What a project is here

When a question becomes a project

Not every question is a project. A project starts once something has to be built that does not yet exist, with an agreed result: a forecasting model in the planning process, visual inspection in the production line, a reporting environment that decisions can rest on.

Before we start, we agree what the result should be recognisable by, which decision changes, which number has to move. If that cannot be said yet, the work starts as an advisory question rather than a project. That is shorter and cheaper, and sometimes the outcome is that you should not build at all. That advisory question is what our AI consultancy is for.

A few examples

Twentynext builds solutions for a wide range of questions, from demand forecasting to image recognition. You will find a full overview under our solutions.

From scope to production

Six steps, from scope to support

Every project goes through the same six steps. Not because every question is the same, but because every project has to answer the same questions: what needs to improve, can the data carry it, and who keeps it running when we are done.

Three colleagues looking at a screen together

Scope: which decision or action needs to improve, and how do we measure it? We settle that before anything is built. If the question cannot be made sharp yet, we say so: sometimes a smaller engagement fits better, or no project at all.
Data exploration: first we check whether the data can carry the answer: where it lives, how reliable it is, what is missing. The outcome can change the scope, better now than halfway through.
Build: models, pipelines and screens, built by the same people who did the analysis. For the data-science core we follow the CRISP-DM methodology, which treats going back a step as normal rather than exceptional.
Testing: not just on a test set, but in the real work process, with the people who will use it. A model that scores well but goes unused is not finished, as far as we are concerned.
Delivery: done, for us, means: it runs in production, on your systems, documented, with people who know how to work with it. A demo is not a delivery.
Support: after go-live someone stays responsible, us or your own team, agreed up front. The world around a model changes; someone has to notice when the model starts to slip.

Agreed up front: what you may hold us to

At the start of every project we agree one measuring point: which number should look different in twelve months, and how we measure it. That one number comes back at delivery and a year later. So you know up front what you may hold us to, and we build up proof that goes beyond a good story.

Where the methods come from

Twentynext R&D

Much of what we apply in projects, we first tested in our own research, together with universities and healthcare institutions. So we know not only that a technique works, but where it stops working.

Realise your project?

Curious whether your question lends itself to a project? Get in touch: we will also tell you honestly if a smaller form will do.

Schedule an introduction

Our services

What else can we do for your organisation?

A few services that may help your organisation further:

Data-driven working

Data architecture

Service & Maintenance

Data & Reporting Platform

Microsoft Copilot implementation

Research & Development

May we measure how this site is used?

We use Google Tag Manager to measure visits and Leadinfo to recognise which company is visiting. Neither loads unless you agree. If you choose essential only, the site works as normal and we measure nothing. If you arrived via an advertisement in ChatGPT, we also use the OpenAI measurement pixel to attribute conversions to that advertisement. Cookie statement · Privacy statement

Projects