What file types can I upload?

More than 45, including PDFs (even scanned ones, read via OCR), Word, PowerPoint and Excel documents, images, audio recordings (transcribed), email files, ebooks, and text and data formats like Markdown, CSV, JSON, and HTML.

Which AI model does it use?

Your choice. Knowledge Hub supports OpenAI, Anthropic Claude, and Google Gemini models. Keep the default that auto-picks a model, set a specific model per assistant, or switch models on an individual conversation.

Can I add the chatbot to my own website?

Yes. Create a public key, choose which collections it can read, set the title, colour, and quick questions, add your allowed domains, and paste one line of script. It appears as an inline panel or a floating chat bubble.

Who can see my documents and assistants?

You control visibility on every item: private, team, or shared with specific roles and members. Data is strictly scoped to your workspace and never mixed with any other tenant, and external callers only see content you have marked visible.

Where are my document vectors stored?

By default in the platform's built-in pgvector store. On an entitled plan you can point an individual Knowledge Base at your own Qdrant index, and each Knowledge Base keeps a separate index.

All apps

AILive now

Knowledge HubAI answers grounded in your knowledge

Upload your documents and get AI answers grounded in them, with a citation for every claim. Build reusable agents and embed a chat widget anywhere.

Start freeView pricing

Grounded answers with page-level citations
Ingests 45+ file types: PDF, Office, images (OCR), audio
Reusable AI Agents you invoke from chat or workflows
Embeddable, white-label KB chat widget

File types

AI providers

Chat widget

Everything it does

Explore every feature

Tap a category to see what is inside. Nothing hidden, nothing bloated.

Knowledge Base and ingestion

Turn any pile of documents into a searchable, cited knowledge source. Upload almost anything and it parses, chunks, embeds, and indexes automatically.

45+ file formatsPDFs, Word, PowerPoint, Excel, images, audio, email, ebooks, and a wide range of text and data files, all handled by one pipeline.
OCR for scanned pages and imagesScanned PDFs and image files are read with a vision model, so text trapped in pictures becomes searchable knowledge.
Audio transcriptionDrop in recordings (MP3, WAV, M4A, and more) and the spoken content is transcribed and indexed alongside your documents.
Collections, folders, and tagsGroup knowledge into named collections so you control exactly what each assistant or widget can read.
Live processing status and retriesEvery document moves through parse, chunk, extract, embed, and index with visible status, bulk and folder uploads, and one-click retry on failures.
Per-collection controlsEach Knowledge Base keeps its own index, so you can set its embedding model, vector store, OCR and transcription toggles, and max file size independently.

How it works

Live in a few steps

Build your Knowledge Base

Upload documents (single, bulk, or whole folders) into named collections. Knowledge Hub parses, OCRs or transcribes, chunks, embeds, and indexes each one.

Create an assistant

Pick a template or start fresh, write its instructions and persona, pin the collections it should rely on, and choose its AI model and tools.

Chat and get cited answers

Ask in natural language. The assistant retrieves the most relevant passages, answers grounded in them, cites each claim, and renders tables, charts, or exports.

Publish the widget or API

Create a public key, choose which collections it can read, set branding and allowed domains, and paste one line of script, or call the API server-to-server.

Control access and iterate

Set private, team, or shared visibility, review answer quality and feedback, retry any failed documents, and add knowledge over time.

Where it shines

Made for real work

Website support and pre-sales chat

Embed the white-label widget on your site, scope it to your product docs and FAQ collections, and let visitors get accurate, cited answers around the clock.

Internal knowledge assistant

Upload policies, handbooks, contracts, and decks, then give your team a private assistant that answers from them with links back to the exact page.

Research and document analysis

Ask questions across a whole library of PDFs, spreadsheets, and reports at once, get tables and charts comparing every document, and export as PDF or CSV.

Operational Q&A over your data

Build an assistant scoped to your invoicing, payments, and approvals data so staff can ask questions in plain language and get grounded numbers.

For developers

Embed it, or drive it from code

Embed snippet

<script src="https://app.softsolz.uk/softsolz.js"
  data-public-key="pk_live_..."
  data-service-id="smart-chat"
  data-mode="floating"></script>

Embed surfaces

Public widgetOne-line JS snippet

API scopes

service.smart-chat.messages.sendservice.smart-chat.threads.manageservice.smart-chat.kb.viewservice.smart-chat.kb.manageservice.smart-chat.agents.manage

Webhook events

smart-chat.message.receivedsmart-chat.kb.document.uploadedsmart-chat.kb.document.readysmart-chat.kb.document.failedsmart-chat.agent.invoked

Product guide Developer reference

Questions

Frequently asked

No. In Knowledge mode the assistant retrieves from your uploaded documents first and grounds every answer in them, citing the source file and page. When nothing relevant is found, it tells you and offers to widen the search rather than inventing content.

Better together

Pairs well with

AI WorkflowsTrigger Knowledge Hub agents as scheduled steps.

InvoicingScope an assistant to your AR data.

BlogsDraft posts grounded in your Knowledge Base.

One subscription covers every service.

Ship Knowledge Hub on your site today

Create a free workspace, install Knowledge Hub, and go live in minutes. 14-day trial on paid plans.

Start free View pricing

Knowledge HubAI answers grounded in your knowledge

Upload your documents and get AI answers grounded in them, with a citation for every claim. Build reusable agents and embed a chat widget anywhere.

Grounded answers with page-level citations

Ingests 45+ file types: PDF, Office, images (OCR), audio

Reusable AI Agents you invoke from chat or workflows

Embeddable, white-label KB chat widget