Skip to Content

AI automation services

AI automation that turns your inbox into finished work

Invoices, order messages, meeting notes, questions about your numbers. We build AI automation that reads them and does the next step in the tools you already use, from your ERP and CRM to email, spreadsheets and WhatsApp, with a person approving what matters. Send a sample through the gate and watch.

Your inbox

In your systems

Vendor BillDraft
Vendor
Sample Packaging Co.
Bill date
12 Sep 2026
Due
12 Oct 2026
ProductQtyAmount
Corrugated box 40×302001,640.00
Packing tape 48 mm60288.00
Delivery145.00
  • Vendor matched by VAT number
  • Lines read from the invoice table
  • You review and post it

Sample documents and data, to show how it works. We build flows like these into the systems you already run.

The difference

What AI automation means for your business

Your software already automates the simple things. The rules and workflows in your ERP, CRM or helpdesk follow what you set up: when a deal is won, create a task; every night, send the payment reminders. They are reliable, and they stop where the rule stops.

AI automation starts where the rules run out. It reads the things a rule can’t: a scanned invoice, a message typed in a hurry, the notes from a meeting, a question in plain English. It works out which customer, product or project they belong to, and prepares the record or the reply. The judgement calls stay with your team.

  1. Built into your software Rules Workflows and rules: when this happens, do that.
  2. What we build AI Reads, understands and drafts: bills, quotations, tasks and answers.
  3. Always in the loop People Approve, correct and post. Every action is logged.

What we build

AI automation services we deliver

Six kinds of work, built into the tools you already run. Most projects start with the one that hurts most and grow from there.

  1. AI workflow automation

    Multi-step work that moves on its own: a document or message arrives, gets read, lands in the right system and the right person is told. Built with tools like n8n and Make, or with custom code where they fall short.

    n8n, Make or custom code
  2. ERP and app integrations

    AI inside the systems your team already works in, such as Odoo and other ERPs, CRMs, helpdesks and spreadsheets: drafts, summaries and alerts appear right where the work happens.

    Odoo, CRMs and your apps
  3. Document processing and OCR

    Invoices, bills, receipts, forms and scans turned into clean records. For Odoo we built our own app: OCR reads the document, AI fills in the record and finds the partner by name or VAT number.

    Includes our AI Invoice Digitization app
  4. AI chatbots and assistants

    Assistants for customers and staff that answer from your own documents, products and policies, on your website, in WhatsApp or in your team’s tools, and hand over to a person when they should.

    Website, WhatsApp and your tools
  5. Predictive analytics and forecasting

    Demand, stock and cash-flow forecasts built from your own history, shown in plain dashboards your team can question, so problems show up before they cost you.

    Forecasts and dashboards
  6. Custom GenAI apps

    A process no off-the-shelf tool fits? We design and build the app, the prompts and the integration around it, using the AI model that suits the job.

    Built for your process

Running Odoo? See our Odoo development and Odoo integration services too.

Works with your stack

We meet your tools where they are

Every automation has three parts: the apps where your work already lives, the workflow that connects them, and an AI model that reads and writes. We work with all three.

Where your work lives

  • OdooERP
  • Google WorkspaceGmail, Sheets, Drive
  • Microsoft 365Outlook, Excel, Teams
  • WhatsAppCustomer chats
  • TwilioSMS and voice

What connects it

  • n8nWorkflows
  • MakeWorkflows
  • PythonCustom code

The AI models

  • OpenAIGPT models
  • ClaudeAnthropic models

Logos belong to their owners and show what we work with, not a partnership.

Our own AI product

Connect Claude, ChatGPT and Gemini to Odoo with MCP

For businesses on Odoo, we built our own bridge between AI assistants and the ERP. The Model Context Protocol (MCP) is the open standard AI assistants use to work with other software, and our Advanced MCP Server puts an MCP endpoint inside Odoo itself, with no extra server to run. Your team keeps the AI assistant they already like, and it can read, analyse and act on live Odoo data.

45
tools to search, report, chart, create, update, message and schedule
9
chart types, from bar and line to funnel, treemap and sunburst
16 to 19
Odoo versions, Community or Enterprise, on Odoo.sh or your own server

Works with Claude, ChatGPT, Gemini, Cursor, Claude Code and any other MCP client.

Setup guides: connect Claude to Odoo and connect ChatGPT to Odoo.

Chart this quarter’s sales by team.

  1. ontology_search“sales team”
  2. read_groupsale.order by team
  3. generate_chartbar

Here are this quarter’s confirmed sales by team.

  1. Direct42k
  2. Partners31k
  3. Online18k
Sample data. Tool names are the MCP Server’s own.

Security

Security, access and human approval

Giving an AI access to your business systems should feel as controlled as giving access to a new colleague. Every automation we build follows the same path.

  1. Scoped accessEach automation can reach only the data and actions its job needs.
  2. Your data stays yoursRecords stay in your own systems. An AI model sees only what a task sends it.
  3. Human approvalThe actions you choose, such as sending or posting, wait for a person.
  4. A record of every runWhat came in, what the AI decided and what changed, logged so you can check it.
  5. You can switch it offEvery automation can be paused or rolled back without touching the rest.

On Odoo, our MCP Server adds OAuth 2.1 sign-in, allow-lists of tools and models, rate and IP limits, an approval queue and a full audit log.

By team

AI agents for sales, support, finance and operations

What your people could ask, and what the agent does in your systems. You decide which actions need approval.

Sales and CRM

  • Which opportunities haven’t moved in two weeks?
  • Draft a follow-up for each of them.
  • Chart this quarter’s pipeline by stage.

Searches the CRM, drafts the emails for you to send and builds the chart.

Customer support

  • Summarise this ticket and its history.
  • Find the answer in our product manual.
  • Draft a reply I can edit.

Reads the ticket and its messages, and answers from your documents with the source.

Finance

  • Turn these supplier PDFs into draft bills.
  • Which customers are more than 30 days overdue?
  • Prepare reminder emails for the top ten.

Digitizes the documents, queries receivables and stages the reminders for approval.

Operations

  • What’s running low in stock this week?
  • Turn today’s meeting notes into tasks.
  • Show late deliveries by carrier.

Reads inventory and deliveries, and creates tasks with owners and dates.

How we work

How an AI automation project runs

  1. Audit

    We look at where your team’s hours go and which tasks are worth automating first.

  2. Blueprint

    A short plan: what to automate, how it fits your systems, and a fixed quote and timeline.

  3. Build and integrate

    We build it into the tools you already use, with approvals where you want them.

  4. Measure and tune

    We check the results with you and adjust the automation as your work changes.

Choosing an approach

Built-in AI features, no-code workflows or a custom build: which fits?

There is more than one way to bring AI into your business. Each is right for a different kind of job.

AI features in your software No-code workflows Custom AI build
Best forQuick wins inside one tool you already useConnecting several apps with clear, repeatable stepsWork that spans your systems or needs your own rules and data
Built withWhat the vendor shipsPlatforms such as n8n and MakeCode, prompts and the AI model that suits the job
FlexibilityLimited to the vendor’s roadmapGood, until a step needs real judgementShaped around your process
ControlThe vendor’s settingsThe platform’s settings, in its cloud or on your serverAccess, approvals and logs designed in

Many businesses use all three. We help you pick per task, and build whichever fits.

We use it ourselves

The AI we run on our own business

Before we suggest AI for your team, we use it in ours.

  1. Meeting notes become action items

    After a call, the notes become action items, decisions and open questions, each matched to the related project task.

  2. Ask AI answers from our docs

    The assistant on our documentation site answers questions from our own guides.

    Try it on our docs (opens in a new tab)
  3. App pages written and published by AI

    Our app pages are generated by AI from each app’s own description and screenshots, then go live automatically, with no manual step in between.

    See our app pages

Sample notes, questions and pages, to show what goes in and what comes out.

Straight answers

Questions about AI automation

Ask us anything else on a short call.

What is AI automation?

It is software that handles the work a fixed rule can’t describe: reading a scanned invoice, understanding a message or answering a question in plain language, then doing the next step in your systems. Ordinary workflow rules still handle the simple, predictable parts.

Do we have to be on Odoo?

No. We know Odoo inside out, but we automate across email, spreadsheets, CRMs, helpdesks, WhatsApp and the other tools your business runs on.

Which tools and AI models do you work with?

Among others: OpenAI and Claude models, Google Workspace, Microsoft 365, WhatsApp, Twilio, n8n, Make, Python and Odoo. We pick what suits each job and what you already use.

Can ChatGPT or Claude work with our business data?

Yes, through a secure connection to your systems. For Odoo we built the Advanced MCP Server, which connects Claude, ChatGPT, Gemini, Cursor and other MCP clients to Odoo 16, 17, 18 and 19. Our guides show the setup for Claude and ChatGPT.

Is it safe to give AI access to our systems?

Each automation gets only the access its job needs, the actions you choose wait for a person, and every run is logged. On Odoo, our MCP Server adds OAuth sign-in, allow-lists of tools and models, rate and IP limits and an audit log.

What happens if the AI gets something wrong?

Wherever a mistake would matter, the automation prepares a draft or waits for approval instead of acting alone, and the log shows what ran. That way errors are caught before they reach a customer or your books.

Which processes should we automate first?

Work that is frequent, repetitive and full of text: supplier invoices, order messages, follow-up emails, routine questions and reports someone builds by hand. We look at your volumes in the audit and start where the payback is clearest.

Who owns what you build, and what happens to our data?

You do. The code, prompts and data stay in your own systems. When a task uses an AI model, only the text that task needs goes to the provider, such as OpenAI or Anthropic, under the account you choose.

How is a project priced, and how long does it take?

We scope it with you on a short call, then agree a fixed quote and timeline before any work starts.

Tell us the task you hate. We’ll show you how an agent would do it.

Bring one real example: a document, a message or a report you build by hand every week. We’ll walk you through how AI automation would handle it in your business.