Ground AI in enterprise data
Build RAG pipelines that combine live, structured, and unstructured data with SQL and hybrid search. Retrieve, rank, and feed real-time context directly to your models.
Do more with your data
0x
up to 100x faster queries
0%
up to 80% cost savings on data lakehouse spend
0x
Increase in data reliability
RAG breaks down without unified, real-time data
Fragmented RAG pipelines force developers to manage multiple search engines, connectors, and model APIs. Models are grounded in incomplete or outdated data, leading to hallucinations, inconsistencies, and production risks.
Build data-grounded AI faster
Deliver accurate, context-rich results by combining SQL federation, hybrid search, and LLM inference in one governed runtime.
Federate structured data
Query operational and analytical data across databases, object stores, and APIs using standard SQL. All in real time with zero ETL.
Learn more about Spice federation
Search and embed unstructured data
Create embeddings from text, documents, or logs using local or hosted models. Run vector similarity search directly inside Spice to find relevant context, semantic matches, and related insights instantly.
Explore hybrid SQL search
Retrieve and augment intelligently
Blend structured SQL results and semantic search results in one query using hybrid search. Deliver precise, context-rich data to your LLMs without manual data engineering or pipeline maintenance.
Learn about real-time hybrid search using RRF
Generate insights with the AI Gateway
Invoke and run models like OpenAI, Anthropic, or local LLMs directly in SQL queries. Feed retrieved context into the model and generate accurate, compliant, and contextual results within the same SQL workflow.
Explore LLM inference and AI model serving
Why choose Spice for RAG
Spice unifies data retrieval, semantic search, and AI generation in a single, high-performance runtime-no pipelines, no orchestration, no drift.
Real-Time Federation
Query all your sources with federated SQL. No data movement or batch syncs.
Built-in Vector Search
Embed and retrieve semantic context alongside SQL filters and full-text search.
SQL LLM Inference
Call, prompt, and evaluate models in SQL via the AI() SQL function
Hybrid Ranking
Blend multiple result sets with Reciprocal Rank Fusion for per-query weighting and tunable relevance.
Governed & Secure
Enterprise-grade access control ensures compliance and auditability.
Deployment Flexibility
Run Spice anywhere: as a sidecar, microservice, cluster, or on the managed Spice Cloud Platform.
Trusted by teams building intelligent applications
Run data-intensive workloads on a high-performance engine trusted by teams building real-time systems at scale.
“Partnering with Spice AI has transformed how NRC Health delivers AI-driven insights. By unifying siloed data across systems, we accelerated AI feature development, reducing time-to-market from months to weeks - and sometimes days. With predictable costs and faster innovation, Spice isn't just solving some of our data and AI challenges - it's helping us redefine personalized healthcare.”
Tim Ottersburg
VP of Technology, NRC Health
“Spice AI grounds AI in our actual data, using SQL queries across many data sources. This brings accuracy to probabilistic AI systems, which are very prone to hallucinations.”
Rachel Wong
CTO, Basis Set
Trusted by global enterprises
Build a scalable RAG app
Guides and examples to learn more about building RAG applications with Spice.
Hybrid Search Docs
Spice provides robust search capabilities enabling developers to query datasets beyond traditional SQL, including semantic (vector-based) search, full-text keyword search, and hybrid search methods.
True Hybrid Search: Vector, Full-Text, and SQL in One Runtime
TL;DR Show Me the (Data)! It’s well established (and maybe even trite) to say that enterprises are going all-in on artificial intelligence, with more than $40 billion directed toward generative AI projects in recent years. The initial results have been underwhelming. A recent study from the Massachusetts Institute of Technology's NANDA initiative concluded that despite the enormous allocation of […]
Real-Time Hybrid Search Using RRF: A Hands-On Guide with Spice
Surfacing relevant answers to searches across datasets has historically meant navigating significant tradeoffs. Keyword (or lexical) search is fast, cheap, and commoditized, but limited by the constraints of exact matching. Vector (or semantic) search captures nuance and intent, but can be slower, harder to debug, and expensive to run at scale. Combining both usually entails standing up multiple engines […]
FAQs
Answers to common questions about building RAG pipelines with Spice
How does retrieval-augmented generation reduce hallucinations?
RAG reduces hallucinations by grounding model responses in data retrieved at query time instead of relying only on training data. Models grounded in incomplete or outdated data are more likely to hallucinate, so retrieval quality and freshness matter as much as the model itself. Spice retrieves live, governed data with SQL and hybrid search so the context passed to the model stays accurate.
Do I need a separate vector database for RAG with Spice?
No. Spice creates embeddings using local or hosted models and runs vector similarity search directly inside the runtime. You can blend semantic results with SQL filters and full-text search using hybrid vector and full-text search, so no separate vector database or synchronization pipeline is needed.
Can RAG pipelines use structured data as context?
Yes. Spice federates queries across databases, object stores, and APIs with standard SQL, so structured records can be retrieved and combined with semantic search results in one query. This grounds model responses in operational data without manual data engineering.
Which models can I use for generation?
Spice invokes models from providers like OpenAI and Anthropic, or local LLMs, through its AI Gateway. Models are called directly in SQL, so retrieved context feeds into generation within the same query workflow. See LLM inference and AI model serving for how model serving works in Spice.
How is retrieved context ranked before generation?
Spice blends multiple result sets with Reciprocal Rank Fusion (RRF), with per-query weighting and tunable relevance. This ranks the most relevant structured and semantic matches highest, so the model receives precise, context-rich input.
See Spice in action
Walk through your use case with an engineer and see how Spice handles federation, acceleration, and AI integration for production workloads.
Talk to an engineer