Skip to main content

Introduction

VectorAmp in one sentence

Governed retrieval for teams that need answers they can trust.

VectorAmp turns documents, application data, and operational knowledge into cited AI answers and low-latency search APIs, backed by SABLE — our filter-aware vector database core built for large, governed datasets.

TL;DR

Use VectorAmp when basic vector search is not enough. The platform combines ingestion, embeddings, SABLE indexing, metadata filters, tenant boundaries, hybrid search, citations, APIs, SDKs, CLI workflows, and MCP access so product and operations teams can build AI search without stitching together a fragile retrieval stack.

What VectorAmp helps you build​

VectorAmp is built for knowledge-heavy products and internal workflows where retrieval has to be fast, scoped, explainable, and repeatable.

Search over company knowledge​

Connect source systems, index documents, and let users search across policies, product docs, customer notes, contracts, security material, and other approved knowledge.

Cited AI answers​

Return answers with source evidence so users can verify claims instead of trusting model output blindly.

Governed retrieval​

Scope every query by tenant, dataset, role, source, document type, timestamp, or custom metadata before results reach an app, user, or agent.

Retrieval APIs for products​

Use REST, SDKs, CLI, LangChain, or MCP to add semantic search and AI knowledge access to SaaS products, internal tools, and agent workflows.

Why teams choose VectorAmp​

Most vector databases can store embeddings. VectorAmp is opinionated around the hard parts that show up after the demo.

StrengthWhy it matters
SABLE vector coreDesigned for large-scale vector search with filter-aware retrieval instead of treating metadata as an afterthought.
Managed ingestionMove from files and source systems to searchable datasets without maintaining a separate parsing, chunking, embedding, and job orchestration stack.
Hybrid retrievalCombine semantic vectors with sparse/text signals and metadata filters when exact terms, SKUs, IDs, policies, or names matter.
Citations and source contextShow where every answer came from, making AI output easier to review, audit, and trust.
Multi-tenant platform modelOrganize datasets, API keys, usage, and access patterns around organizations and application boundaries.
Developer surfacesUse the dashboard for operations, REST for any backend, SDKs for app teams, CLI for automation, and MCP for AI agents.

The mental model​

VectorAmp has five moving pieces:

  1. Sources — files, cloud storage, web content, or application data you want to make searchable.
  2. Pipelines and jobs — ingestion workflows that parse, chunk, embed, and index source material.
  3. Datasets — the retrieval boundary your application queries. A dataset owns vectors, metadata, index configuration, and search behavior.
  4. SABLE retrieval — the search layer that combines vector similarity, filters, optional hybrid signals, and ranking.
  5. Intelligence — cited answers and agent-ready context built on top of retrieved evidence.

Common first use cases​

  • Internal knowledge search: let employees search policies, docs, meeting notes, and project knowledge with citations.
  • Support escalation assistant: retrieve product, customer, and incident context before an agent writes a response.
  • RFP and security questionnaires: draft answers from approved source material and preserve evidence links.
  • Tenant-scoped SaaS search: add semantic search to a product while enforcing organization and permission boundaries.
  • Agent knowledge access: expose approved retrieval through MCP so AI agents can answer using company context.
  • Compliance and policy Q&A: keep answers grounded in the current version of governed documents.

How to get started​

If you want the shortest path to value, follow this sequence:

  1. Create an API key.
  2. Run the quickstart and create your first dataset.
  3. Upload a small real document set, not toy text, so you can inspect citations and search quality.
  4. Add metadata filters that match how your product or team scopes knowledge.
  5. Choose your integration path: REST API, SDKs, CLI, LangChain, or MCP.