Skip to content

Book a 30-minute intro call

Loading the calendar…

Booking runs on Cal.com and loads only once you open this panel. See our privacy statement. Open the calendar in a new tab.

We build workflow automation, AI agents, energy forecasting, websites and custom software for teams that want the manual work gone.

louis2 is an AI automation studio in Rotterdam. We build automation, AI agents, custom software and websites, and we prove it with working projects.

Automation, AI and web

Our services

We provide all types of automation, website building and anything related to AI. Each area below is backed by something we have built. Open one to see the projects.

  • 01

    Workflow and process automation

    We remove the manual steps between the tools you already use: forms, spreadsheets, calendars, email, Telegram and CRMs. We build on n8n, Python, scheduled jobs and webhooks. Failed sends are retried and stay visible.

    • n8n and Python
    • Google Workspace
    • Telegram and email

    Projects: Hiring autopilot

  • 02

    AI agents and assistants

    Assistants that read documents and emails, score and classify what comes in, draft replies and answer questions. We run local models such as Ollama when data must stay in house, and cloud models when that fits. We say so when a simple rule is better.

    • Local or cloud models
    • Scoring
    • Drafting

    Projects: Hiring autopilot

  • 03

    Websites and web design

    Design, build and hosting on your own domain. The sites are fast, accessible and animated, with 3D visuals and live demos where they help explain the product. This website is the example.

    • Design system
    • 3D visuals
    • Your own domain

    Projects: This website

  • 05

    Forecasting and data

    We forecast energy consumption per meter from weather and calendar inputs, scored against a baseline. We also rank new companies to approach from public data, with a hand-off to your CRM or a CSV.

    • Energy forecasting
    • Public data
    • CRM hand-off

    Projects: Energy forecasting Optima Prospector

  • 06

    Document and email intake

    We pull names, dates, references and instructions out of emails and files into the format your system takes. A person reviews before anything is sent on.

    • Email intake
    • File parsing
    • Review step

    Projects: Logistics suite

Not sure what fits? Start with the task you repeat most.

Or write to us

How we work

From a messy task
to a working hand-over

  1. Discover the task

    We talk to the people who do the work and write down every step, tool, file and exception. The aim is one page that says what goes in and what has to come out.

    You get a one-page plan and a short list of real samples, such as files, emails or pages.

  2. Build the smallest version that does the core job

    One input, one output, nothing extra. You can try it early and say what is wrong while it is still cheap to change.

    You get a working first version, running on your own data or content.

  3. Test it on your real data and content

    We run it on your files, emails or pages, not ours, and count what it gets right and wrong. You see the results as they are, misses included.

    You get a short test report with the failures listed.

  4. Hand over, with a way back

    It goes live where you want it. For software that can mean an install script, nightly backups and a rollback if an upgrade goes wrong. For a website, your own domain and hosting.

    You get the finished build, installed where you want it, with backups and rollback where the work needs them.

Four rules we build by

  • Nothing wrong goes forward silently

    Every field is checked against what you expect. A mismatch is flagged for a person, and failed sends stay visible. Where AI is involved, it is a first filter and a person decides.

  • Fits the tools you already use

    We connect to what you have: Google Sheets and Calendar, a CRM, a CSV import or your own domain. No new system to learn.

  • Your data stays inside your network where possible

    The hiring autopilot uses a local AI model, so candidate data stays on your own machine.

  • Scored against a baseline, win or lose

    Every forecast is compared with a simple baseline on data it has not seen. We publish the result even when we lose, as we did on one meter in the energy pilot.

Under the hood

Technologies

  • Python
  • FastAPI
  • PostgreSQL
  • scikit-learn
  • n8n
  • Ollama
  • Google Workspace
  • HubSpot API
  • Docker
  • HTML, CSS and JS
  • Blender

Contact

Tell us what you want to automate, build or fix

Describe it in a few lines: what you do today and what it should do instead. It can be a manual job, an AI idea, a website or a tool. We reply with what we would build first.

Send us a message

What do you need?

How we handle your message is in our privacy statement.
Your message is delivered to our mailbox by Web3Forms. Nothing is stored on this site itself.