Summarized answers to your users’ questions are generated by our platform by first looking across all your data before responding.

Provide the best answers to your users’ questions through search that is optimized to your data with a custom blend of large language model (LLM) based retrieval (i.e. "Semantic Search") and boolean exact match.

Neural model for document classification, retrieval, and ranking.

Extract a snippet and highlight the relevant information from any retrieved document.

Extract file metadata such as document access control, source URI, document author, and other attributes.

Get direct access to the ranking scores assigned by the platform.

Neural model that re-ranks the top ranked documents for a given query.

Index documents in different file formats.

Index and search in multiple languages.

Search an index of a given language using queries in a different language.

Allow searches to be prioritized by customer-defined measures of relevance.

Submitted text data is not retained on Vectara for utmost data privacy.

Search across multiple indices or corpora at the same time.

Access search and events analytics information.

Improve fault tolerance, latency, and throughput of the service.

Configure backup retention to meet business needs.

Protect data while stored and when moved between two services.

Customer ownership of the encryption keys that protect their data.

Get support from user peers in a Vectara community forum.

Get support by contacting us directly via email or on our website.

Designated hours to answer any question or help with any issue.

Service level agreement defining the level of quality and availability of Vectara’s services.

Count of eligible creator(s) of support requests.

We make every reasonable effort to respond promptly. For production system outages, the target is within 1 hr.

The maximum number of results that will be returned for each query.

A corpus is a client-defined collection of data with a unique ID created via Vectara API or the Vectara Administrative Console.

Allocated number of unique users within a customer account.

Allocated number of users in a given account that can control user access and define account configurations.

The number of results that will be returned by a re-ranker request.

Number of queries that can be issued per month for the account.

Number of generative requests for summarizations that can be issued per month for the account.

Sum of size (measured in MB) of all corpora within the customer account after indexing, before any replication factor is applied.

Aggregate number of MB of data ingested via any supported Vectara ingest API.