Skip to content

Guztia AI Infra Ops

Guztia /About

The business

About Guztia

Guztia is an infrastructure operations business. We have run production systems for other organisations since 2015, across Europe and Asia-Pacific, and our current focus is the cost and control of enterprise AI.

What the business does

Guztia keeps other organisations' production systems running, available and affordable. That has meant cloud migrations, platform builds, managed hosting and round-the-clock operational cover since 2015, for companies from the mid-market up, mostly across Europe and Asia-Pacific.

Since 2024 the centre of gravity has moved to enterprise AI cost. The work is recognisably the same discipline — measure it, control it, run it, report on it — applied to a line item that arrived without passing through a budget process and is now growing faster than headcount in most of the companies we speak to.

We are a specialist business rather than a generalist consultancy. We do not have a strategy practice, we do not produce slide decks as a deliverable, and we do not staff engagements with people who have never operated a system. What we sell is a small number of things we are demonstrably good at, and we decline work that falls outside them.

The business also operates a long-standing China and Asia-Pacific practice, running since 2017: connectivity for mainland staff, site performance for mainland customers, regulatory filings and cross-border links. That practice continues unchanged and funds a meaningful share of the business.

Founded
2015. China and Asia-Pacific practice from 2017.
Registered
Operating across EU and Asia-Pacific hours.
Practice areas
Enterprise AI cost and governance · Infrastructure operations · China and Asia-Pacific market access
Organisations served
100+

01 Leadership

You will know who is accountable for your engagement.

Both partners work on client engagements directly. Neither of them is a layer you speak to before reaching the people doing the work.

PARTNER

Alberto Roura

Infrastructure & AI Operations

PARTNER

Xavier Macià

Client Strategy & Governance

Delivery is supported by a standing network of specialist associates the partners have worked with over many years — brought in by discipline where an engagement needs it, and named to you before they start.

02 How we engage

Commercial terms designed to be easy to leave.

Every term below exists because it removes a reason to distrust us. They are in the contract, not only on this page.

Fixed fees
Projects are quoted, not metered. If the work takes longer than we estimated, that is our problem — which is a considerably better incentive than billing by the hour.
No resale margin
Infrastructure and capacity are billed to you directly by your own providers, at your own rates. We take no commission from any platform and hold no reseller agreements, so we have no interest in you consuming more than you need.
Thirty days' notice
On every ongoing agreement. No minimum term, no multi-year commitment, no automatic escalation. An arrangement a client cannot leave is one that stops trying.
You own the output
Infrastructure code, documentation and accounts are yours throughout. We place nothing proprietary into a client environment, so ending an engagement requires no extraction project.
Named accountability
A partner is accountable for every engagement and an engineer is named on every operational agreement. Escalation reaches a person, not a queue.
We decline work
Where an assessment shows the numbers do not support proceeding, we say so and bill for the assessment only. We would rather lose a project than deliver one we do not believe in.
Confidentiality
Under your NDA as standard. Client names are not published without written permission, which is why this site carries no logo wall.
Written by default
Work spans European and Asian hours, so decisions and updates are documented rather than relayed in a standing call. You can read what happened without having attended.

03 Why a specialist business

The two obvious alternatives, and where they fall down.

A large consultancy will produce a considered strategy and a capable team, and will bill for the months it takes them to learn what your systems do. The people who write the recommendation are generally not the people who implement it, and neither group holds the pager afterwards.

Hiring for it works if you can find and keep a platform engineer who has done this before, which at present is difficult and expensive. It also concentrates the knowledge in one person, which is its own risk — we are often called precisely because that person has moved on.

We are the option in between: a business small enough that the partners do the work, established enough to carry operational commitments, and specialised enough to have done this particular thing before.

2015
Operating since
100+
Organisations served
Alibaba Cloud MVP
2
Continents covered daily
  • We run the same arrangement we sell, in our own production — it is not a demonstration assembled for the pitch.
  • The technical archive on this site goes back to 2015 and is not gated. Clients routinely arrive having already decided whether they agree with us.
  • Deep specialisation in Asia-Pacific and China infrastructure, which most European businesses in this space simply do not have.
  • Governance and regulated-environment experience through the partners' post-quantum security work at QuReady.

Own the stack before the bill owns you.

Thirty minutes. Bring your current AI invoice. We look at what your teams send to model APIs, what it costs today, and what a self-hosted stack would cost instead. No deck, no discovery phase.