# BuildSpark AI

> A practical map and cards for builders who already use AI.

Concept cards, the map, and Ask are free. The Foundations pack includes a token budget set on its page. The other track packs are not written yet. There is no account and no payment on this site. Coming soon concepts are listed on the track pages and do not have cards yet.

## Live concepts

- [Tokens](https://buildsparkai.com/tokens): The unit models read, write, and bill in. Plain text: https://buildsparkai.com/tokens.md
- [Context Window](https://buildsparkai.com/context-window): The fixed text one request can hold: instructions, history, documents, and the answer. Plain text: https://buildsparkai.com/context-window.md
- [Prompting](https://buildsparkai.com/prompting): How you state the task this turn: the reader, the shape, and what to leave out. Plain text: https://buildsparkai.com/prompting.md
- [Prompt Caching](https://buildsparkai.com/prompt-caching): Reusing an unchanged prefix so a repeated prompt can cost less than sending it from scratch. Plain text: https://buildsparkai.com/prompt-caching.md
- [Model Selection & Routing](https://buildsparkai.com/model-selection): Picking a model, and sometimes a route between models, for the job in front of you. Plain text: https://buildsparkai.com/model-selection.md
- [Structured Outputs](https://buildsparkai.com/structured-outputs): Asking for a reply that follows a shape, such as fields you name, so code can read it. Plain text: https://buildsparkai.com/structured-outputs.md
- [Hallucination](https://buildsparkai.com/hallucination): A reply that states something the prompt and the sources do not support. Plain text: https://buildsparkai.com/hallucination.md
- [Evaluation basics](https://buildsparkai.com/evaluation-basics): Checking a model on examples you already know the answer to, before you trust a change. Plain text: https://buildsparkai.com/evaluation-basics.md
- [Temperature](https://buildsparkai.com/temperature): A sampling control that shifts how often less likely next tokens are picked. Plain text: https://buildsparkai.com/temperature.md
- [RAG](https://buildsparkai.com/rag): Retrieve relevant text, then ask the model to answer from that text. Plain text: https://buildsparkai.com/rag.md
- [Chunking](https://buildsparkai.com/chunking): Splitting documents into pieces that retrieval can store and return. Plain text: https://buildsparkai.com/chunking.md
- [Embeddings for retrieval](https://buildsparkai.com/embeddings-retrieval): Using embeddings to find stored chunks that are close to a query. Plain text: https://buildsparkai.com/embeddings-retrieval.md
- [Hybrid search](https://buildsparkai.com/hybrid-search): Combining word matching and embedding search so exact terms and paraphrases can both hit. Plain text: https://buildsparkai.com/hybrid-search.md
- [Reranking](https://buildsparkai.com/reranking): A second pass that reorders retrieved chunks before they go into the prompt. Plain text: https://buildsparkai.com/reranking.md
- [Agent loop](https://buildsparkai.com/agent-loop): A cycle of model step, tool result, and another model step, until a stop. Plain text: https://buildsparkai.com/agent-loop.md
- [Tool calling](https://buildsparkai.com/tool-calling): The model asks your code to run a named function, then reads the result. Plain text: https://buildsparkai.com/tool-calling.md
- [Tool design/MCP](https://buildsparkai.com/tool-design): How a tool is named, described, and exposed, including over MCP. Plain text: https://buildsparkai.com/tool-design.md
- [Context engineering](https://buildsparkai.com/context-engineering): Deciding what goes into the window for this step, and what stays out. Plain text: https://buildsparkai.com/context-engineering.md
- [Prompt injection](https://buildsparkai.com/prompt-injection): Untrusted text that tries to override the instruction or the tool rules. Plain text: https://buildsparkai.com/prompt-injection.md
- [Production evaluation](https://buildsparkai.com/production-evaluation): A repeatable eval set you run on changes that will face real users. Plain text: https://buildsparkai.com/production-evaluation.md
- [Observability](https://buildsparkai.com/observability): Logs and traces of prompts, tool calls, and replies so a failure can be inspected. Plain text: https://buildsparkai.com/observability.md
- [Cost forecasting](https://buildsparkai.com/cost-forecasting): Estimating spend from token volume, rates, and how often a workflow runs. Plain text: https://buildsparkai.com/cost-forecasting.md
- [Latency](https://buildsparkai.com/latency): How long a request takes, including retrieval, tools, and the model. Plain text: https://buildsparkai.com/latency.md
- [Fallbacks & rate limits](https://buildsparkai.com/fallbacks): What the product does when a model errors, slows down, or hits a rate limit. Plain text: https://buildsparkai.com/fallbacks.md
- [Privacy & data handling](https://buildsparkai.com/privacy-data): What you send, what is stored, and what should never enter a prompt. Plain text: https://buildsparkai.com/privacy-data.md
- [Safety](https://buildsparkai.com/safety): Limits on harmful or disallowed use, in the product and in the model. Plain text: https://buildsparkai.com/safety.md
- [Monitoring & drift](https://buildsparkai.com/monitoring-drift): Watching quality and behavior after launch, when data or models change. Plain text: https://buildsparkai.com/monitoring-drift.md
- [Fine-tuning vs prompting vs RAG](https://buildsparkai.com/fine-tuning): Choosing among a prompt change, retrieval, and a trained update. Plain text: https://buildsparkai.com/fine-tuning.md
- [Licensing](https://buildsparkai.com/licensing): What the model and the data allow you to ship. Plain text: https://buildsparkai.com/licensing.md

## Tracks

- [Foundations](https://buildsparkai.com/tracks/foundations): How models read text, how much fits in one request, and how you state the task.
- [Retrieval & Grounding](https://buildsparkai.com/tracks/retrieval): How a product finds source text and keeps an answer tied to it.
- [Agents](https://buildsparkai.com/tracks/agents): How a model asks for tools, loops on the result, and when a person still decides.
- [Ship & Maintain](https://buildsparkai.com/tracks/ship): How a workflow is evaluated, watched, and kept within cost and privacy limits after it ships.

## Packs

- [Foundations pack](https://buildsparkai.com/packs/foundations): Token budget set, with a worked cost example and a pre-launch checklist. Listed at €39 once. No checkout.
- [Retrieval & Grounding pack](https://buildsparkai.com/packs/retrieval): Coming soon. €39 one-off for practice beside the free Retrieval & Grounding cards.
- [Agents pack](https://buildsparkai.com/packs/agents): Coming soon. €39 one-off for practice beside the free Agents cards.
- [Ship & Maintain pack](https://buildsparkai.com/packs/ship): Coming soon. €39 one-off for practice beside the free Ship & Maintain cards.

## Site

- [Home](https://buildsparkai.com/): Practical map and cards for builders who already use AI.
- [Map](https://buildsparkai.com/map): The curriculum and its prerequisite links.
- [Ask](https://buildsparkai.com/ask): Keyword match against concept titles, summaries, and live card text. No model call.
- [Pricing](https://buildsparkai.com/pricing): Free cards, €39 track packs, all-access, and teams. No checkout.
- [About](https://buildsparkai.com/about): What BuildSpark AI is, who it is for, and how cards are sourced.
- [Changelog](https://buildsparkai.com/changelog): Updates that have shipped on the site.