Digital self-service

The compelling alternative to a chatbot, Viko works out of the box without custom development or scenario training.

Benefits

Live answers to any question

Backed by your existing information sources

Better than a chatbot

Laser like targeting

Viko treats each of your visitors as an individual, providing direct answers to their specific questions.

Reduced costs

Viko can handle a wide range of sales or customer service questions, significantly reducing inbound queries.

Preserve your investment

Unlike chatbots, Viko doesn't require coding or training. Viko finds answers hidden deep within your information sources.

Automated question answering for websites

Features

Unlocking your data

Natural language text mining finds hidden answers

Surfacing information

Unlike pre-defined chatbot dialogs, Viko answers questions on the fly, surfacing information from your existing data sources.

Automatic FAQs

Don't wait for someone to ask the question. Viko can automatically generate page specific SEO compliant FAQs.

Insights & visibility

What are your users really interested in? Viko removes the guesswork by identifying and clustering topics of interest.

Self learning

The true power of artificial intelligence is the ability to learn and improve based on feedback. That's exatly what Viko does.

Screenshots

Dialog widget

Simple popup style dialog box. Best suited to simple flows

How it works

Data ingestion

Data is ingested via API calls or by our web scraper. It's then processed, optimised and stored in a specialised full text database.

Document ingestion

Documents are ingested by our web crawler or through the API and placed in a redundant, persistent queue

Parsing & pre-processing

The documents are split into smaller chunks, then pre-processed by our natural language processing models

Indexing

The chunks of text are stored in an optimised, tree like structure for fast retrieval at query time

Querying

Questions are parsed by our natural languages models to understand the true semantic meaning. They are then passed down through a chain of handlers.

Prior answer search

First we check if the question is semantically similar to a prior question/answer pair

Answer generation

If a prior answer cannot be found we extract or generate an answer, using the ingested documents for context

Forwarding

Finally if no suitable answer can be generated, or the answer is rejected by the user, we forward it to an agent