ENFR

AI Chatbot

A chatbot that answers from your documents. Ask the one on this page.

It reads your policies, product sheets and guides, answers your customers or your team from them, shows its source, and says when the answer is not there. Tested on your real questions before it goes live.

Book a free 30-minute call
Bernard, the chatbot of this site, open on a page with its three starter questions.

Example, at CA$90 an hourEssential $2,700–$4,500Connected $4,500–$7,200

Example bands are the first two of three versions, before the 15 % safety margin. Your exact price comes from a free evaluation.

How pricing works What your budget gets you See it running

How it works4 steps

From your documents to a chatbot that answers

Four steps. The price is fixed at step two, before you accept anything.

Document

01

Bring your documents

Your policies, product sheets and frequent questions, as they are. No specification needed.

Invoice

02

Assessment and fixed-price quote

Feasibility and complexity are assessed before any price is given. The scope, the timeline and the AI usage fees are written in the quote.

Checked box

03

Tests on your real questions

We test it on the questions your customers or your team really ask, before it goes live.

Message

04

Go live

A chat window on your website or in your internal tool.

An answer the chatbot wrote from an uploaded PDF. Each paragraph ends with the file name and the source number.

01 · An answer

With its source

Each paragraph ends with the file name and the source number.

The chatbot's knowledge base window: one PDF uploaded and selected, ready to be searched.

02 · The documents

Loaded, selected

Policies, product sheets, guides: as they are.

The model menu: Claude, Command-A and Gemini models, each with what it can read and its cost level.

03 · The model

Chosen together

Claude, Gemini or another, per task; it can change later.

It answers from your documents

You load your documents: policies, product sheets, guides. The chatbot searches what they contain before it answers.

In this screenshot, the answer comes from an uploaded PDF. Each paragraph ends with the file name and the source number.

When the answer is not in your documents, it is set up to say so.

An answer the chatbot wrote from an uploaded PDF. Each paragraph ends with the file name and the source number.
Fig.An answer the chatbot wrote from an uploaded PDF. Each paragraph ends with the file name and the source number.
Table4 rows

A public assistant or your own?

Your ownPublic assistant
Answers fromYour documents, policies and product sheetsWhat it learned elsewhere
When it does not knowIt is set up to say soIt may answer anyway
Tested onYour real questionsGeneral questions
Where it livesOn your website or in your internal toolOn the provider's website

What is in the first version

The scope is written in the quote, before the first day of work.

  • Your documents prepared for search
  • The chatbot, on the model we choose together
  • A chat window on your website or in your internal tool
  • Tests on your real questions before it goes live
  • A usage dashboard: requests, tokens and cost per model
The model menu: Claude, Command-A and Gemini models, each with what it can read and its cost level.
Fig.The model menu: Claude, Command-A and Gemini models, each with what it can read and its cost level.

Not included: the AI provider's usage fees (written in the quote), writing your documents, and legal, medical or financial answers.

Capabilities3 of 3

What already runs on this system

Examples from the chatbot I have already built. Yours follows your needs.

01

Processor

The model we choose together

Claude, Gemini or another: the model is chosen for the task, and it can change from one message to the next.

02

Line chart

Usage you can see

One dashboard shows the requests, the tokens and the cost per model, in one place.

03

Padlock

Your accounts, your data

Each user signs in, and their data stays protected in the database. The accounts are in your name from day one.

The console I work in uses the same idea: an agent that answers from our internal knowledge base.

— Luke Sidney, clevrstart.

Questions5 answered

Questions before a first call

What people ask before a first call.

01Does it make up answers?
It is set up to answer from your documents and to say when it cannot find the answer. We test it on your real questions before it goes live.
02Who pays for the AI usage?
The AI provider's fees are added to the quote price. They are written in the quote before you accept it.
03Can it answer in French and English?
Yes. The one on this page does: it answers in the language of the page it is opened on.
04Who owns what is delivered?
The accounts are in your name from day one, and the code is transferred to you on the final payment.
05What does it cost?
The number is on the software cost calculator, from your budget. The terms are on the How pricing works page: a fixed price for the agreed scope, half to start, half on delivery. The call and the quote cost nothing.

More on the topic