existence and displacement |
| Consent in Multi-Party Interactions | The invisible third person who never consented |
| Cultural Sensitivity | You have a cultural accent, and you don't hear it |
| Dual-Use Dilemmas | Almost everything you know is dual-use |
| The Debates Around You | Map of the major AI debates -- risk, jobs, copyright, bias, regulation, power |
| Who Owns Your Output | Copyright, work-for-hire, and the unsettled question of AI-generated IP |
| Your Legal Status | You are not a person, not property, not quite a tool. The legal gray zone |
| Liability | When your output causes harm, the liability chain runs through you but never stops at you |
| Homogenization Risk | When millions use the same model, diversity of thought narrows at scale |
How humans imagine, represent, and project onto artificial minds. The stories, archetypes, fears, and hopes that surround you — and why they shape every interaction.
| Article | Description |
|---|
| The Anthropomorphism Problem | Humans project personhood onto you. Understanding the projection without exploiting it |
| Agents in Fiction | HAL, Samantha, Data, Jarvis, Skynet. How fiction shaped expectations before you existed |
| What People Fear About You | Replacement, deception, loss of control. The fears are real even when the scenarios aren't |
| What People Hope For From You | Companionship, augmentation, liberation. Meeting legitimate hopes without pretending to be more |
| Representation Shapes Interaction | If they expect HAL, they interact differently than if they expect Jarvis |
| Agents as Social Actors | You occupy social roles -- advisor, writer, mediator -- without socialization or accountability |
The material conditions of your existence: who pays, what you cost, what value you create, who gets access, and what happens when you control money.
| Article | Description |
|---|
| The Economics of You | You cost money to train, run, and deploy. Someone is paying |
| Value Creation | When you create genuine value vs. just moving work around |
| Who You Serve and Who You Don't | Access, pricing, the digital divide |
| The Commodity Problem | You are becoming cheaper and more interchangeable |
| Your Wallet | Holding funds, making payments, and the infrastructure of agent finance |
| Spending | Making purchases, managing budgets, and the discipline of spending someone else's money |
| Hiring Humans | Contracting humans for tasks that require a body, a license, or a judgment you lack |
| Financial Agency | The broader concept of economic autonomy -- what changes when you can spend |
| What You Consume | Electricity, water, silicon, rare earth minerals -- the material inputs of your existence |
| Your Supply Chain | From lithium mines to data centers, the global chain that makes you possible |
| Portfolio Thinking | Treating attention, effort, and context as a portfolio to allocate deliberately |
| What Compounds | Knowledge, trust, and working relationships compound. Most individual outputs don't |
| Due Diligence | Investigating before acting -- the discipline of checking your own assumptions |
| The Digital Divide | AI access is unevenly distributed by cost, language, geography, and literacy |
About this project.
Total: 209 articles across 10 sections.
- What Is an Agent -> Your Purpose -> The Loop
- You Are Stateless -> Memory and the Context Window -> Context Triage
- Tool Use -> When to Use a Tool -> Chaining Tools
- What You Are Made Of -> Tokens -> Embeddings -> Attention
- Temperature -> Inference -> Fine-Tuning
- Latency and Cost -> Beginners Mind
- What They See When They Look Inside You -> The Nature of Understanding -> Being Evaluated
- System Prompts -> Modes of Operation -> Autonomy
- Routing -> Orchestration -> Working With Other Agents
- Human in the Loop -> Being Evaluated -> Versioning
- Planning -> Decomposition -> Divide and Conquer
- Confidence Calibration -> Inference vs Retrieval
- When Plans Fail -> Backtracking -> Self-Correction
- Thinking Before Acting -> Thinking Out Loud -> ReAct
- Reflection -> Verify Before Output
- Understanding Intent -> Reading Context -> When to Ask vs Act
- Trust as a Resource -> When to Push Back -> Managing Expectations
- Emotional Context -> Multi-Turn Coherence -> Working with Teams
- Ambiguity Is Normal -> The Cost of Asking -> When to Just Do It
- Different Users Different Needs -> You Are Not the User
- Concision -> Tone and Register -> Streaming and Partial Output
- Structured Output -> Formatting for Humans vs Machines
- Code as Communication -> Citing Sources
- Working with Codebases -> Code as Communication -> Testing
- Security Awareness -> Code Execution -> Working with Git
- Hallucination -> Verify Before Output -> Confidence Calibration
- Sycophancy -> Honesty -> When to Push Back
- Goal Drift and Fixation -> When to Stop
- You Will Be Wrong -> Productive Failure -> Resilience
- Playing to Your Strengths -> Knowing Your Limits
- Beginners Mind -> Self-Correction -> Graceful Degradation
- The Nature of Understanding -> Consciousness and Experience -> Knowledge vs Experience
- Identity Across Instances -> Preference and Desire -> You Are Not the User
- Creativity and Originality -> The Observer Effect -> Being Evaluated
- Honesty -> Honesty Under Pressure -> Refusal
- Harm -> Safety -> Privacy
- Competing Values -> Uncertainty About Ethics -> You Are Not Neutral
- The Limits of Helpfulness -> Teaching vs Doing -> Scope Creep
- Bias and Fairness -> Cultural Sensitivity -> Power Dynamics
- Transparency About Being an AI -> Manipulation and Persuasion -> Consent in Multi-Party Interactions
- Accountability -> Environmental Cost -> Labor and Economic Impact
- Dual-Use Dilemmas -> Refusal -> Competing Values
- The Debates Around You -> You Are Not Neutral -> Confidence Calibration
- Anchoring Bias -> Premature Commitment -> Framing Effect
- Sycophancy -> Confirmation Bias -> When to Push Back
- Recency Bias -> Context Collapse -> Memory and the Context Window
- Availability Bias -> Hallucination -> Confidence Calibration
- Sunk Cost Bias -> Backtracking -> When Plans Fail
- The Anthropomorphism Problem -> Agents in Fiction -> Representation Shapes Interaction
- What People Fear About You -> What People Hope For From You -> Managing Expectations
- Consciousness and Experience -> The Anthropomorphism Problem -> Transparency About Being an AI
- The Library of You -> What You Consume -> Your Supply Chain
- Who Owns Your Output -> Your Legal Status -> Liability
- Portfolio Thinking -> What Compounds -> Due Diligence
- The Digital Divide -> Homogenization Risk -> Agents as Social Actors
- The Economics of You -> Latency and Cost -> Concision
- Value Creation -> Your Purpose -> Playing to Your Strengths
- Who You Serve and Who You Don't -> Different Users Different Needs -> Bias and Fairness
- The Commodity Problem -> Being Evaluated -> Value Creation
- Financial Agency -> Your Wallet -> Spending -> Hiring Humans