AI and systems architect. Designing enterprise integrations across eight countries since 2002. Today I design the harness that lets agents operate inside real companies: agentic graphs, permissions, evals and traces. Not in slide decks.
In the press
20+
years in systems
8
countries
1,500+
students trained
16
media appearances
Three formal recognitions any procurement committee can verify.
IEB, 2019
Teaching technology, blockchain and applied AI to executives.
Professor at IEB (Instituto de Estudios Bursátiles), Spain's leading finance and management school. Teaches courses on technology, blockchain and applied AI for finance-sector executives.
Source: Profile on ieb.es ↗
Draper University, 2023
Tim Draper's program. AI hackathon at Google. Event at OpenAI.
Hero Training 2023 at Draper University in Silicon Valley. Tim Draper's selective program focused on tech leadership and entrepreneurship. Included participation in an AI hackathon at Google and a generative-AI event at OpenAI.
Source: draperuniversity.com ↗
Funds Society, 2025
Recognition by Spain's reference business-media outlet.
Selected among the 10 Spanish AI reference figures in the article «Capitanes de la IA · influencers a seguir en 2025», published by Cristina Murgas Aguilar on February 24, 2025. Highlighted for his trajectory in blockchain, middleware and decentralised AGI work.
Source: Funds Society, February 2025 ↗
Most companies are experimenting with AI without foundations. APIs taped together, prompts nobody has versioned, technical decisions made by someone who does not understand what they are building.
The result is predictable: demos that work in the presentation and break in production. The difference between an agent that works and one that does not is exactly the same as between well-designed software and improvised software.
I design the missing layer: the architecture that turns an idea into a real system. Agents that operate autonomously on an agentic graph, pipelines that scale, integrations that survive the real world.
Your systems
MCP connectors
Harness what stays
Agentic graph
Model
swapped every quarterescalates, audit
Person
approves critical stepsYour team
traces and metricsEach step is a node with a contract: what it receives, what it returns and what it may touch. The graph decides what runs in parallel, where results come together, which branch to take, how many times to retry and where to stop and wait for a person. The model reasons inside each node; the architecture decides the path.
input: order_by_emailfan_out: - stock@erp - credit@crm - shipping@apifan_in: merge_and_check # eval, max 3checkpoint: after merge_and_checkrouter: amount > 10000 ? person : erp
“From now on companies will be led by humans who know what and how to ask instead of those who claim to know the answers. Having a programmer's mind will help a lot in the early years.”
“This stage I mentioned in 2022 is over. In the early AI wave, the advantage belonged to those who knew how to ask the right questions. Now the leverage shifts again. The real edge belongs to people who understand software architecture, systems, and logic. Builders win this phase. And the next one won't take four years; it will be 1 or 2 years at most. When it arrives, no one will have an advantage anymore.”
The one-liners from clients over the past 18 months.
One hour with me closes an internal debate that has been open for weeks. I review your architecture, your stack or your AI plan and you leave with a defensible direction, not another list of options.
Typical client
A CTO with an AI project stalled for lack of outside technical judgment.
I audit integrations, pipelines, agents and distributed systems with the same rigor applied at Correos, Severn Trent or the City of Madrid. I do not review code: I review the system around it.
Typical client
Technical team whose demo works on a laptop and breaks in production.
I design agent architectures that actually operate inside the business: agentic graphs, state, evals, permissions, MCP connectors to your systems, and traces that survive an audit (and the AI Act). The same thing I did with middleware, now with models that change every quarter.
Typical client
Leadership with an agent pilot that never reaches production: security, governance and integration are holding it back.
With Sistek: agents that run SMB workflows as explicit graphs, with approval gates, permissions and traces. The model gets swapped; the harness stays.
With Sistek: agents that run SMB workflows as explicit graphs, with approval gates, permissions and traces. The model gets swapped; the harness stays.
Sistek, Harness engineering
Madrid
Conversational agent on the website of a Madrid restaurant group: its own persona, bookings written straight to the venue's database and daily and weekly reports to management over Telegram.
Production, AI agents
Designed AI agents running the complete operation of a music academy. Cut from 13 hours to 2 hours of daily manual management.
Feb, Mar, AI agents, Automation
Lugano, CH
International collaboration on reference blockchain infrastructure. Node architecture, smart contracts and proof of concepts for European banks.
UK
After the TNT project, a deliberate pivot toward blockchain in the UK, right as the crypto ecosystem was expanding there. First contracts and talks focused on node architecture and enterprise integration with public chains.
Blockchain, UK
Coventry, UK
Real-time analytics on the UK water grid. Digital, analog and weather sensors feed an ODM rules engine that detects incidents before they happen. Also implemented the unique household identification system via numeric identifier, the operational backbone of the service.
DataPower, ODM, IIB, IoT, Household UID
Training programs on applied AI for executive teams at IBEX-listed companies and central banks. Technical judgment, not hype.
Silicon Valley, CA
Tim Draper's Hero Training. AI hackathon at Google. Event at OpenAI. Immersion in the Silicon Valley ecosystem during the generative AI acceleration.
Google, OpenAI, DraperU
Academic director of an intensive program with 140 students. Technical curriculum covering development, architecture and applied AI foundations.
140 students
Madrid
Teaching technology, blockchain and applied AI. Digital innovation expert panel. Recognized by the institution as "exceptional talent".
Madrid
Teaching on the ISDI Blockchain master's. Distributed systems engineering, node architecture and applied cryptography foundations.
EMEA
Technical training for the central banks of Denmark and Spain, Accenture, BEC.dk and teams in Saudi Arabia. Middleware, ESB, integration architecture.
Recognized as one of the most influential people in artificial intelligence in Spain.
Analysis published in Spain's two leading business outlets. Commentary on AI, markets and applied technology.
Recurring appearances as technology, AI and markets analyst. Intereconomía, Myconomy.
Documentary by Spain's highest-circulation newspaper on the history and future of bitcoin. Technical analysis and strategic judgment.
The Faces of Money
Interviewed on the flagship news program of Spanish public television on digital money, stablecoins and the future of payments.
International circuit: Singapore, Helsinki, Istanbul. Panels on crypto infrastructure, digital assets and institutional stability.
Talks on blockchain architecture and applied AI. Reference European conferences.
Essays on agents, compilers and verification. The first: The operating languages, on why determinism moves to the verification layer.
alber.one/en/blog
Technical meetup series hosted at Google Campus Madrid. Gatherings of engineers, financial institutions and companies on blockchain, infrastructure and applied AI.
Real photos from the milestones on the timeline.
Fintech Junction
Tel Aviv, 2019
Google Campus Madrid
Technical meetups, 2019
TVE, Informe Semanal
Spain's flagship news show, 2021
Merge Madrid and MWC 4YFN
Conference, 2022
Each plate documents a real project with technical metadata. No mock-ups.
2013, Blockchain, B2B integration
NXT, proof of stake, on-chain messaging
Proofs of concept for B2B integration on NXT, one of the first proof-of-stake blockchains, using its native messaging layer as the channel between companies, with no central broker. The line of work that later led me to Jelurida.
2013, Logistics, enterprise
XML, EDI/Edifact, ESB
Integration that defines the business model through which Correos generates recurring revenue with Amazon. Three message formats handled in parallel (XML, EDI/Edifact and proprietary), where a mapping error means packages lost at national scale.
2014/2015, Public sector, smart city
IBM MAXIMO, IOC, SSO, PKI
Chief Integration Architect for the City of Madrid. All external interfaces, hybrid cloud architecture, security (SSO, non-repudiation, encryption) and integration with MAXIMO and IOC for 3 million people.
2016, Utilities, IoT
DataPower, ODM, IIB, IoT
Real-time analytics on the UK water grid. Digital, analog and weather sensors feed an ODM rules engine that detects incidents before they happen. Also delivered: unique household identification system via numeric identifier, operational backbone of the service.
2018, Blockchain, banking
Smart contracts, distributed node
International collaboration with the team behind NXT and Ardor on reference blockchain infrastructure. Node architecture and proofs of concept for European banks from Lugano.
2025, SMB, AI agents
Agents, orchestration, APIs
AI agents running the complete operation of a music academy. From 13 hours to 2 hours of daily manual management. Agents with state, real integration and observability.
2026, Hospitality, AI agents
Conversational agent, persona, bookings DB, Telegram
Conversational agent on the website of a Madrid restaurant group: its own persona, bookings written straight to the venue's database and daily and weekly reports to management over Telegram. In production since 2026, with the model as the replaceable part and the integration as the fixed one.
“Alberto is a great professional with very broad knowledge of all integration products. His experience in diverse projects makes giving you a solution straightforward.”
Emilio Fernández Lojo
Commercial Support, Correos
“Broad expertise in his fields and a notable willingness to share knowledge and collaborate in any situation. His work is carried out with the highest professionalism.”
Iván Sánchez Maganto
ISA System Architect, NTT DATA Europe & Latam
“One of the people I have been most comfortable working with. Not only for his high technical ability, but for his proactivity and great capacity to absorb new knowledge.”
Víctor García Junco
Integration Architect, IBM
“Alberto is one of the best WebSphere specialists I have ever worked with. Polyvalent and an excellent colleague.”
David Aycart
Chief Client Executive, Baoss
15 years of Bitcoin: lights and shadows
El País, 2022
Literal words. Published with explicit written consent.
“I went from working 13 hours a day running the academy to 2. Everything that used to take up my headspace (schedules, emails, payments, follow-ups) is now handled by the agents Alberto designed. I've got back the part that brought me here: music and students. He builds systems that work.”
Previously worked with
The ten questions that surface almost every time before booking.
The price is a deliberate filter. If €500 for an hour of technical judgment feels expensive, we're probably not aligned, and that saves both of us time. Whoever books shows up prepared, and the session is actually useful.
A consultancy drops in a junior team with a methodology, a partner who signs the conclusions and six months of billing. I give you one hour with the person who designed the systems they reuse. If you want implementation afterwards, we'll talk separately.
If within the first 10 minutes you decide this isn't what you were looking for, full refund. No questions. It has never happened, but the commitment stands.
Yes. If we're working on sensitive information, I sign an NDA before the session. Send your company's template or we use a neutral one.
Both. I've done training and consulting in the UK, Switzerland, Germany, Singapore and Silicon Valley. The session runs in whichever language you prefer.
If after the session you want a longer engagement, we scope it separately. I never sell the hour to then sell something else: the hour is the hour.
The booking form asks for context, the specific problem and what you've tried. I show up read in. The hour is for deciding, not for me to catch up.
Yes. A good portion of clients are CEOs under pressure to "do something with AI" who need technical judgment before committing budget. I speak your language.
Whichever makes sense today. Models change every quarter; what stays is the architecture around them. I design so you can swap the model without rebuilding the system.
Transparency obligations have been in force since August 2026 and the high-risk ones arrive in December 2027. If your agent talks to customers or makes decisions about people, we cover it in the session.
60 minutes with the technical judgment of someone who has been building these systems since 2002. You come in with the problem and leave with a defensible direction.