geg gegz.nl

From raw data to decisions. Data science and AI, grounded in science.

gegZ builds data pipelines, predictive models and private AI systems for organisations, researchers and nature projects. We handle the messy data, test the models properly and hand over tools people actually use.

Start a project Try our labs
200M+data records processed and analysed
TBterabytes of tabular, spatial and sensor data handled
400K+images processed for computer vision
And still counting
What we do

Four ways we put your data to work

Every engagement starts with a question someone has to answer, and ends with something they can use: a clean dataset, a tested model, a dashboard or an AI assistant that knows your organisation.

Custom data analysis and engineering

We collect, clean, link and analyse data at any scale, from one messy spreadsheet to hundreds of millions of records. Pipelines, databases and statistics that give you tailored answers for informed decisions.

  • Python
  • R
  • SQL
  • ETL pipelines
  • statistics
  • dashboards

Predictive modelling and machine learning

Forecasts, classifiers and computer vision built with precision and validated on data the model never saw. From time series and risk scores to image recognition across hundreds of thousands of photos.

  • forecasting
  • classification
  • computer vision
  • deep learning
  • model validation

Local AI as a service

Private AI assistants that run inside your organisation and answer from your own documents, databases and archives. No data leaves your premises. We set it up, tune it, train your team and keep it running.

  • open-weight LLMs
  • RAG
  • on-premise
  • GDPR
  • EN · NL · AR

Ecological and spatial solutions

Species distribution models, GIS suitability maps, biodiversity analytics and visual digital twins. Science-based designs for regreening, restoration and nature-based solutions.

  • GIS
  • SDM
  • remote sensing
  • digital twins
  • biodiversity
Local AI as a service

Your own AI, on your own ground

Most organisations sit on years of reports, datasets, protocols and email that nobody can search well. We build AI assistants that know that material, run on your own hardware and never send it to an outside provider.

Private by designModels run on your servers or a dedicated machine we supply. Your data stays in your building.

Answers with sourcesEvery answer points to the documents and records it came from, so people can check it.

Speaks your languagesEnglish, Dutch and Arabic, including documents that mix them.

More than chatIt can run analyses on your data, draft reports and summarise projects, not only answer questions.

Built by gegZResearch group · NL

An AI research assistant for ecologists

A private assistant for a university research group. Researchers ask questions in Dutch or English about their own datasets, methods and literature, and it answers with sources and can run analyses on uploaded data.

Built by gegZResearch institute · NL

An institutional knowledge and analysis system

A knowledge system for a research and technology institute. It connects project documents, reports and data across themes, so staff can find what the organisation already knows and analyse it in one place.

Labs · live in your browser

Two small experiments you can run right now

Both run entirely on this page. Nothing you type or set is sent anywhere.

Ask a private archive a question

This is the retrieval step behind a local AI assistant. Below is a small sample archive of a fictional research organisation. Your question is turned into a TF-IDF vector and compared with every document by cosine similarity, all in your browser.

Extracted answer

Sample documents only. A local language model would write a fuller answer from these same sources, on your own hardware.

How a project runs

One pipeline, from raw data to the boardroom

01 · Collect

Every source

Databases, spreadsheets, images, documents, sensor feeds, field surveys and satellite time series.

→ raw data
02 · Engineer

One clean foundation

Cleaned, linked and documented to FAIR standards, in pipelines that scale to terabytes.

→ analysis-ready data
03 · Model

Statistics and AI

From classical statistics to machine learning, computer vision and language models, validated before anyone relies on them.

→ tested predictions
04 · Deliver

Tools people use

Dashboards, web apps, maps, local AI assistants and reports, handed over with training.

→ decisions
Our work

Projects we have delivered

Data science, modelling and AI behind regreening, restoration and digital twins on three continents.

Web app · open to allNL

Netherlands Biodiversity Explorer

An effort-aware, multiscale explorer of biodiversity in the Netherlands across space, time and taxonomy. It corrects for where people looked hardest, and every view runs as a live query in your browser.

Open the explorer ↗
RegreeningEgypt

Lake Bardawil and Sinai initiative

Regreening an arid region of northern Sinai.

  • Selected plant species suited to the site
  • Produced GIS maps of plant distribution and potential planting locations
  • Predicted suitable environmental conditions for planting
  • Led the data analysis
Master planNuweiba · Egypt

Green Corridors, Sinai

A master plan for green corridors that connect farms across an arid coastal region, designed from all available ecological and environmental data and validated with experts and the local community.

106farms connected
893corridors
183 kmtotal length
192,772plants planned
Visual digital twinsNL & abroad

New Horizon Initiative

Visual digital twins of environments in the Netherlands and abroad, made with artists and designers.

  • Identified the target plants
  • Modelled plant distribution and environmental conditions
  • Verified the final visual products against the data
  • Helped design the visual outcome
Digital twin · ongoingEemshaven · NL

The New Polder

A digital twin of the Eemshaven region in the north of the Netherlands.

  • Identified target plants and animals
  • Produced GIS maps of their distribution
  • Modelled the environmental conditions
  • Building the models behind the digital twin
Arid regreeningBotswana

Regreening arid regions in Africa

Design work for regreening dry regions of southern Africa, in the spirit of the Great Green Wall.

  • Supported the design phase
  • Identified the target region in Botswana
  • Mapped the relevant environmental conditions
  • Validated the outcome and co-wrote the final report
Nature-based solutionMozambique

Ecological mangrove restoration

With the community of Icidua, we designed and implemented hydrological mangrove restoration to reduce erosion, flooding and salinity.

1,000+new mangroves rooted
13months after the channels
~100residents trained
Where we work

From 53° north to 19° south

Our projects run from the Dutch coast through the Sinai to southern Africa. The landscapes change. The data questions stay surprisingly similar.

EnglishNederlandsالعربية
Why gegZ

How we work

Science-gradeMethods that hold up in peer review, with uncertainty stated plainly.
Your data stays yoursLocal and private setups by default. No silent uploads to third parties.
Built to hand overDocumented code, training and tools your team can run without us.
Three languagesWe work in English, Dutch and Arabic, across Europe, the Middle East and Africa.
Contact

Have data and a question? Let's talk.

Tell us what you are trying to decide and what data you have. We reply within two working days, in English, Dutch or Arabic.

info@gegz.nl