FAQs

About Vectara

What is Vectara?

Vectara is the trusted platform for enterprise conversational AI and RAG with extraordinary accuracy. As an end-to-end Retrieval Augmented Generation (RAG) service, deployed on-prem, in VPC, or utilized as a SaaS, Vectara delivers the shortest path to a correct answer/action while mitigating hallucinations and providing high-precision results. Vectara provides secure and granular access controls and comprehensive explainability, allowing companies to avoid risks and provide iron-clad data protection.

How does Vectara work?

Vectara is RAG-as-a-service, encapsulating the various components required for a scalable and high-performance RAG pipeline (document processing, multiple best-in-class embedding models, a retrieval engine, multiple rerankers to unlock different capabilities for your app, and an LLM) behind an easy-to-use developer API. Developers use Vectara to build RAG and semantic search applications by using the API to index their documents and respond to user queries with the full power of RAG, while Vectara works behind the scenes to execute the RAG ingest and query flows in a secure and scalable way while maintaining low latency and low TCO.

How can I benefit from Vectara?

You can benefit from Vectara if:

You are looking to build conversational AI based on your documents and data.
You are looking to build a centralized agentic RAG system to boost productivity and automate information delivery.
You need Guardian Agents to build out your always-on AI governance.

What are the use cases for Vectara?

Conversational AI
Agentic Retrieval Augmented Generation (RAG)
AI Agents
AI Governance

How is Vectara different from other Agentic solutions?

Vectara gives all types of builders an end-to-end platform for embedding powerful generative AI capabilities into your app or site without the need for data science and machine learning experience.

Some of Vectara’s unique differentiators include:

Accuracy: Vectara provides superior accuracy and recall based on your enterprise data. Complete with tools for explainability and scores for each generated response. Vectara also provides a factual consistency score (FCS) with every response and many ways to configure your queries and responses.
Security: Vectara provides enterprise role-based access control, customer-managed keys, and multiple deployment options, including on-prem and airgapped.
Explainability: Vectara provides full observability and provides citations with every response, allowing users to audit results and understand how answers are generated.

Is Vectara a vector database?

No, Vectara is not a vector database. It is a RAG-as-a-service platform that includes multiple components required for RAG, such as a document processing engine, chunking, a state-of-the-art embedding model, and its own internal vector database, which is used by its high-quality retrieval engine.

Is Vectara an embedding service?

No, Vectara is not an embedding service, although it does have state-of-the-art embedding models it uses within the platform. It is a RAG-as-a-service platform that includes multiple components required for RAG, such as a document processing engine, chunking, state-of-the-art embedding models, and its own internal vector database, which is used by its high-quality retrieval engine.

In which countries is Vectara available?

Vectara is a cloud-based GenAI platform running on AWS or GCP infrastructure in Vectara’s SaaS environment or alternatively in your own VPC or on-premise install. Vectara can be deployed in most regions in AWS or GCP. See the AWS Regional Services and GCP Regions/Zones pages for more details. Reach out to sales@vectara.com if you have specific requirements.

On-premises installs are currently available for the North America, APAC, and the UAE regions only.

How do I apply for a job at Vectara?

You can check all our available openings on our Careers page.

What is Retrieval Augmented Generation or RAG?

Retrieval Augmented Generation, otherwise known as RAG, is an approach to building GenAI applications that builds on semantic search (or retrieval) to provide answers to user questions by retrieving the most relevant facts and providing them to a generative LLM for summarization. This approach has several advantages:

It reduces GenAI costs by telling the LLM to only use relevant information when providing answers instead of providing much larger amounts of data.
It drastically reduces hallucinations, copyright issues, and keeps answers focused on the types of answers you’d want to provide in your business because it only uses the LLM for its knowledge of language: not for knowledge of how to answer end user questions. The answers to the questions come from the data you provide.
It increases GenAI security, since ACLs can be used for filtering data out that the user does not have access to before it ever gets to the LLM.
It provides explainability of GenAI responses by citing references to where it found the answers.
It keeps information up-to-date and eliminates costly, time-consuming, and privacy-concerning fine tuning based on your enterprise data. Information can be added and removed in seconds just as it would with a traditional keyword search application.

Software

How do I get up and running with Vectara?

All you need to do is sign up for a 30-day free trial with a company email address. You will then get access to the Vectara Console to get started with ingesting documents and testing the platform.

For more information on setting up Vectara, you can check out our Docs.

How long does it take to implement Vectara?

Setting up and implementing Vectara in production can be done on the same day. Index your first document and issue your first batch of queries in under 5 minutes.

Does Vectara's offering come with onboarding and training?

Yes, as part of all enterprise standard plans, users will receive onboarding and training within the scope of their first use case. Resources for training are delivered through online platforms and individual engineer training as part of production implementation.

What file types does Vectara support?

Vectara’s file upload API supports PDF, Microsoft Word, Microsoft Powerpoint, Open Office, HTML, JSON, XML, email in RFC822, text, RTF, ePUB, and Common Mark. Audio data (via a speech-to-text engine) and image data (via optical character recognition – or OCR) are available upon request by reaching out to support@vectara.com.

What languages does Vectara support?

Vectara supports over 100 languages and dialects. This support is integrated across the platform, including data ingest, the embedding models, retrieval and generation with the LLM.

Can I index from any data source?

You can index data from any supported file format, as well as raw text from data source systems, via Vectara APIs or the file upload feature within the Vectara Console.

Can I search across multiple indexes?

Yes, you can issue a single query or multiple queries in parallel to one or multiple indexes.

Deployment

What are the deployment options for Vectara?

Vectara is available as a fully managed cloud platform, maintained by the Vectara team, in a VPC install, or on-premises. Like other multi-tenant SaaS services, Vectara’s SaaS platform employs a release process designed to ensure features ship faster, the product can scale seamlessly when any client load increases and it leverages enterprise-grade security. This eliminates the responsibility for server maintenance, upgrades and capacity provisioning.

What resources do I need to deploy Vectara?

No additional or specialized engineering, hardware, or infrastructure resources are needed to successfully deploy Vectara if you use our SaaS service. Vectara was developed to make it easy for web and application developers to integrate generative AI in sites and applications without the need for additional training or resources.

For more information on setting up Vectara, you can check out our Docs.

Running in a VPC on self-managed/on-premise install requires resources to be allocated that depend on your specific usage and data requirements. If you need a VPC or on-premise install, reach out to sales@vectara.com to let us help size the right resources for your use case.

Does Vectara have any scalability limits?

Vectara can autoscale from small text volumes to millions of documents. Our SaaS platform automatically adds capacity as needed to handle higher query volumes.

I'm having trouble setting up Vectara. What should I do?

If you are having any trouble setting up Vectara or need any help with implementation, you can send a message on our Vectara Discuss Community or contact Vectara support directly at support@vectara.com.

You can also visit https://support.vectara.com.

What are Vectara’s SLAs?

Vectara offers SLAs in support, availability, and performance.

For a full list of available Vectara SLAs, please review our Pricing page and order form documents. If you have more aggressive SLA needs, we often can meet them: reach out to sales@vectara.com to start the conversation.

Pricing

What are the different options for buying Vectara?

There are three Vectara Enterprise subscription plans to choose from: Small, Medium, and Large. Vectara also offers VPC and on-premise options. Customers can also add a forward-deployed engineer or subscribe to platinum support services.

You can visit Vectara’s pricing page to view the different features for each plan and determine the best one for your needs.

What are credits?

Credit is a usage unit representing the value and cost for using a specific API, data storage, and compute for processing different kinds of data in the RAG pipeline (e.g. table, image etc). Credits is the usage currency of Vectara’s platform features.

Is there a free plan?

Vectara offers a 30-day free trial, complete with nearly all of the enterprise features of the platform. There is no free-forever program.

What is Vectara’s pricing model?

Vectara's pricing model is usage-based and based on the number of search queries processed and the account data size. These consumption metrics can be scaled by purchasing additional credit. Visit our pricing page for more details.

How do you count queries in your pricing model?

Any queries that are issued to indexed content – via the Vectara console or the Vectara API – are counted towards the query count and applicable to the credits.

What is the definition of account size in the pricing model?

Account size is the sum of text and metadata size (measured in MB) within all corpora in the customer account, before any replication factor is applied.

Which payment methods and currencies are accepted by Vectara?

Vectara accepts payments made through contracted agreements or on the AWS Marketplace to use AWS credits. Vectara supports billing in United States Dollar (USD).

Is there any commitment once I start paying for Vectara?

Each plan has its own minimum commitment. For details, see our pricing page.

What happens if I exceed my committed plan usage?

Once you have exceeded the minimum commitment, you will automatically be billed for any additional bundles you have consumed at the end of the month.

How do I switch from one plan to another?

Some upgrade functions can be performed directly within the Vectara Console. For any other requests, please reach out directly to sales@vectara.com.

How can I change my account details and billing information?

Account details and billing information can be accessed and changed from the Billing tab within the Vectara Console.

Security and privacy

How does Vectara handle and process sensitive personal and customer information?

Vectara supports full client control over how data is preserved, including the support of a full deletion of customer data via API. Clients decide what data they transmit to the service and what data remains within the service, with the exception of billing and billing contact data.

SOC II Compliance

Vectara’s platform is independently validated through annual SOC 2 Type II audits and HIPAA compliance, ensuring our data protection and security practices meet the highest standards. Customer data is always encrypted in transit and at rest, with backup and disaster recovery programs tested regularly. We continuously monitor for vulnerabilities, conduct penetration tests, and follow strict SLAs for remediation. These controls give our customers confidence that their data remains protected, resilient, and available.

Does Vectara implement any encryption standards for information processing?

Vectara ensures sensitive data is encrypted and the keys are managed both logically and physically by the appropriate teams.

Encryption at rest is AES 256-bit symmetric key encryption. Separate keys are managed per corpus, and access to keys is through an account master key managed on FIPS 140-2 compliant hardware. Vectara also provides the option of a customer managed account master key. Encryption in transit is TLS 1.3.

Does Vectara have a security program?

Vectara has a documented security program that is audited periodically for major security program objectives, status of security program non conformities, and risk logs. The system architecture was designed to enable ready compliance with SOC 2 and ISO 27001 standards. Vectara systems are SOC 2 and HIPAA compliant. Vectara’s SOC 2 Type 2 and HIPAA Evaluation Report are available in the Vectara Trust Center. If you would like to report a security concern, contact security@vectara.com.

Does Vectara have a privacy policy?

Vectara has a documented privacy policy that is reviewed periodically. Vectara’s privacy policy covers responsibilities under GDPR regulations. CCPA regulations are not applicable to Vectara. Vectara’s privacy policy can be found here.

Does Vectara have a disaster recovery plan?

Vectara has policies and procedures for disaster recovery and backup. Vectara has established, documented, implemented, and maintained processes, procedures, and controls to ensure the required level of continuity for information security during an adverse situation.

Information processing facilities, infrastructure, and application architecture are implemented with redundancy sufficient to meet and support high availability requirements.