Anhora
Pricing
English
  • English
  • Українська
Anhora

Create configurable, branded AI assistants powered by your products, documents and company knowledge.

  • Dashboard
  • API Status
  • Privacy
  • Terms
  • Contact

© 2026 Anhora

Site Chat guide

Configure an assistant that can answer and convert

A practical order for turning website content into grounded answers, useful product claims, and a restrained sales conversation.

Open Site Chat Connect a website

Setup order

01Connect website content02Review Product facts03Set voice and answer limits04Configure conversion behavior05Preview, save, and publish

1. Connect facts before writing behavior

Start with pages, FAQs, policies, and product documentation. Run Sync and check the committed item count before generating a draft. Website content is the evidence layer; the assistant configuration decides how that evidence is used.

Public product and service pages
FAQ, shipping, returns, and policies
Pricing only when it is current

2. Treat Product facts as publishable claims

Keep only claims supported by a source. Mark planned, limited, and unavailable items explicitly. Disable a claim when it should not be used in answers. Generated facts are a draft: review the status, summary, source, and enabled switch before saving.

Product facts

Enabled claims are available to the assistant

3 facts

WordPress integration

Source attached

AvailableEnabled

Shopify integration

Source attached

PlannedEnabled

Private beta access

Source attached

LimitedEnabled

3. Set voice after the evidence is stable

Use Auto language for multilingual sites. Choose the sales approach, then tune response length only when the default is too short or expensive. Auto language responds in the visitor's language; locale patterns help greetings and short conversational turns without tying the assistant to one market.

4. Add one next step, not a sales script

Decide when contacts are requested, add fit terms only when useful, and connect a webhook only if another system needs the lead payload. Start with the default consultative approach and change cooldowns only after reviewing real conversations.

Conversion behavior

Advance only qualified conversations

Sales approach

Consultative

Ask for contacts

After clear interest

Qualification

Optional fit signals

Fit roles

owner, support lead

Fit use cases

product questions, support

5. Test the boundaries, then publish

Test factual questions, planned integrations, a greeting, and an off-topic request. Save the reviewed draft before publishing the deployment key.

  1. 01Ask what the company offers.
  2. 02Ask whether a planned integration is already available.
  3. 03Ask for a price that is not in the sources.
  4. 04Say hello in another supported language.
  5. 05Show buying interest and check that only one next step appears.
Published Site Chat widget with welcome copy and suggested questions