Knowledge base
Your information in one place, the part you approve
Address, prices, opening hours and common questions: we collect them from your website, your documents and answers you add by hand, and you review them. Answering calls from this knowledge automatically is not switched on yet and is planned. For now our team prepares the assistant from your materials.
Trial: 7 days and 60 minutes, with no card. It starts at your first test call from your phone

A wrong answer is worse than no answer
Your information lives in ten places
Prices are on the website, hours on Facebook, terms in a Word file. The assistant needs one correct place to read from.
AI can make things up
A confident but wrong price or opening time leaves a poor impression of your business. On anything unknown the assistant must say it does not know.
Information changes
A price changed, a branch opened. If updating takes a lot of work, the knowledge goes stale fast.
How the knowledge gets filled
Step 1
Add a source
Paste your website address, upload a document or type a question and answer by hand.
Step 2
Review and confirm
iCall prepares a draft: services and prices, hours, contacts, common questions. Anything it is unsure about is marked "needs review". You confirm before anything goes live.
Step 3
Fill the gaps
You see the questions the assistant could not answer in the workspace. You write the answer and confirm it.
Three sources, one knowledge base
All sources sit in one list. You see what the assistant knows and where it came from.
Website import
Enter your address and iCall reads your public pages (up to 10): services, prices, hours, contacts and Q&A. Whatever the page does not say stays empty rather than guessed.
Documents
PDF, Word and text files: a price list, terms, a service description. The assistant answers from their content.
Missed questions
Questions the assistant could not answer appear in one list. The answer enters the knowledge base when you confirm it.
Fill a missed question in the workspace
You see what the caller asked and what the assistant could not answer. You write the right answer and confirm it.
- You see the question the assistant could not answer
- You write the answer
- Nothing changes until you confirm
How far you can trust an answer
Knowledge is not a black box. You see what is published.
Published knowledge
Only what you confirmed goes into the knowledge. Using it automatically on calls is not switched on yet and is planned.
Honest about the unknown
If it has no information, the assistant says so and saves the request for your team.
Call → Action → Evidence
One call, from a question to an added answer. The example uses a fictional clinic.
- Call
A caller asks for a price
How much does teeth whitening cost?
Inbound call, the price is not in the knowledge base yet
- Action
The assistant does not guess, you correct it
- Found no price in the knowledge and did not invent one
- Told the caller we will confirm the exact price by phone
- You wrote the price in the workspace and confirmed it
- The confirmed price is in the knowledge; using it on calls automatically is planned
- Evidence
What the workspace shows
- Topic
- Teeth whitening
- New answer
- 350 GEL per visit
- Source
- Added by hand, 19 Oct, 14:05
- State
- Confirmed and published
ExampleDemo data: a fictional clinic and price, not a real tariff.
Questions about knowledge
Let's gather your knowledge
Send us your website address or a price list. We review the first draft together.
Trial: 7 days and 60 minutes, with no card. It starts at your first test call from your phone
More modules
iCall is one platform: phone lines, knowledge, actions and evidence work together
- Phone lines and SIPYour number or any carrier's SIP trunk
- IntegrationsTelegram for your team; Google Sheets and SMS are being prepared
- Call evidence and qualityTranscripts, action receipts and review
- Campaigns and remindersOutbound calls with consent; not switched on yet
- SurveysCSAT after a visit; NPS planned, not switched on yet
- When a person is neededA message to your team; direct transfer is planned
- Security and trustData location, access and AI disclosure
- Why iCallMarket, product, moat, business model and roadmap