The problem
Support breaks when knowledge stays scattered.
Answers live across product pages, documentation, PDFs, policies, and internal notes. Basic chatbots fail when the right context is missing.
AI support infrastructure
Your websitealready knowsthe answer.Avento makesit respond.Your website alreadyknows the answer.Avento makes it respond.
Index your website and documents, deploy grounded assistance, uncover knowledge gaps, and add voice support without rebuilding your stack.
Knowledge intelligence
Answer pipeline
Customer question
“How do I cancel my subscription?”
The Avento system
One knowledge layer for every customer conversation.
Avento connects sources, retrieves context, records conversations, surfaces missing knowledge, and deploys the same intelligence across chat and voice.
Grounded retrieval
Observable conversations
Continuous improvement
From knowledge to support
Build a support system in four connected steps.
Connect knowledge01
Bring your sources together.
Index website pages and upload supported documents so every answer starts with the right context.
WebsiteDocumentationPDFPolicy
Configure behaviour02
Shape how Avento responds.
Set assistant identity, response guidance, allowed origins, fallback behaviour, and customer-facing branding.
IdentityGuidanceDomainsBranding
Deploy anywhere03
Add support without a rebuild.
Install the existing widget script in Next.js, React, HTML, Framer, or WordPress and keep project-level control.
Widget SDKProject keyAllowed origins
Learn and improve04
Turn conversations into insight.
Review analytics and unanswered questions, then improve the knowledge your assistant can retrieve.
ConversationsAnalyticsKnowledge gaps
The Avento control layer
The complete support system, not only the chat bubble.
Sources, conversations, analytics, deployment controls, and voice outcomes stay connected.
Knowledge
A source of truth your assistant can retrieve.
Track indexed pages and documents, manage source status, and refresh knowledge when your content changes.
Representative preview using product-supported capabilities.
Project knowledge
4 sources, 200 pages and chunks available
Documentation
128 pagesIndexedProduct website
42 pagesIndexedBilling policy.pdf
18 chunksIndexedTerms of service
12 chunksIndexedKnowledge-gap intelligence
See what Avento could not answer before customers ask again.
Avento tracks unanswered and low-confidence questions so your team can see which source content needs more coverage.
Unanswered clusters
Illustrative examples
Low source coverage18 similar questions
How do annual plan upgrades work?
Affected source
Billing documentation
Recommended action
Add upgrade policy and proration examples
“Integration is exceptionally clean. Having chat and phone support on a single platform keeps the workflow focused.”
“We love how it brings our website and custom documentation into one support workflow.”
“The voice agent feels natural, and the support experience remains easy to review.”
Deployment
Install once. Control it from Avento.
Use a public project key, allow the deployed origin, and add the verified SDK to your existing stack.
Public project key
The key identifies the project. It is not a private backend credential.
Origin enforcement
Requests are accepted only from origins configured for the project.
Next.js Integration
- 1.Add the public project key to .env.local.
- 2.Add the Avento script to the root layout.
- 3.Restart the development server and deploy.
- 4.Add the deployed domain to allowed origins.
.env.local
NEXT_PUBLIC_AVENTO_KEY=YOUR_PUBLIC_PROJECT_KEY
app/layout.tsx
import Script from "next/script";
<Script
src="https://avento-ai.vercel.app/sdk.js"
data-key={process.env.NEXT_PUBLIC_AVENTO_KEY}
strategy="afterInteractive"
/>Add your deployed domain to the project's allowed origins before testing the widget.
Voice AI · Callback agent
Your chatbot can pick up the phone.
Add callback support beside the embedded assistant. Avento uses project knowledge, captures the outcome, and keeps the conversation visible to your team.
Shared project knowledge
Structured call outcomes
Project-level language configuration
Customer callback
Project support line
Live response
Call in progress“I found the relevant billing policy. I can walk you through the next steps.”
Matched sources
4
Language
English
Outcome
Summary ready
Callback completed and outcome stored with the project
Built beyond the chat bubble
Support today. Agent workflows tomorrow.
Avento organises sources, retrieval, conversations, deployment controls, and observable outcomes into a reusable support layer.
Structured knowledge sources
Web pages and supported documents organised by project.
Grounded retrieval
Relevant source chunks selected before an answer is generated.
Conversation observability
Customer questions and answers remain available for review.
Knowledge feedback
Unanswered patterns guide the next source improvement.
Widget and SDK surfaces
One knowledge layer delivered into supported website stacks.
Voice-ready support
Callback experiences can use the same project context.
Deployment controls
Project keys and allowed origins control where the widget runs.
Future workflow foundation
Reusable support context for more advanced automation over time.
Make your knowledge respond
Build support on the knowledge you already have.
Frequently asked
Clear answers before you build.
Avento can crawl website pages and accept supported document uploads. Your project knowledge is chunked, indexed, and used to retrieve context for customer questions.