AGENTIC AGENCY/ RESOURCES

Resources / AI fundamentals

AI fundamentalsField guide12 MIN READ

20 AI concepts worth understanding, from basic to advanced

A progressive field guide from simple model vocabulary to how modern AI systems actually work.

↓ Download guide Read online ↓

Free to read and download. No signup required.

/// THE_OUTCOME

Learn the concepts that make AI products, agents, and trade-offs easier to reason about.

WHO THIS IS FOR

Students, professionals, creators, and curious builders

/// THE_GUIDE

20 actionable checks.

Tick off each item as you work. Your progress is saved in this browser.

PART 01

The foundation

01. ModelA trained system that maps an input to a prediction or generated output.
02. Training vs inferenceTraining adjusts the model; inference uses it to answer requests.
03. TokensPieces of text the model processes, not necessarily whole words.
04. Context windowThe amount of material available during a single interaction.
05. HallucinationA fluent but unsupported or incorrect model output.

PART 02

Getting useful answers

06. PromptingInstructions and context that shape a model's response.
07. TemperatureA parameter influencing the randomness of sampling.
08. EmbeddingsVector representations used to compare meaning or similarity.
09. Retrieval-augmented generationSupplying retrieved external information at answer time.
10. Tool callingLetting a model request operations performed by external software.

PART 03

Building systems

11. Fine-tuningFurther training a model on task-specific examples.
12. QuantizationUsing reduced numerical precision to shrink model memory and compute.
13. Latency vs throughputTime per response versus total work processed over time.
14. Evaluation setsRepeatable examples used to assess performance and regressions.
15. Agent orchestrationCoordinating planning, tools, and multi-step execution.

PART 04

Advanced ideas

16. AttentionA mechanism for weighing relationships between token representations.
17. Mixture of expertsRouting computation through selected model sub-networks.
18. DistillationTraining a smaller model to approximate a larger system.
19. Inference-time computeSpending more computation to improve a response at runtime.
20. Alignment and guardrailsMethods to guide behavior and constrain risky actions.

/// NEXT_STEPS

Useful work starts with clearer decisions.

Keep a copy of this guide or explore another resource.

↓ Download guide All resources →