NEWVectors or files. Pick a path.Start →

    Guides

    Vendor-neutral, engineer-written guides to the concepts behind multimodal AI: perception, retrieval, embeddings, and the infrastructure agents use to see, hear, and search unstructured data. Learn the idea first; then see how Mixpeek applies it.

    141 guides across 21 topics

    Retrieval
    12 min read

    Why Do I Get Different Search Results Every Time I Run the Same Query?

    Almost every vector search engine is approximate: it walks a fraction of the index instead of comparing your query against every vector, and the fraction it walks can differ between runs. Five things produce run-to-run variation, they leave different fingerprints, and telling them apart takes about ten minutes. This is how to find which one you have and what to change.

    Vector Search
    ANN
    Recall
    Sep 2026Read guide
    Visual Search
    10 min read

    Why Does My Image Search Return Similar-Looking Items Instead of the Exact One?

    An embedding model is trained to place things that look alike near each other, so a different shoe in the same colourway scores almost as well as the shoe you photographed. Similarity and identity are different questions. This explains why one embedding cannot answer both, how to tell which one you are asking, and what to add so the exact item comes back first.

    Image Search
    Visual Search
    Product Search
    Sep 2026Read guide
    Search & Discovery
    10 min read

    How Do I Find Out Where a Video Came From?

    Someone sent you a clip and you want to know who filmed it, when, and whether it is what it claims to be. Reverse image search on a frame is the first move and it fails more often on video than people expect. This covers what each method actually catches, how to pull the right frames, how to spot recycled footage and AI-generated video, and what to do when the clip is cropped or re-encoded.

    Reverse Video Search
    Video
    Verification
    Sep 2026Read guide
    Search & Discovery
    9 min read

    Why Do My Photos Have No Date or Location, and Can I Get It Back?

    Almost every platform strips EXIF when you upload, so the photo you downloaded back has no date, no GPS and no camera. This explains which fields survive, which are gone for good, how to tell a stripped file from one that never had the data, and what you can still recover from the picture itself.

    EXIF
    Image Metadata
    GPS
    Sep 2026Read guide
    Search & Discovery
    10 min read

    Why Can't My Search Find Exact Part Numbers, Codes or Names?

    Semantic search compares meanings, and a part number, an error code or an unusual surname carries almost none, so the index returns records that look alike instead of the one that matches. Run a keyword (BM25) search beside the vector search, merge the two lists by rank, and use an exact filter when the identifier lives in its own field.

    Hybrid Search
    BM25
    Keyword Search
    Sep 2026Read guide
    Architecture
    18 min read

    Multi-Index Search Architecture: How to Combine Visual, Audio, and Text Embeddings for Rich Media

    A systems-design guide to building search over rich media by decomposing assets into multiple feature streams, storing them in separate indexes, routing queries, and fusing scores. Covers index-per-modality vs. fused-space design, RRF and weighted fusion, query routing heuristics, and production trade-offs.

    Architecture
    Search
    Multimodal
    Sep 2026Read guide
    Agent Perception
    18 min read

    Object Decomposition and Layered Indexing for AI Agent Perception

    A practical architecture guide for turning video, audio, images, and documents into searchable evidence that agents can inspect, cite, filter, and reason over.

    Agent Perception
    Multimodal Search
    Video Understanding
    Sep 2026Read guide
    Retrieval
    20 min read

    Hybrid Search Fusion: How to Combine Dense and Lexical Retrieval Without Breaking Ranking

    An agent searching transcripts, OCR text, and captions needs both meaning (dense vectors) and exact terms (BM25), but the two return scores on incompatible scales that you cannot simply add. This guide teaches the real fusion mechanics: why score distributions make naive normalization fail, the exact math of Reciprocal Rank Fusion and how its k parameter behaves, weighted convex combination with proper normalization, and how to choose and tune a fusion method against a labeled set.

    Hybrid Search
    Reciprocal Rank Fusion
    BM25
    Sep 2026Read guide
    Retrieval
    19 min read

    BM25 and the Inverted Index: The Lexical Retriever Every Hybrid Search Treats as a Black Box

    Every hybrid search pipeline pairs dense vectors with BM25, but almost no one can say where the BM25 number actually comes from, which is exactly why fusion, tuning, and exact-match failures stay mysterious. This guide opens the box: how an inverted index turns transcripts and OCR text into posting lists, the precise BM25 scoring formula with its term-frequency saturation and length normalization, what the k1 and b parameters really do, and why the tokenizer is the silent decider of whether an agent ever finds a serial number.

    BM25
    Lexical Search
    Inverted Index
    Sep 2026Read guide
    Search & Discovery
    13 min read

    Why Do My Search Results Get Worse as I Add More Data?

    Search that worked on ten thousand items and fails on ten million usually has one of five causes: a candidate budget that is a count rather than a fraction, near-duplicates filling the top ten, a coarse index trained on data that no longer looks like yours, a score floor calibrated on a smaller corpus, or a filter starving the candidate pool. This explains each mechanism, the single test that identifies it, and the order to run the tests in.

    Vector Search
    Recall
    Search Quality
    Sep 2026Read guide
    Data Infrastructure
    3 min read

    Marengo 3.0 on Amazon Bedrock: Architecture, Benchmarks and Vector Storage

    TwelveLabs Marengo Embed 3.0 is generally available as an embedding model in Amazon Bedrock Knowledge Bases. What the managed path gives you, what it takes away, and when to hold the vectors yourself. Includes the 1024 to 512 dimension change that breaks an in-place upgrade.

    TwelveLabs
    Marengo
    Amazon Bedrock
    Sep 2026Read guide
    Retrieval
    9 min read

    Why Does My Search Only Work in One Language?

    Search over your media library works in English, then a user types the same question in Japanese and gets nothing, or gets confident nonsense. Three separate causes produce that. Each leaves a different fingerprint in the results, and a better embedding model fixes only one of them.

    Multilingual
    Cross-Lingual
    Transcription
    Sep 2026Read guide

    All guides

    Every guide in one place, 141 in total.

    From concept to production

    These guides explain how multimodal perception and retrieval actually work. Mixpeek is the platform that runs them: point it at your storage and get back relevant, timestamped results.