AI Engineering · Enterprise Software · Cloud
We design, build and run AI systems, enterprise software and cloud platforms for public institutions and businesses across Europe, connected to the systems you already depend on.
A three by three grid of nodes connected by a single path that draws the FromNine mark: from business context through intelligence, software, cloud and operations to one system in production.
01 Scale and principles
Our focus is AI, cloud and modern engineering, backed by the scale and delivery discipline of a large firm. That means enough capacity to staff a multidisciplinary programme, with teams that stay from first design to daily operation.
2,400+people designing, building and running systems for public and private organisations.
600+European government projects, delivered and ongoing, for public bodies across the European Union.
400+projects for private companies
How we work with the public sectorBuilt around your systems
We work with the platforms, data and processes you already run, and replace only what needs replacing.
Designed for maintainability
Tested, documented and handed over, so your own teams can still change it years from now.
Human oversight where it matters
Consequential steps wait for a person, and every automated action leaves a trail you can review.
and public bodies across the European Union
02 Services
Useful AI depends on everything around it: software people want to use, data it can trust, integrations that keep working and a platform that keeps it running. We bring all of it under one roof, so the system works as a whole and one team answers for all of it.
Generative AI that holds up in production: grounded in your data, evaluated before release and observable once it runs.
Agents that act within the permissions you define and ask before consequential steps.
Governed data that decisions and AI can rely on.
Portals, internal platforms and SaaS products that stay easy to change.
Cloud foundations where releases are routine and costs are visible.
Systems that exchange data reliably, including when something fails.
Technical direction before large commitments: options, trade-offs and a roadmap you can defend.
Working on Salesforce, SAP or Odoo? See enterprise platforms
03 Solutions
Start with a problem, not a model. Each example follows the same path: the friction your teams face today, the system we would build, the benefit to measure and where people stay in control.
04 Public sector
600+European government projects, delivered and ongoing, for public bodies across the EU.
Government organisations need technology that is accessible, secure, transparent and built to last beyond a single tender. Across these projects we have learned to work within procurement rules, multilingual requirements and strict data protection.
We bring the same rigour to AI: documented decisions, human oversight and design that takes the EU AI Act into account from the start.
Dot map of Europe. All 27 EU Member States are highlighted, other European countries are shown in a lighter tone, and the offices in Antwerp and Ghent are marked.
05 Enterprise platforms
Customer, finance, supply chain and operations data already lives in Salesforce, SAP and Odoo. We implement, extend and integrate these platforms, and connect AI to them with the right permissions, so automation happens inside the processes your people already use.
Summit Partner
What we do
With AI
Assisted case handling, with drafts your service staff review before sending
Gold Partner
What we do
With AI
Document intake into SAP processes, with exceptions routed to people
Gold Partner
What we do
With AI
Automated back-office workflows on Odoo data
Salesforce, SAP and Odoo are trademarks of their respective owners.
06 Engineering depth
The model is one component. Around it sit the integrations, permissions, evaluation, approval steps and operations that decide whether it still works on a Monday morning, under real load. That is where AI projects succeed or quietly fail.
AI is only as useful as the systems it can read from and write to. We design the integrations before the prompts: mapping, retries, duplicate handling and errors people can see.
The system only retrieves what the person asking is allowed to see. Permissions are enforced in the retrieval layer, never left to the prompt.
Answers and actions are tested against agreed criteria before and after release. Actions with real consequences wait for a person to approve them.
Infrastructure as code, staged environments and monitoring of quality, latency and cost, so the system can be run, audited and improved after go-live.
The diagram shows six layers. From top to bottom: agents and workflows, with a human approval gate before consequential actions; models and evaluation; data and retrieval; the integration layer; and the business systems where records live, such as CRM, ERP, case management and document stores. A cloud platform with observability runs underneath all of them.
Access control runs vertically across every layer: identity, roles and audit apply at each step. A request enters at the top and passes down through every layer to the business systems.
Four annotations: integrations come first (integration layer and business systems); retrieval that respects permissions (data and retrieval, access control); evaluation and approval gates (models and agents); operated, not only deployed (cloud platform and observability).
07 Approach
Four stages, each ending in something you can review and keep. You see progress at every step, and you own what we build.
We map the business problem, the people involved and the systems already in place, then agree what is worth solving first.
OutputProblem definition
We choose an architecture, agree how success will be measured and plan a first release small enough to ship.
OutputArchitecture and success measures
We deliver in short, tested increments that your people can see, use and challenge.
OutputTested implementation
We hand over or run the system, monitor it in production and improve it with you.
OutputHandover and operating plan
What we learn in operation feeds the next round of understanding.
08 Trust and sovereignty
Responsible AI is an engineering discipline, not a policy document. Four layers of control wrap the systems we build, with your data and decisions at the centre.
We design for EU data residency and agree hosting locations with you before we build, including sovereign options where your requirements call for them.
Our information security management is ISO/IEC 27001 certified. Our standard delivery includes least-privilege access, threat modelling and secure engineering practices from design onwards.
Risk classification, documentation, audit logs and decision records, designed with the EU AI Act and GDPR in mind, so you can show who did what and why.
Approval steps, escalation and the ability to override, wherever an automated decision has real consequences for people.
Your data and decisions
Credentials
We design for GDPR, the EU AI Act and NIS2 in the systems we build.
How we handle security and data09 In your market
Few of our clients work in one country only. A public service launches in several Member States at once, a manufacturer runs sites under different national rules, a group moves its subsidiaries onto one platform. We design for that from the start: one architecture, with national differences handled where they belong rather than in a separate copy per country.
Our teams work in your language: English, Dutch, French, German, Spanish, Italian and other European languages. In Asia, teams work in English, Malay, Chinese and Thai. This website is available in six languages.
Public work is where we learned to build for accountability: 600+ European government projects, delivered and ongoing, for public bodies across the European Union. We know what European administrations expect: interoperability across borders, the once-only principle, accessibility under the European Accessibility Act, and documentation that holds up in procurement and audit.
A public service must launch in several Member States, each with its own language, identity scheme and accessibility law.
An organisation rolls out an AI assistant across departments and must show how it works, what it was tested against and who is accountable.
A group with entities in several Member States must notify significant incidents to different national authorities on NIS2 timelines.
Careers
Engineers, architects, consultants and specialists building AI systems, enterprise software and cloud platforms for public institutions and businesses across Europe. Work on problems with real users and real consequences, alongside colleagues who care about doing it properly.
Practices
A first conversation is about your business problem, the systems you already have and the most useful next step. No pitch deck required.
We run cross-border programmes with multilingual teams and a single point of accountability.
