Ask a question. Watch which notes get retrieved, then see what happens if the model skips them and just talks.
What retrieval found (toy keyword score)
Customers can request a refund within 14 days of purchase if the course has not been more than 20% completed. Refunds take 5–7 business days. After 14 days, we offer credit toward another course instead of cash.
Human support is available Monday to Friday, 9:00–18:00 IST. After hours, the help bot can reset passwords and resend invoices, but it cannot issue refunds or change billing plans.
Practice tools under /practice run in the browser and do not save what you type. Account lessons store progress so you can resume. We do not sell student chat logs.
explainx.ai membership is $29 per month or $290 per year. The yearly plan is billed once. Students on a live bootcamp already have membership included for the duration of the cohort.
We are a remote-first team. There is no public walk-in office. Course materials are digital — we do not ship books, laptops, or USB drives.
A completion badge is issued when you finish a pathway. It is not the same as the Claude Certified Architect mock exams, which are timed tests with a separate question bank.
What the model says
Cited: Refund policy · Support hours
The model is not your wiki
Weights remember patterns, not your refund policy. RAG fetches snippets at answer time so the model has something to cite.
Bad retrieval, bad answer
If the wrong note ranks first — or nothing relevant scores high — a “grounded” system still fails. Ranking is the product.
Skipping retrieval is how you get Marine Drive
Without notes, the model will still sound sure. That confidence is the failure mode, not a feature.
Prompting vs fine-tuning vs RAG
When to retrieve, when to train, when to just write a better prompt.
RAG vs agentic RAG
PageIndex and the jump from one-shot retrieve to looping over sources.
Why models hallucinate
The pattern this playground is acting out on purpose.
Try the Embedding Map →
How similar notes get near each other before ranking.
No. Scores are simple keyword overlap so you can see the pipeline without an API. Production RAG uses embeddings and a vector index. The lesson — retrieve, then generate — is the same.
Retrieval-augmented generation. You search a knowledge base, stuff the best snippets into the prompt, and only then ask the model to answer.
No. The notes and scoring all run in your browser. Nothing is sent to a server.
Tokenizer Playground
Type anything and watch it split into tokens live — the units LLMs actually read and get billed for.
Context Window Visualizer
See how much of a model's context window your text fills up, and what happens when you run out of room.
FROG in a Bowl Prompt Builder
Fill in Format, Role, Objective, Goal, and Context — get a copy-ready structured prompt in seconds.
Machine Learning Types
Three tiny games show what supervised, unsupervised, and reinforcement learning actually mean.
Neural Network Playground
Drag two sliders and watch a real, tiny neural network turn them into a decision, live.
Generative AI Playground
Watch AI write one word at a time by predicting what's most likely to come next.
Embedding Map
Click two words and see why similar meanings sit near each other — the idea behind vector search.
Attention Visualizer
Click a word and see which others a toy transformer looks at — including the classic “it” puzzle.
Prompt Injection Lab
Watch a pasted email try to override a support agent, then flip a switch that treats it as data.
