Artificial intelligence

AI agents and process automation for businesses

The AI that does not merely answer. It acts.

Inside Web builds AI agents that carry out concrete tasks inside your tools: sorting incoming enquiries, preparing quotations in Odoo, chasing up, enriching product records, producing reports. Each agent works with precise permissions, keeps a record of what it does, and leaves you to approve what matters.

Assistant, automation, agent: the difference.

What it doesExample
AI assistantAnswers a question from your content“What is the delivery time to Ghent?”
Classic automationChains fixed actions, triggered by a ruleNew order → confirmation email
AI agentUnderstands a situation, decides the steps and acts in your toolsReads a price enquiry, identifies the products, prepares the quotation in Odoo and submits it for approval

Classic automation excels where everything is predictable. The AI agent takes over where you have to read, understand and choose: an email written freely, an attachment, an incomplete enquiry. The two work together very well.

Automate what is repetitive. Keep people for what is important.

What an agent can take on.

  • Sort incoming enquiries

    Read emails and form submissions, identify the type of enquiry, the urgency and the right person, then create the task or the ticket.

  • Prepare quotations

    Extract the products and quantities from an enquiry, check prices and stock in Odoo, and prepare a quotation ready for approval.

  • Chase up at the right moment

    Quotations with no reply, overdue invoices, incomplete files: personalised reminders, sent out or submitted for approval.

  • Enrich product records

    Write descriptions, attributes and translations from supplier sheets, in your brand's own voice.

  • Produce reports

    A weekly summary of sales, enquiries or incidents, written and sent to the right people.

  • Process documents

    Extract the useful information from purchase orders, supplier invoices or forms, and record it in the right place.

How we build an agent.

  1. 01

    Choose the right process

    We target a task that is frequent, time-consuming and well defined. We measure how long it takes today: that is the yardstick for success.

  2. 02

    Map the steps and the exceptions

    Who does what, with which tools, and what happens when an enquiry falls outside the ordinary.

  3. 03

    Connect the tools

    Odoo, mailbox, CRM, shop, document storage: the agent receives only the access its task requires.

  4. 04

    Set the approval points

    What the agent does alone, what it prepares for approval, and what it passes straight to a person.

  5. 05

    Test, measure, extend

    A pilot phase on real cases, a comparison against the starting yardstick, and then extension to other tasks if the gain is real.

Under control, by design.

  • Minimal permissions

    The agent reaches only the data and actions it needs. It cannot delete what it has no right to touch.

  • Human approval

    Sensitive actions, such as sending something to a client or making a financial change, go through approval.

  • A complete log

    Every decision and every action is recorded: you always know what was done, when and why.

  • GDPR and the AI Act

    Personal data kept to what is necessary, processing documented, transparency towards the people concerned.

With an eye on what comes next.

AI agents are moving fast, and the standards for connecting them to business tools are becoming widespread. By building today on clear processes, controlled access and clean data in Odoo, you prepare a company where each new agent is added in days rather than months. And your people spend their time on what AI will never do for them: the relationship, the judgement, the decisions.

Your questions.

(AI agents and automation)

What is an AI agent?

An AI agent is a program that understands a situation, decides the steps to follow and acts in your tools: creating a task, preparing a quotation, updating a customer record. Unlike a simple assistant, it does not merely answer, it carries out the work.

How does an AI agent differ from classic automation?

Classic automation chains fixed actions according to rules set out in advance. An AI agent can handle variable situations: an email written freely, an attachment, an incomplete enquiry. The two approaches often sit side by side in one process.

Which tasks can an SME hand to an AI agent?

Sorting incoming enquiries, preparing quotations, chasing quotations or invoices, enriching product records, extracting information from documents and writing regular reports are among the most rewarding cases.

Can an AI agent work inside Odoo?

Yes. Inside Web connects agents to Odoo to read and create quotations, contacts, tasks or invoices, with rights limited to what the agent has to do. Sensitive actions go through approval.

Can the agent make mistakes?

Yes, like any person or any piece of software. That is why we set human approval points for important actions, limit permissions and record every action. A pilot phase on real cases comes before any wider roll-out.

Do we stay in control of what the agent does?

Yes. You decide what it does alone, what it prepares for approval and what it passes to a person. A detailed log lets you review every decision and every action at any time.

Is our data safe with an AI agent?

The agent reaches only the data its task requires, through dedicated access. We favour model providers that do not use data sent through their API for training, options for European hosting and, where necessary, self-hosted models.

Do we have to change our tools to use AI agents?

No. Agents connect to the tools you already have, such as Odoo, your mailbox, your CRM or your online shop, as long as they offer a way to exchange data such as an API. The agent adapts to your organisation, not the other way round.

Where should we start?

With a process that is frequent, time-consuming and well defined, and whose current cost in time can be measured. A first agent on that ground proves its value quickly before you automate anything else.

How long does a first agent take to set up?

A pilot agent on one well-defined task usually takes a few weeks. The timing depends above all on how many tools have to be connected and on how clear the starting process is.

Will AI agents replace our employees?

They replace tasks, not people. The time freed from re-keying, sorting and chasing goes back into client relationships, advice and decisions, where people make the difference.

How do we measure the return on an AI agent?

By comparing the time spent on the task before and after, along with the volume handled, the response time and the error rate. Those indicators are set at the outset so you can decide objectively whether to extend the automation.

Let's talk about your project.

A first conversation with no strings attached. One of the team gets back to you within 24 working hours.