Lattice
Skip to content

Tell me what to use

Build my AI stack.

Describe your case — or start from a real one below. The stack updates as you answer, every answer becomes part of a shareable link, and the decision copies out as a report. Drawn from the 112 tools already in the index; estimates are heuristic bands, not vendor quotes.

Start from a real case

01 — What are you building?
02 — Your case

How much does retrieval depend on filtering?

How often does the corpus change?

How fast must an answer come back?

03 — Constraints
04 — Priority

Your stack — updates as you answer

RAG application · 0 req/mo

  1. RoutingWorkable
    Bifrost

    Why for you: A high-throughput OpenAI-compatible proxy with a small footprint.

    Watch out: You need the breadth of a full routing stack.

    Consider Cloudflare AI Gateway when edge caching and latency dominate, on Cloudflare already.

  2. RetrievalStrong fit
    Elasticsearch

    Why for you: One index carrying BM25 and dense vectors, with ranking you can tune.

    Watch out: You need a permissive licence — the Elastic licence is not OSI-approved.

    Consider pgvector when under a few million vectors, or when retrieval joins rows you already have.

  3. InferenceWorkable
    KoboldCpp

    Why for you: Local generation with a UI tuned for interactive, long-form use.

    Watch out: You need headless serving at scale.

    Consider llama.cpp when the constraint is hardware, not throughput — no GPU, or an edge box.

  4. PromptsStrong fit
    Agenta

    Why for you: Open-source prompt management with a playground and versioning.

    Watch out: You need eval depth rather than prompt ergonomics.

    Consider DSPy when compiling prompts into optimised programs from labelled examples.

  5. EvalsWorkable
    Arize Phoenix

    Why for you: OpenTelemetry-native evaluation you can run yourself.

    Watch out: You want a vendor support contract behind it.

    Consider Braintrust when making evaluation, rather than tracing, the primary workflow.

Estimated cost

$0–$65/mo

Driven by Elasticsearch — the usage-billed picks. Heuristic band, not a quote.

Confidence

87%

Lower when constraints narrow the field.

Biggest risk for your case

Retrieval quality decides the outcome — most failures are ranking or chunking failures, not model failures. Build the eval set before tuning anything.

Every why and watch-out comes from the tool's own use-when / skip-when. Read the method →