Skip to content

ragwise

The pip-installable Python RAG library built for 2026 — hybrid search, retrieval observability, temporal filtering, and agent tools on by default.

No Docker. No background server. No framework lock-in.

pip install ragwise

from ragwise import RAG, QueryConfig

async with RAG(llm="openai/gpt-4o-mini", reranker="flashrank") as rag:
    result = await rag.ingest("./docs/")
    print(result)  # IngestResult(chunks_created=42, skipped=0, failed_files=[])

    answer = await rag.query("What is the refund policy?")
    print(answer.text)
    print(answer.citations[0].text)    # passage text, not just filename
    print(answer.trace.retrieval_ms)   # 34 — always populated
    print(answer.trace.cost_usd)       # 0.00021

Hybrid BM25+dense retrieval runs automatically on every query. answer.trace is always populated — no setup needed.

Why ragwise?

Feature ragwise LangChain LlamaIndex RAGFlow
Lines to get started 4 40+ 20+ Docker setup
Hybrid search by default ✅ ❌ opt-in ✅ (Docker)
pip install, no server ✅ ✅ ✅ ❌
Async-first ✅ partial partial ❌
Streaming responses ✅ partial partial ❌
Retrieval trace (always-on) ✅ ❌ ❌ ❌
Passage-level citations ✅ ❌ partial ❌
Temporal filtering (as_of) ✅ ❌ ❌ ❌
Agent tool built-in ✅ ❌ ❌ ❌
Multi-tenant isolation ✅ ❌ ❌ ❌
Built-in eval ✅ ❌ partial ❌

Chunking accuracy: matches the FloTorch Feb 2026 benchmark winner at 69% end-to-end accuracy using RecursiveChunker.

Core Features

Hybrid search — on by default

BM25+dense retrieval fused with Reciprocal Rank Fusion. Catches exact keywords and semantic meaning on every query, no configuration needed.

Retrieval observability

answer = await rag.query("...")
print(answer.trace.retrieval_ms)    # 34
print(answer.trace.cost_usd)        # 0.00021
print(answer.citations[0].text)     # passage text
print(answer.citations[0].score)    # 0.91

Temporal filtering

# Query as of any date — no competitor has this
answer = await rag.query(
    "refund policy?",
    config=QueryConfig(as_of="2024-06-15"),
)

Semantic cache

async with RAG(cache=True, cache_threshold=0.92, ...) as rag:
    a = await rag.query("How do refunds work?")
    print(a.trace.cache_hit)  # True on similar queries — <10ms

Streaming

async for token in rag.stream_query("What changed in v2?"):
    print(token, end="", flush=True)

Agent tools (stateful)

from ragwise.agent import as_claude_tool_suite

tools = as_claude_tool_suite(rag, max_iterations=5)
# search_documents + get_document_context + check_context_budget

Multi-tenant isolation

await rag.ingest("./org_a/", tenant_id="org_a")
await rag.ingest("./org_b/", tenant_id="org_b")
answer = await rag.query("policy?", config=QueryConfig(tenant_id="org_a"))

Store upgrade path — no code changes

store="memory"                # dev / CI — zero setup, volatile
store="lance://./ragwise-index"  # persistent dev — no server
store="postgresql://..."      # production — pgvector

Next Steps