News
We are always taking measures.
What we improve, fix and deliver, month by month.
-
55% less time on our internal test run
The full automated tests of our internal system went from 61 to 27 minutes in our internal measurement of 22 October 2026. That is waiting time our team gets back on every review.
-
We prepared our servers against PHP vulnerabilities
We keep PHP (the language many of our apps run on) and the parts they rely on up to date, and test them before publishing to close known flaws in time.
-
27% fewer fixed instructions for the AI
We cut the fixed instructions our AI reads at the start of each work session by 27% (internal measurement, 17 October to 10 November 2026). Less fixed reading means less usage when each task starts.
-
We hardened our setup against WordPress vulnerabilities
We tightened the settings of the WordPress sites we manage to reduce the entry points attackers use most.
-
We handle first the errors that affect users
We prioritise what most affects the users of each live project.
-
Security patches for the parts our apps use
When a security notice appears for a component we use, we update it and publish the fix.
-
597 versions and counting
Our own system has published 597 versions since March 2026, according to its change log. Because we built it ourselves, we improve it without depending on anyone.
-
Security reviews of our AI agents
We run regular security reviews on what our AI agents are allowed to do, and more and more of our working rules are checked automatically instead of by hand.
-
A page to book appointments that sends every request to the team
We delivered an appointment-booking page for a clinic: every request goes straight into the team's contact system, with no copying by hand.
-
89% less waiting on one pre-save check
In an app of more than 5,000 files, one of the automatic checks that run before each change is saved went from 77 seconds to between 6 and 8, in our internal test of 21 October 2026.
-
We watch the security notices for our apps
A routine check compares the parts of each app against official security notices and flags which ones need updating.
-
We update without slowing your project
Improvements to our system are applied without disrupting ongoing work.
-
Tests already passed are not repeated
In an audit of 192 full test runs, up to 19 (about 10%) repeated code nobody had changed. Since 17 October 2026 a version that already passed is not tested again, which saves time and AI usage.
-
An online community for gamers, in Spanish and English
We built an app where a gaming community can come together, with everything available in two languages.
-
We test every project in conditions close to production
Before publishing, apps are tested with versions similar to the ones they will run on live. Fewer surprises on delivery day.
-
Before-and-after simulators for dental and beauty clinics
We delivered simulators that show an illustrative image of how a treatment could look, to support the conversation between patient and specialist.
-
An AI chat that answers only with confirmed prices
The assistant answers with confirmed information and delivers contacts by email.
-
An online store for a beauty brand
We built a store with catalogue and cart for a beauty brand, ready for its launch.
-
A service that creates videos and clips with AI
We delivered a service that assembles short videos and clips automatically from existing content, ready for social media.
-
A 3D space game you play in the browser
We built a spaceship game with waves of enemies that runs in any browser, with nothing to install.
-
75% less waiting in a pre-save check
In a test with 536 files, one of the checks before saving changes went from 64 to 16 seconds (6 August 2026). Less waiting means more useful working time on every project.
-
Centralised monitoring of live projects
We see in one place how each live project is running, so we can step in when needed.
-
A website that attracts patients for dental clinics
We delivered a website designed for dental clinics to attract new patients, in Spanish and English.
-
Automated tests on access controls
We check with automated tests who can see each piece of data, so every user reaches only their own.
-
Daily backup of every WordPress site
Every WordPress site we manage saves a daily copy of its database, and we check that the copy can be read.
-
61 versions in one month
In July 2026 we published 61 versions of our own system. Small, steady improvements instead of big leaps.
-
We check that no keys are left inside saved code
Every change goes through an automatic check that looks for forgotten passwords or keys before it is saved.
-
An AI agent that looks after SEO and publishes to WordPress
We delivered an agent that analyses a site's search ranking and publishes improvements directly to WordPress.
-
Its own design on every screen
In June 2026 we made it mandatory for every screen we deliver to have its own design, not a generic template.
-
We delivered a 2D role-playing game that runs in the browser
A 2D role-playing game with combat and character progress, ready to play from any browser with nothing to download.
-
A follow-up chat between doctors and their patients
We delivered a platform for doctors to stay in touch with patients between visits, with organised conversations.
-
A mini app for emotional support
We built a simple web app that keeps people company through difficult moments. It does not replace professional care.
-
A bot that monitors real-time data
We delivered a bot that follows market data in real time and automates repetitive tasks.
-
45 versions of our system in April
April 2026 added 45 versions of the framework we built in-house to work with AI in an orderly way.
-
We automated invoice reading with AI
We delivered a system that watches a folder, reads each invoice with AI and builds the spreadsheet reports with no manual entry.
-
We reinforced the security of connections between systems
We reviewed how our tools talk to each other so they validate what they receive and keep a record of what happens.
-
Visual documentation for each app
We document each app visually to understand it at a glance and plan changes with care.
-
Our own way of working with AI
We built our own framework for working with AI: 74 versions in March 2026 alone. Thanks to it we can do everything you read about on this page.
-
We delivered an AI chat for websites
An assistant that answers a website's visitors at any hour, ready for several companies to use, each with its own information.
-
A bot for live streams with giveaways and polls
We delivered a bot that lives in a stream's chat: it registers giveaway entrants and collects poll votes in real time.
We share what we do without client names. Figures are internal measurements with the date and scope stated.
Ready for us to take your measurements?
Tell us your idea and receive a reference pattern and a breakdown in credits. It's preliminary and doesn't commit you to anything.
Take my measurements