Results

Results. Not promises.

Real cases with verifiable metrics. No template filler. No made-up testimonials.

00

Dogfooding · our own site

skanndar.top

3 versions · same brand, another league

We practice what we sell: three generations of our own site, from a template CMS to a 3D experience. Every version is still live.

v1

Odoo Website

CMS del propio ERP. Rápido de montar, genérico.

Odoo 16 WebsiteQWebTema de plantilla
Visit
v2

Next.js

Primera reescritura headless: diseño propio y control total.

Next.jsReactTypeScript
Visit
v3live

Next.js + 3D

Experiencia scroll inmersiva con eclipse 3D. El front actual.

Next.js 15React Three FiberThree.js
Visit
01

AI architect and implementer

skanndar (in-house R&D)

In-house AI suite · in production

Challenge

Generative AI promises a lot and delivers little when it hallucinates or fails to cite sources. Before offering it to clients, we wanted to prove it actually works in a real business.

Solution

An in-house AI suite on Odoo, in production and dogfooded: we run it in our own business before selling it. It includes a multilingual chatbot that answers from the site's real content citing the source and captures leads; an internal expert advisor that handles the tax calendar and generates reminders; a copilot that administers the system by conversing, with confirmation before any deletion; and voice to talk to the assistant. All connected to the business's real data, with the option of self-hosted models to keep information under control.

02

Reference solution · live demo

AI Helpdesk for marketing units

Reference solution · live demoAI tickets · same-day response

Challenge

A property developer's marketing unit splits every request across 8-10 external providers (SEO, design, video, web…). The result: 5-8 days per asset, requests lost between handoffs, and zero traceability of who does what.

Solution

A single Odoo Helpdesk where every sales request becomes a ticket. AI classifies it (segment, asset type, priority), drafts and sends the first response the same day, and routes it to the right internal cell —Design, Video or Development—. Everything documented, measurable and without middlemen. This is a reference solution with a working, navigable demo, not a client testimonial.

03

CEO + Technical Architect

Cymit Química

Exit to PALEX Group · ×2 revenue

Challenge

Chemical company with 3 unintegrated legacy systems (EuroWin, Access, ad-hoc CRM) and a catalogue of over 2 million references with no automation.

Solution

Unified Odoo architecture on a digital platform, mass crawling to automate references, and a technical team scaled from 1 to 5 people (QA, Front, Backend).

04

Head of technical strategy

Rehabmedic

Top 3 Google in 60 days · zero downtime

Challenge

New business unit (medical equipment rental) with no web visibility and inherited infrastructure without high availability.

Solution

Odoo Multi-web with Schema.org architecture, live PostgreSQL replication, and observability with Telegram bots.

05

Architecture + custom development

Marta Solivellas

Custom coaching system on Odoo 19

Challenge

An entrepreneur launching her coaching business needed a solid digital foundation from day one —session and client management, online booking— without relying on scattered tools.

Solution

Odoo 19 with a custom-built coaching module: a portal where her clients book and reschedule, an integrated calendar with ICS feed, and PDFs with corporate branding. All on its own VPS infrastructure with Nginx HTTPS, automated backups and CI/CD on every deploy.

06

Leader and architect (Founder)

Kreathon

All-in-One innovation suite in production

Challenge

Organisations manage innovation with chaotic ideas and disconnected tools, with no clear flow from backlog to execution.

Solution

All-in-One suite that centralises the innovation backlog, strategic prioritisation and agile project management. Odoo 19 + Next.js 15 stack with a modular architecture built to scale.

07

ML architect + developer

Aplantida

In-browser AI · 171k images

Challenge

Computer vision model for plant identification, runnable in the browser without a dedicated server.

Solution

ML pipeline with PyTorch, 171,000 training images, knowledge distillation and TensorFlow.js deployment. Full-stack with React, Node.js/Express, MongoDB and CI/CD.

Have an Odoo project or an implementation that never quite works?