Vakav Autonomous product tour
Your organization's knowledge assistant, on your own servers
It reads your procedures, process diagrams and technical documents.
In this tour
From document to sourced answer: workspaces, access levels, custom bots and flows, all inside your organization. About 4 min.
- 01 · KNOWLEDGE AI
Your organization's knowledge assistant, on your own servers
It reads your procedures, process diagrams and technical documents. Everything runs on your own servers, with a large language model hosted in-house. People ask in plain language, much as they would in ChatGPT. Each answer comes with its exact source: the document and the page.
- 02 · HOW IT ANSWERS
From document to answer, with sources
PDF, Word and Excel documents and audio files are added to a workspace. Their content is indexed into your organization's knowledge base. For each question, the most relevant passages are retrieved. This method is called RAG, retrieval-augmented generation. The language model answers from those passages and shows which document each part comes from.
- 03 · EXAMPLE · CHAT
Ask, and get the answer with its source
A user opens the DR unit's workspace. and asks what changed in the new version of the safety procedure. The answer is drawn from that procedure. The file name and page of each source appear with the answer.
- 04 · WORKSPACES · ACCESS
A workspace for each unit, access for each user
Each unit has a workspace, with its own documents and settings. Each workspace's documents are kept apart from the others. Each user's access to workspaces follows your org chart. Each user gets answers only from the workspaces they can access.
- 05 · BOTS · FLOWS
Custom bots and flows
Each workspace has a bot you call with @ in the chat. Ready-made skills, such as answering from documents or querying a database, give the bot its abilities. In the flow builder you build a custom bot block by block; each block is one step of the work. In the chat, the bot uses the skill or flow that fits the request.
- 06 · OFFLINE BY DESIGN
Your data never leaves
The language model and the assistant are installed on your own server. Documents are read from inside the same network. No data crosses your network boundary, and no internet connection is needed. Answers reach only users inside your organization.
- 07 · ENGAGEMENT
From assessment to operation
Needs assessment: your documents, users and needs, and one measurable goal. Design and setup: workspaces, access levels and bots that match your organization's structure. Installation and testing: deployment on your server and testing with real documents. Training and support: training your users and continuously improving the assistant.