Question 1

What does Arazon actually do?

Accepted Answer

We design, build, and deploy AI systems. Strategy, custom model development, end-to-end applications, automation, integration, data engineering. The output is production code you own, not a slide deck you shelve.

Question 2

What industries do you work with?

Accepted Answer

We're industry-agnostic. Financial services, logistics, healthcare, media. The common thread is a real business problem where AI can make a measurable difference. We'd rather understand your specific challenge deeply than claim vertical expertise we haven't earned.

Question 3

Do you work with startups or only large companies?

Accepted Answer

Both. The qualifying factor is the problem, not the company size. If AI can create measurable value for your business, the engagement model scales to fit.

Question 4

Where is Arazon based?

Accepted Answer

EU-registered company, co-founded by an enterprise architect with over 20 years of experience, including roles at Razorfish, Akamai, and Deloitte. We deliver globally: on-site, remote, or through our partner network.

Question 5

How do I get started?

Accepted Answer

Reach out through our contact page and describe the challenge. We'll schedule a call to understand whether and how we can help. No commitment. No pitch deck. A direct conversation about the work.

Question 6

How is pricing structured?

Accepted Answer

Three models. Discovery: fixed-scope strategy for assessment and roadmapping. Build: custom-quoted development for designing and shipping a specific system. Partnership: monthly retainer for ongoing iteration and support.

Question 7

What does a typical engagement look like?

Accepted Answer

Most clients begin with Discovery to map where AI fits their business. That leads to Build, where we design and ship the system. Some continue as a Partnership for long-term iteration. Each phase has clear deliverables and exit points.

Question 8

How long does a typical project take?

Accepted Answer

Discovery runs a few weeks. Build engagements range from a few months to longer, depending on scope. We provide detailed timelines before work begins. No open-ended commitments.

Question 9

Do you provide ongoing support after delivery?

Accepted Answer

Every Build engagement includes a post-delivery support period for handover. For ongoing development, the Partnership model provides continuous access to our team.

Question 10

Is there a minimum engagement size?

Accepted Answer

No published minimum. Discovery is our smallest offering, designed to be accessible. For Build and Partnership, scope and pricing are determined by the work itself.

Question 11

Do I own the code you write for me?

Accepted Answer

Everything. Source code, trained models, data processes, documentation. No licence fees, no proprietary dependencies. You can hand the entire codebase to another team tomorrow. Nothing ties you to us except the quality of the work.

Question 12

What does 'no vendor lock-in' mean in practice?

Accepted Answer

No proprietary frameworks. No closed platforms. No licensing structures that require you to stay. Your codebase runs on standard, open technologies. If you part ways, you walk away with everything. Fully functional, fully documented, fully yours.

Question 13

Can I have the code independently audited?

Accepted Answer

Yes, and we encourage it. Full source code handover means you or any third party can audit, review, and inspect everything we've built.

Question 14

What cloud platforms do you work with?

Accepted Answer

AWS, Google Cloud, Microsoft Azure, hybrid, and on-premises environments. We recommend infrastructure based on your requirements, not partnership commissions.

Question 15

Do you use open-source or proprietary AI models?

Accepted Answer

Whichever fits the problem. Llama, Mistral, and other open-source models when you need full control. OpenAI or Anthropic when commercial services are the right trade-off. Custom-trained models when nothing off the shelf is good enough.

Question 16

Can you integrate AI into our existing systems?

Accepted Answer

Yes. Your systems stay. We connect AI to databases, cloud services, customer platforms, business tools, and custom applications. The integration fits your business. You don't rebuild around ours.

Question 17

What programming languages and frameworks do you use?

Accepted Answer

Python and TypeScript primarily, with others as needed. PyTorch, TensorFlow, and Hugging Face for AI. We choose tools to match the problem, not the other way around.

Question 18

Can you fine-tune or train custom AI models?

Accepted Answer

Yes. We fine-tune foundation models on your data for domain-specific performance and build custom models when off-the-shelf falls short. Every trained model ships with full weights, training code, and documentation.

Question 19

How do you handle our data?

Accepted Answer

Terms agreed before work begins. GDPR-aligned practices, encryption at rest and in transit, data minimisation. Your data is never used to train models without explicit written consent.

Question 20

Are you GDPR compliant?

Accepted Answer

EU-registered company. GDPR-aligned data handling across all engagements. Privacy built in from the start: minimal collection, anonymisation where possible, clear processing agreements.

Question 21

What security practices do you follow?

Accepted Answer

Architecture reviews, code review, dependency management, role-based access, multi-factor authentication, least-privilege principles. Industry-standard encryption. Cloud security best practices across AWS, GCP, and Azure.

Question 22

Will my data be used to train your other models?

Accepted Answer

No. Client data is never used for other clients or for our own purposes without your explicit written consent. Your data stays yours.

Question 23

What is your approach to responsible AI?

Accepted Answer

Six principles govern every project: safety first, transparency, privacy by design, fairness and bias awareness, human oversight, accountability. These are built into our development process and delivery contracts. Not aspirational.

Question 24

Do you test for bias in AI systems?

Accepted Answer

At data selection, model training, and evaluation. We test for fairness across relevant dimensions and document findings. Bias testing is ongoing, not a one-time step.

Question 25

Are there projects you won't take on?

Accepted Answer

Yes. No surveillance tools. No weapons systems. No user manipulation. No systems designed to deceive end users. If a project conflicts with our principles, we decline it.

Question 26

How do you ensure AI decisions are explainable?

Accepted Answer

We design systems where decisions can be traced and understood. For high-stakes decisions, human-in-the-loop controls ensure final authority stays with people. Every delivered system includes documentation of capabilities and limitations.

Your Questions, Answered.

Straight Answers. No Jargon.

Ask Us Directly.