Why start now?
AI sets the tone, you decide the tempo
Automate
Streamline repetitive tasks and win back precious time for creative work.
Predict
Leverage historical data to map demand, risks and promising niches with precision.
Personalise
Deliver content and offers that feel hand-crafted for every individual customer.
Iterate
Keep learning — models continuously refine themselves based on new feedback and data.
Data tasting and algorithm seasoning.
Data tasting and algorithm seasoning
Like cooking with flavours
Data is your pantry
Without fresh, clean data you are cooking with spoiled ingredients. Check sources, remove noise and structure everything — a messy cupboard leads to unpredictable flavours.
Model is the recipe
You can serve haute cuisine with a simple recipe if the technique is right. Convolutional, recurrent or transformer? Choose what fits your business question, not what is trending on GitHub.
Humans remain the chefs
AI suggests; you taste. Keep weighing, tweaking, seasoning. In practice the real game-changer is the blend of domain expertise and algorithmic power.
Clean data prevents a nasty aftertaste
Start small, learn fast, scale big
Do not forget governance — compliance is the salt that keeps everything in balance
Roadmap to AI success.
A practical route from idea to impact.
Step 1 — Peel back the problem.
First determine which concrete business friction you want to solve. Think smaller than big: "triage customer questions" instead of "revolutionise customer service".
Step 2 — Draw your data map.
Map sources, ownership and quality. Many people underestimate how much work data cleaning really is.
Step 3 — Build a prototype.
Use open-source libraries or managed platforms to train a minimum viable model. Test it out loud with end users — feedback is gold.
Shadow mode: run your model alongside the current process without direct impact. Measure the difference, fine-tune parameters.
Step 4 — Productise.
Integrate the model with APIs, monitor performance and set up automatic retraining. No more manual tinkering.
Step 5 — Embed the culture.
Make AI literacy part of regular meetings. Technology fails more often on the human side than on the technical side.
A philosophical view of algorithms
Can machines truly understand what we mean?
Our relationship with technology is a hall of mirrors. The smarter the code, the more we wonder what "smart" actually means. You often hear the question: is AI conscious? For clarity: no. A neural network optimises a mathematical function, nothing more. Yet the outcomes can feel as if there is understanding behind them — that is exactly where both confusion and potential lie.
Homo in machina
We love to project humanity onto systems. That can be useful (think conversational agents) but risky when we shift moral responsibility to software.
Limits of interpretation
Algorithmic decisions remain statistical. That means probability distributions, not certainties. Organisations that underestimate this nuance often end up with disappointed stakeholders.
Ethics as a built-in feature
Governance should be designed from day one: fairness checks, explainability, audit trails. Otherwise you will later find yourself in a swamp of legal and reputational damage.
Practical tips for quick wins
Small budget, big impact — start tomorrow
Start with existing SaaS tools that already offer AI functionality (think automatic tagging, anomaly detection)
Use transfer learning: take a pre-trained model and fine-tune it on your domain. Saves weeks of training time and energy consumption.
Set up a data contract between teams. It may sound dull, but agreeing on definitions prevents endless discussions later.
Monitor drift: models age faster than you think. Schedule periodic health checks in your sprint backlog.
Document successes and failures. An internal "AI logbook" accelerates learning and prevents repeating mistakes.
A critical note
When you might want to wait with AI
There are scenarios where the hype is bigger than the value. For example when:
Data is minuscule
A dataset of 300 rows? Then a simple Excel model could make more sense.
Regulation is strict
Financial sector? E-privacy? Sometimes the compliance burden outweighs the benefits.
Legacy refuses to bend
If your core systems date back to 1998, integration can be more expensive than a rewrite. Upgrading infrastructure will solve it eventually, but budget needs to be allocated first.
Brainstorm about smart deployment?
Interested in testing your idea against our practical experience? Send us a message — we are happy to think along about a small-scale experiment that shows value quickly.
"A deal is a deal"
Yield.inc is a new asset manager in the Netherlands that focuses on excellent customer experience, sustainability and technology. When faced with the question of whether to develop our platform in-house or with a partner, we got in touch with Spartner. We are still 100% behind the choice for Spartner because we own the source code and with Spartner, a deal really is a deal, whether it concerns the delivery date or the agreed budget!
Hans Lugtenberg Partner at Yield Inc.
"A professional software partner since 2010"
Spartner is involved and contributes in our search for innovations and always delivers top-notch work. After the first Huurda.nl version got out-dated, we collaboratively launched a completely new version in 2020.
Custom-built search engine for rental propertiesNorbert Wegter Owner of 123wonen and Expat Homes Holland
"Pragmatic and flexible collaboration"
Spartner has created a customized portal for us for the accountability of large-scale collaborative projects. This enables us and our partners to execute and account for their plans in a shared environment that is secure, transparent, and manageable. The pragmatic and flexible collaboration with Spartner has ensured that we have quickly achieved a functional and user-friendly environment.
Jurjen Terpstra Managing Partner at Wecreate Consulting
"AI and machine learning as legal tools"
In 2020, we transferred the development of Lynn to Spartner. As a result of their fast development process, the Lynn platform achieves an increasingly central role in the legal world.
Lynn, the smart AI legal robotMarco Caspers Software Development Manager at Lynn Legal
"As a business you must dare to move forward"
Innovation and change are not always welcomed with equal enthusiasm by our employees, man is a creature of habit but as a company you have to dare to move forward and the need for a contemporary drawing program for the realization of various constructions with our Masterbloc bricks was high. For us but also for Spartner, this challenge was a bit of pioneering and together we had to overcome some hurdles to arrive at a beautiful and efficient business tool. A development and learning process for both parties, in which the feedback of questions and desired adjustments by our employees to Spartner was always in good consultation and we could count on quick feedback and targeted solutions. We continue to exchange experiences with Spartner and look back and forward to a successful collaboration.
Powerful 3D construction drawing softwareBjorn Gubbels Owner of Masterbloc
"Spartner extends our capabilities and development capacity"
With weekly calls, using our Azure DevOps sprint system, Spartner actively works together with our innovation, communication and software development departments.
Multi-website CMS with Statamic 3Marco Leenders Head of Automation
"Flexible collaboration with yearly updates"
The projectteam of the city of Groningen is very enthusiastic about the "Roomfinder" platform.
Real estate management software for student roomsGemeente Groningen Projectteam
"Have been using Laravel Excel for years"
We requested custom help to a performance challenge we had in our implementation of Laravel Excel. I was amazed how quickly these Laravel artisans achieved significant performance gains, which saved us a lot of development time.
Open source package for LaravelJosh Mountain Co-Founder at IncentivePilot.com
"Higher occupancy rate thanks to our user-friendly platform"
Throughout the great years that we cooperate with Spartner (previously Maatwebsite, Ed.), our software has been developed continuously. Students and landlords actively work with features like allocation, payments, chat, contract generation and more.
Custom built ERP for all student residencesMaurice Evers Head of Student Housing Department
"Innovating in the digital identification network of The Netherlands and Europe"
The pro-active approach of Spartner was crucial in understanding the complexity of the eHerkenning network.
Development of eHerkenning connection and servicesMaarten Bremer CTO / Founder Ensured
.webp)
"Innovation in legal technology"
As jurists with knowledge of IT, we highly value quality; within code, but also in process. The high work-level and trustworthiness of Spartner gives us the capacity to continuously create innovative features.
Legal tech platformNiels Winters Managing Director JuriBlox
What is the difference between machine learning and AI? 🙂
Machine learning is a subset of artificial intelligence where systems learn patterns from data. AI is the broader umbrella that also covers logic, planning and language understanding.
How much data do I need at minimum? 🤔
There is no magic number, but as a rule of thumb: the more complex the problem, the more examples you need. For simple classification a thousand rows might work; for nuance-rich language models you are talking millions.
Can I use AI without a data scientist in-house? 😅
Yes, through low-code platforms and ready-made APIs. But as you go deeper, specialist knowledge becomes indispensable to steer bias and performance.
How do I prevent bias in my model? ⚖️
Start with representative data, use fairness metrics and let diverse teams test. As mentioned earlier, ethics is not an afterthought but a core component.
Is AI expensive to maintain? 💸
Maintenance mainly costs time from people who handle retraining and monitoring. Cloud costs are falling, but do not forget the expertise needed to interpret alerts.
When will I see the first ROI? ⏳
With well-defined use cases a proof of value can provide insight within weeks. Full ROI depends on integration, scale and adoption.
Do I need a GPU cluster? 🚀
Only for heavy deep-learning workloads. Many business models run fine on CPU or shared cloud GPUs.
Will employees be replaced by AI? 😟
Some tasks will, but new roles are emerging: prompt engineer, AI ethicist, model steward. It is about reskilling and task redistribution, not pure substitution.
