Pillar — Knows
Knowledge base
The knowledge base is your reference docs as markdown documents that an AI canvas fills with real data straight from your conversations: frequent topics, metrics, and characteristic snippets from calls and correspondence. Those documents then plug into automations as AI context — so knowledge from your conversations feeds back into handling new messages. Any AI edit can be reverted, so filling the base out is safe.
Key facts
- What it is
- Reference docs — markdown documents that serve as the knowledge base for AI processing of conversations.
- AI canvas
- An editor with a chat assistant that fills the document with real data from your calls and correspondence.
- Channel-neutral
- Works with any conversation — calls and correspondence alike — in one loop.
- Context for AI
- Documents plug into automations and are used by AI as reference context during processing.
- Reversible
- Any edit by the AI assistant can be reverted in one action — filling the base out is safe.
Reference docs as a knowledge base
Your team's knowledge lives in structured markdown documents instead of scattered notes and people's heads.
- Documents describe products, policies, frequent topics and the right answers.
- One place the AI draws reference context from when processing conversations.
- A clear markdown format: anyone can edit and versions are easy to read.
The AI canvas fills the doc from your data
An assistant next to the editor pulls facts from your conversations and writes them into the document — you don't start from a blank page.
- 'Fill the frequent-topics section from my data' — the assistant gathers them from conversations.
- Real metrics, topics and characteristic snippets from calls and correspondence flow into the doc.
- The chat with the assistant happens right in the canvas — edits are visible instantly.
Documents as automation context
The knowledge base isn't dead weight — it plugs into your processing rules and makes the AI more accurate.
- Attach a reference doc to AI automation actions — tagging, extraction, text generation.
- The AI answers and scores messages with your policies and terminology in mind.
- This way knowledge from past conversations works on handling new ones — the flywheel closes.
Edits are reversible
Any change the AI assistant makes can be reverted — so you can fill the base out with confidence.
- Every assistant edit is visible and can be undone in one action (Revert).
- Experiment with the contents without the risk of spoiling the document.
- You stay in control: the AI proposes, you decide.
Product
Knowledge base
- 1
Create a document
Set up a reference doc in markdown — a policy, a product description, a guide to message topics.
- 2
Fill it from your data
Ask the AI canvas to gather facts from your conversations: frequent topics, metrics, characteristic snippets.
- 3
Plug it into automations
Attach the document to AI actions as context — tagging, data extraction, text generation.
- 4
Improve it as data grows
As conversations accumulate, extend the base — and reversible edits make that safe.
FAQ
Frequently asked questions
What are reference docs and why do I need them?
They're markdown documents that serve as a knowledge base for AI processing of conversations: they hold your policies, product descriptions, frequent topics and the right answers. The AI uses them as context to score and process new messages more accurately.
How does the AI canvas fill a document?
A chat assistant works next to the editor and pulls facts straight from your conversations. Ask it to 'fill the frequent-topics section from my data' and it gathers real metrics, topics and characteristic snippets from calls and correspondence into the document text.
Is this for calls only or for correspondence too?
For any conversation. The knowledge base is channel-neutral: it's built the same way from calls and from correspondence — chats, emails, tickets — in one loop.
How does the knowledge base affect handling of new messages?
Documents plug into automations as AI context: when tagging, extracting data or generating text, the AI takes your policies and terminology from the reference doc into account. This way knowledge from past conversations works on handling new ones.
What if the AI writes something wrong?
Any edit by the AI assistant can be reverted in one action. So filling the knowledge base from your data is safe: you always see what the AI proposed and decide whether to keep it or roll it back.
Related capabilities
Ready to start?
Turn every conversation into data, knowledge and action
Start by analyzing your conversations — no risk, no bots required. The platform turns your archive into data and a knowledge base, and voice agents plug in when you're ready.
Free plan, no card required.