Chat · Voice · Video agents

Agents that know what your business knows

Volator builds knowledge bases from your websites, documents and media, then runs them as chat, voice and video agents. Every conversation goes back in as new material.

See how it works

Embedded on your own site

Ask our team
ChatVoice

Which sensor fits the 400 series?

The DS-4 fits every 400-series panel. Wiring differs on the 412, I can walk you through it.

transcript → knowledge base
  • EU · US · AU data residency
  • Chat, realtime voice, realtime video
  • Bring your own vector or graph database

One loop, four moves

Retrieval is the easy half. The last step is the one that usually goes missing. What an agent learns in a conversation becomes knowledge the next agent can use.

01

Ingest

Crawl entire sites with per-source crawler configs, upload PDFs and documents, write markdown in-app, or pull from connected integrations.

02

Structure

Chunking, embeddings and hybrid dense-plus-sparse search turn raw content into a knowledge base. Open any source, fix what came out wrong, extract it again.

03

Deploy

Push the same knowledge into your own vector or graph database, export it as markdown or llms.txt, or wire it straight into an agent.

04

Learn

Volator transcribes each conversation into a knowledge base you own. The next agent reads it.

transcripts return to the knowledge base

The platform

Support AgentCapabilities
Chat
Realtime voice
Realtime video

AI Studio

Chat, voice and video from one definition

Each agent gets its own capabilities: text, realtime voice, realtime avatar video. You write behaviour as a persona and a scenario instead of one long prompt you keep editing. Pick the model, attach the tools it may call, then put it on your site with one script tag and a domain allowlist.

  • Realtime voice and avatar video over WebRTC
  • Personas and scenarios instead of raw prompts
  • Tool calling and per-agent model selection
  • Embeddable widget with domain allowlisting
  • docs.yourcompany.comCrawled
  • Installation manualsExtracted
  • Training videosTranscribing
  • Agent transcriptsReturning

Knowledge bases

Sources in, structure out

Crawled websites, uploaded documents, markdown you write in the app and connected integrations land in one place, anti-bot pages included. Search runs dense and sparse vectors together, so exact terms and paraphrases both hit. Every source keeps its extraction history.

  • Site crawling with per-source crawler configs
  • PDFs, documents, images and video transcripts
  • Hybrid dense and sparse retrieval
  • Every source keeps its extraction history
  • Qdrant
  • Pinecone
  • pgvector
  • Neo4j
  • FalkorDB
  • OpenAI
  • llms.txt
  • S3

Destinations

Your stack, not ours

One knowledge base can feed several destinations at once. Write it to Qdrant, Pinecone, pgvector, Neo4j, FalkorDB or OpenAI vector stores. Export markdown, llms.txt, JSON or CSV to S3 or an endpoint you control. Set a schedule and each destination refreshes on its own.

  • Qdrant, Pinecone, pgvector, Neo4j, FalkorDB
  • Markdown, llms.txt, JSON and CSV exports
  • REST and GraphQL endpoints, S3 delivery
  • Scheduled refresh per destination

Workflows & operations

Pipelines you can see

Drag scraping, bulk extraction, PDF parsing, filters and loops onto a canvas, then run the result on a schedule or on demand. Every job reports its status while it runs. Usage logs show what it cost. The roles are owner, admin, editor and viewer.

  • Visual workflow builder with reusable nodes
  • Scheduled runs and change monitoring
  • Live job status and usage logs
  • Role-based team access and API keys

Your data stays where you put it

Pick a region for each business when you sign up. Volator stores and processes that data in the European Union, the United States or Australia, and never copies it between them. Ask us if you need another region.

  • European Union
  • United States
  • Australia

What people build with it

  1. 01

    A support agent that cites your docs

    Put a chat or voice agent on your site that answers from your documentation instead of guessing. Every transcript lands in a knowledge base you own, so you can read what customers actually keep asking.

  2. 02

    An internal expert on call

    Technicians, new hires and field staff ask in plain language and get the procedure back, taken from the manuals and work instructions you already have. Answers come in English or German.

  3. 03

    Question-and-answer over a document pile

    Papers, contracts, reports and archives become searchable, with equations, tables and figures intact. Point an agent at them, or point your own retrieval stack at them.

  4. 04

    Sources that never go stale

    Schedule recrawls and watch named pages for the changes you care about. The knowledge base updates itself instead of slowly going out of date.

See it running on your own content

Bring a website, a stack of PDFs or the questions your team answers all day. We'll run the demo on your material, not ours.