RAXARDel Caos al Éxito
Loading ecosystem
Managed AI · measurable
Enterprise AI, managed.
And measurable.
You know where your data is hosted and who it is transferred to. In the systems we run, personal data is redacted before it leaves (with the exceptions we list in our Privacy Policy). Every action the system takes is traced, and RAXAR runs it for you.
We tell you who your data is transferred to
The list of processors and transfers outside the EU is in our DPA, provider by provider: some AI models come from providers outside the EU, and we say so. The databases of the systems we run are self-hosted. Audience measurement (Google Analytics) only runs if you accept it.
We redact personal data — and we say where we don't
In the systems we run, the message is cleaned before it goes out to an external model: email, phone, national ID, NIE, passport, IBAN and card. With local models nothing leaves, so there is nothing to redact. And we say it ourselves: the public chat on this website and the other cases we list in our Privacy Policy do not carry it yet, and detection is designed for Spanish formats.
Real AI Act transparency
When someone talks to our AI, they know it — that's the transparency obligation under Art. 50. Our systems are mostly classified as limited or minimal risk; Annex III (high-risk) doesn't apply to us as-is, and if your case would require it, we assess that with you before building.
The difference
RAXAR vs. generic cloud AI
Honesty first: we use quality models from US providers, with each one's safeguards in our Privacy Policy, and in the systems we run we redact personal data before it leaves, with exceptions that we also set out there. Not yet in the public chat on this website — which is why we ask you not to type personal data there. What we offer is control, knowing where the data is and a real local option.
Axis
Generic cloud AI
RAXAR
Data residency
Generic cloud AI
Where data is hosted and who it is transferred to is not always stated.
RAXAR
Hostinger VPS in the EU, with self-hosted PostgreSQL and vector store. Model providers and their transfers are listed in the DPA.
Personal data (PII)
Generic cloud AI
The full prompt — personal data included — travels to the model provider.
RAXAR
In the systems we run, we redact personal data (email, phone, national ID, NIE, passport, IBAN, card) before calls to external models, but not in every case: the exceptions are in our Privacy Policy. The public chat on this website does not apply it either, and we flag it right there.
Local option
Generic cloud AI
Doesn't exist. You depend on the provider's cloud for everything.
RAXAR
Real private mode: we route to local models on our own machine. PII never leaves the box.
AI Act transparency
Generic cloud AI
Generic “Powered by”, with no standardized AI disclosure.
RAXAR
AI disclosure on every surface (chat, email, analysis), implemented and verified in code, in ES/EN/PT.
Synthetic content marking
Generic cloud AI
Absent or in a proprietary format only they can read.
RAXAR
C2PA marking plan + machine-readable metadata — an interoperable industry standard.
Tenant isolation
Generic cloud AI
Logical multi-tenancy: you trust their separation policy.
RAXAR
Mathematical isolation (RLS) on every table with client data — verified database-level policies.
Human oversight (HITL)
Generic cloud AI
Opaque, end-to-end automation.
RAXAR
Human-in-the-loop on every sensitive action: collections, hiring, operations. Explicit approval.
Legal basis for transfers
Generic cloud AI
You accept their terms. Opacity around clauses and risk assessment.
RAXAR
The safeguards of each provider, set out in our DPA (in some cases the provider's mechanism is not specified), and prior redaction of personal data in the systems we run (with the exceptions we list in our Privacy Policy).
Who runs the infrastructure
Generic cloud AI
You, or a separate team you hire and coordinate.
RAXAR
RAXAR runs it. You use it; we operate, monitor and maintain compliance.
GDPR data-subject rights
Generic cloud AI
Generic forms and variable timelines.
RAXAR
Rights channels (access, erasure, portability, objection) with a 30-day SLA and written procedures.
Verifiable technical claims. Legal conclusions are validated with legal counsel before being published as final.
Local-first and measurable
Predictable cost. And numbers we show.
Local-first = predictable cost
Routine work runs on low-cost in-house models; only the genuinely hard part scales to frontier models. The retainer is stable and doesn't spike with volume, unlike the typical per-token SaaS.
Measurable, no make-up
Every action is traced: cost, latency, outcome. We even measure our own use of local compute — without dressing it up — and we publish it. That transparency is part of the product.
Proof number one
RAXAR runs itself
We don't show you a dossier: we show you our own company. We use AI agents in our own operation — infrastructure, sales, customers, intelligence, finance, among other areas — and our infrastructure has continuous automated monitoring with alerts.
“Before we sell it to you, we use it to run our own company.”
AI agents in our own operation
Continuous automated monitoring with alerts
Internal first, then external
How it works in practice
A ladder, not one-size-fits-all
You start where it hurts most and scale with judgment. Every proposal is tailored: we close the price with you in the assessment, no small print.
Entry
AI Start
Your first piece of AI, in production
One concrete bottleneck solved with a system in production, with the timeline set in the proposal. Try it before you scale.
- 1 turnkey AI system for one specific process, defined in the proposal
- Access to the control panel
- Managed hosting (Hostinger VPS in the EU)
- Monitoring + one monthly tweak included
Tailored proposal · we close it in the assessment
Let's talkRecommended
Recommended
AI Operation
Several AI systems in your business processes
Three AI systems integrated into specific processes of your company, defined in the proposal, with a person approving critical decisions.
- 3 integrated AI systems, tailored to the business
- Human-in-the-loop on critical decisions
- Full dashboard with KPIs and real cost, up to 5 users
- Managed hosting + two improvements a month
- Team onboarding
Tailored proposal · we close it in the assessment
Let's talkEnterprise
AI Core
Tailored AI systems for critical processes
Critical processes, complex integrations and SLA demands, with the option to deploy in your own house.
- Architecture tailored to your areas and processes
- Integration with any existing system (ERP, CRM, private APIs)
- Unlimited dashboard + written availability SLA
- On-premise option: the data stays on your infrastructure
- Extended training + dedicated support
Tailored proposal · we close it in the assessment
Let's talkPrograms
RAXAR is a member of NVIDIA Inception, NVIDIA's program for startups building on accelerated AI.
Where to start?
30 minutes.
With your numbers.
We analyze your most inefficient process and you walk away with a concrete plan: which system fits, where your data lives and what results to expect.
Diagnostic, no commitment