Intranow—Private Q&A over your content

Intranow lets you build chatbots that answer questions using your own material—think docs, sites, PDFs, wikis, or Slack archives. It can learn from your human engagement and tickets, and also use general AI knowledge to provide context-aware responses.

Start with How Intranow Works, then move on to Concepts and Bots.
When you’re ready, connect your knowledge sources and, if needed, embed the chat widget on your site.

How Intranow Works

Intranow follows a retrieval-augmented generation (RAG) pattern: a user asks a question; Intranow searches your indexed content to grab the most relevant snippets; those snippets are passed to the language model to craft an answer. Only the selected pieces are shared with the model—your entire corpus isn’t exposed at once.

Tip: Phrase questions specifically (e.g., “Which plans support Slack integration?”) rather than broad prompts like “Summarize everything.”

Where Intranow shines

Customer support for product docs
Internal Q&A for employees
Research across papers or technical notes
Developer help grounded in documentation

Less suitable

Whole-corpus summarization
General open-ended questions not tied to your data
Spreadsheet analysis or calculations

Be specific: specific keywords improve retrieval.
Keep instructions separate: add formatting/voice directions as bot instructions rather than cramming them into the question.

Conceptual Overview

In Intranow, everything revolves around two building blocks: bots (the chat interfaces people talk to) and knowledge sources (the collections of content the bots can draw from). You can mix and match many bots with many sources without duplicating data.

Example

Imagine a company with two products. It could have public bots for website visitors and private bots for support teams. Each bot selects the appropriate sources (public docs, internal wikis, etc.) so that public bots never see private material.

Knowledge Sources

In Intranow, a knowledge source is any connected dataset your bots can reference when answering questions. The platform currently supports the following types:

File uploads — common docs like Word and PDF.
Websites — add individual URLs or point to a sitemap.xml to crawl.
Confluence — connect your wiki space.
ZIP archives — upload compressed collections of files.
GitHub — index content from repositories.
Slack — include messages from a workspace.

Need something not listed? email our team to request another source type.

The Intranow Chat Widget

The chat widget lets you add a “Chat with us” button to any website. First create a bot in Intranow and set it to public. :contentReference[oaicite:1]{index=1}

Install

1) Load the widget script in your site’s <head>:

  
  
  <script type="module" src="https://unpkg.com/intranow-chat@0.2.7/dist/scriv-chat/scriv-chat.esm.js"></script>

2) Place the custom element where you want the button:

  

  <Intranow-chat visible="false" team="YOUR_TEAM" bot="YOUR_BOT" button-text="Let's Chat"></Intranow-chat>

Replace YOUR_TEAM and YOUR_BOT with the IDs from your Intranow account.

Customize

You can restyle colors and positioning via CSS (including CSS variables), and change the button label with button-text. Here’s an example that uses static positioning and custom colors:

<style>
  Intranow-chat {
    /* Use fixed/bottom-right for a floating button, or static for in-flow placement */
    position: static; /* default floating example shown in docs is fixed at bottom-right */

    /* Color tweaks */
    --button-background-color: #448c74;
    --button-background-color-hover: #2f6f5b;
    --button-text-color: #ffffff;
    --button-text-color-hover: #ffffff;
    --button-border-color: rgba(0,0,0,.12);
    --button-border-color-hover: rgba(0,0,0,.22);
  }
</style>

Z-index

If the widget ends up behind other elements, raise the --Intranow-z-index variable (the default is 50):

<style>
  Intranow-chat { --Intranow-z-index: 1000; }
</style>

Bots

A bot is the user-facing chat endpoint. Each bot references one or more knowledge sources and can be configured with behavior and presentation options.

Setting	Type	Description
Is Default?	Yes/No	The project’s fallback bot (used in Slack or when no bot is specified).
Is Public?	Yes/No	Public bots have shareable URLs; private bots require login.
LLM	Single Choice	Choose the model (e.g., GPT-3.5 for speed/cost, GPT-4 for higher quality).
Knowledge Sources	Multi-select	Which sources the bot can consult while answering.
Show Source Citations	Yes/No	Display links to the passages used in the response.
Expand Lookups	Number	How many neighboring chunks from a source to include. Higher values can improve answers but cost more and may hit limits if set too high.
Additional Instructions	Text	Style or behavior hints (e.g., “Answer in Markdown,” “Adopt a friendly tone”).