It has read all your documents, and tells you where.

On the website it answers customers, inside the company it answers colleagues. It uses only what you have written (pages, manuals, procedures, price lists) and puts the document it came from next to every answer.

Where
website, intranet, internal chat
Used by
customers, customer service, colleagues
First prototype
2–3 weeks

The problem

The same questions arrive every day: opening hours, terms, shipping, how to do a certain thing, where that form is. The answers already exist, written down somewhere, but whoever asks cannot find them and whoever answers rewrites them every time.

The chatbots of a few years ago left a bad memory: trees of hand-written answers that did not understand the question and could not say they did not know. The idea was right, but those chatbots did not read anything.

How the data flows

  1. Where it comes from: Question (customer or colleague); Website pages (indexed and kept current); Manuals and procedures (PDFs and shared documents).
  2. What happens, Assistant: Searches your content (hybrid search on the index); AI, Writes the answer (only from what it found); Adds the source (page or document).
  3. Where it ends up: Answer with its source (in chat, on the site or inside); A person (when there is no source).
  4. questions without a source show what to write.
Steps marked AI are the model's. The rest is ordinary automation or people.

How we solve it

04
01

It reads what you already have

Website pages, PDFs, shared documents, FAQs. There are no answers to write by hand, and the index updates when the sources change.

02

It answers only with what it finds

If a piece of information is not in your content, it says so and hands the conversation to a person instead of making something up.

03

The source next to every answer

Whoever reads can open the page and check. That is what makes an automatic answer credible, for a customer as much as for a colleague.

04

Two audiences, two indexes

What customers see and what colleagues see stay separate, with the permissions of the systems the documents come from.

What we measure

03

These are the numbers we set in discovery, before writing any code. The starting value is measured on your data, and the prototype is judged against it.

Questions closed without a person

The share the assistant resolves on its own, read together with the ratings of those who asked.

Unanswered questions

The ones for which it finds no source. They are the list of content missing from the website or the documents, and often the most useful discovery.

Repetitive requests to customer service

How many still arrive by email and phone on the topics the assistant covers.

When it is not worth it

If the questions are few and always the same, a well-written FAQ page does the same job. And if the content is out of date, the assistant confidently repeats old information: that needs fixing first.

We have already built it

Widgeto.ai is exactly this, ready-made for websites. It installs with one line of code, cites the page every answer comes from and speaks the visitor's language. For internal documents we build it to measure.

How Widgeto is built

Services involved

Other use cases

11 scenarios
06 · RAG and knowledge base

Technical support

Manuals, technical sheets and resolved tickets become answers with the source next to them, for the technician on site and for whoever answers the phone.

07 · MCP servers

ERP assistant

How the month went, which orders are stuck, what is missing from stock. Questions in plain language to the assistant you already use, with the answers taken from the ERP.

Tell us the process. We will tell you if it makes sense.

A free 30-minute call, with no commercial follow-up unless you want it.

Request a call

Companies that have already worked with us

  • Velux
  • ISU
  • MovePay
  • AB Group
  • Archiva Group
  • Onda SpA
  • PDE
  • Bynary System
  • COOPI
  • Gimasi
  • Creactives SpA
  • Rev AI
  • Odditt Inc.
  • Si Sociale
  • Stones
  • Turboden
  • Schindler