Add relevant documents, retrieved passages, or stored information to the current request. This supplies context beyond the model’s training without retraining it. Missing, stale, or misleading material can still produce a poor answer.
Level 02
Supply relevant information
Searching your documents and giving the results to the model.
Storing facts as entities and relations, for questions that span several documents.
Each line names the failure that justifies moving to a higher level.
The person asking wants an answer with the passages behind it, not a ranked list of passages to read themselves.
The answer needs facts joined across documents, or you must show where each fact came from.
One retrieval is not enough and the next query depends on what the last one found.
Which entity to follow next depends on what the last hop returned, so the queries cannot be written in advance.
What has to survive between sessions is work still in progress, not facts about the person you are talking to.
Each one needs this level and no higher, and says why.
Uses RAG, structured output and an eval set. Level 2 is enough because a single search answers most questions.
Retrieval over the documents an engineer already has, answered with citations that can be checked. The case that matters is a change notice contradicting the datasheet on one number, where the right answer depends on the board revision. Level 2 is enough because one search finds the passage.
Names listed 09/19/2026. 223 of 223 registry entries have been checked against the maker's own page; the registry marks the rest as unchecked.
Open problems at this level, what people are trying, and the source each rests on. Read 09/19/2026. This block ages faster than the rest of the page, and nothing in it predicts which approach wins.
Which embedding model you pick changes which questions your search answers, and no model is best at everything. Choosing one is still a matter of testing on your own documents rather than reading a ranking.
What people are trying
We find that no particular text embedding method dominates across all tasks.
Where it bites: Embeddings and search
A model that advertises a large context window can still fail to use a fact sitting inside it, once the fact does not share wording with the question. The window is what the model will accept, not what it will reliably find.
What people are trying
performance degrades significantly as context length increases
Where it bites: Context engineering
Retrieval cuts invention down and does not end it, and most checkers score a whole answer at once. An answer that is right except for one made-up sentence scores like a right answer, which is the case a reader most needs flagged.
What people are trying
Retrieval-augmented generation (RAG) reduces but does not eliminate hallucination, and existing detectors return a single answer-level score that does not indicate which sentence is unsupported, or why.
Where it bites: Retrieval-augmented generation (RAG)
When a memory store holds two facts that contradict each other, an old preference and the one that replaced it, there is no settled rule for which should govern the answer. A system can retrieve the stale one and still answer correctly, or retrieve the right one and answer wrongly, and an outcome score cannot tell those apart.
What people are trying
providing limited insight into how systems retrieve and rank temporally valid, factually correct, and contextually applicable memory evidence under conflicting alternatives
Where it bites: Memory
Pages at this level last reviewed 09/19/2026. Pages unreviewed for 90 days are flagged for another pass. Markdown version of this page