N.01
The meeting is where the system fails in public.
Lost proxy votes were a symptom. Information, questions and debate all converged on one evening, so that is where it broke.
24 questions answered before the room
I’m Álvaro Velasco. Twenty years designing products across games, regulated platforms, consumer AI and B2B SaaS. I look for the system underneath the interface problem, then I help build the fix.
Agenda · 19 items
Proposal to renew the pool’s filtration and tiling before next summer, funded from the reserve fund. Votes are weighted by each owner’s coefficient.
Questions · 7
Pre-votes so far
Owners can change their vote until the item closes.
Before the meeting
75
owners have already voted
Languages
Doors open · 23 July
Hover the product. The surface is what people see. The lens shows what decides.

20M+
players on the multi-brand system I designed at GVC
GVC · Entain · 2016 → 2018
127
countries where Fretello was App of the Day
Fretello · WWDC19
65x
monthly contributions after the incentive redesign
SellerCrowd · 2021 → now
10,975
owners on vecinOS, across 64 communities
vecinOS · 2026 → now
01 · Selected work
Hover a project.Scroll through the projects. You’ll see the brief I was handed, then the problem that was actually controlling the outcome.

The brief
Stop losing proxy votes at the annual meeting.
The real problem
The meeting was where the governance system failed in public. Move it upstream.

The brief
Fix the information architecture so people contribute more.
The real problem
Contribution was an incentive problem, not an interface problem.

The brief
Build one component library for about fifteen brands.
The real problem
Every brand had reasons to stay different. Adoption was the design problem.

The brief
Ship the AI that listens to you play.
The real problem
Proving the tech works is not proving the interaction premise works.

The brief
Make a portfolio.
The real problem
Decide which conversations deserve my time, and answer the rest for me.

The brief
Align a product team around what to build.
The real problem
Long specs let teams agree without aligning. Four lines force a commitment.

The brief
Pick the better of two prototypes.
The real problem
Both prototypes scored well. The answer was to combine them: above 75% on every task.
02 · Field notes from vecinOS02 · Field notes · vecinOS
vecinOS started because proxy votes got lost. It became a lesson in governance, trust and legitimacy. Figures from the July 2026 annual meeting of a 165-owner community, in four languages.
3h 26m
annual meeting, down from 9 hours the year before
60.2%
of the coefficient decided before the doors opened (best ever with a full room: 48%)
24
questions answered in advance, in four languages
Same community, same kind of agenda. What changed was everything that happened before anyone entered the room.
75 owners voted before the meeting and 24 questions were answered in advance, in four languages.
N.01
Lost proxy votes were a symptom. Information, questions and debate all converged on one evening, so that is where it broke.
24 questions answered before the room
N.02
Owners who couldn’t travel still read, listened, asked and voted in their own language before the doors opened.
60.2% of the coefficient · best with a full room: 48%
N.03
Understanding what Yes, No and Abstain actually do makes voting slower. The goal is a legitimate decision, not a conversion.
Votes can change until the item closes
N.04
My minutes engine assumed the agenda runs in order. Real meetings revisit items, correct records and reinterpret the rules mid-debate.
The model had to tolerate reality breaking it
N.05
A trusted president’s message did more for adoption than any feature. The best products amplify human systems instead of replacing them.
76 of 76 voted to keep using vecinOS
N.06
The assistant answers from the statutes and the law. Anything it does can be done by hand, and judgment calls go to the administrator.
50 assistant conversations · judgment stays human
Now live: 64 communities · 72 meetings · 10,975 owners
Read the vecinOS case03 · Judgment
Twenty years of accumulated judgment, not uninterrupted victories. Here is one where I was wrong, step by step.
Assumption · SellerCrowd · 2021 → now
Fix the architecture and activity will follow.
SellerCrowd had grown organically and the product lacked structure. It looked like an interface problem, so I treated it like one.
04 · How I work now
AI shortened the distance between my judgment and working software. The judgment still does the work.
Every action an assistant can take stays doable by hand, and judgment calls go to a human. In vecinOS, the assistant sits next to the manual path, never instead of it.
Agents kept inventing CSS values instead of using tokens. More documentation didn’t fix it. Locked primitives, a screen scaffold and CI gates did.
A second model reviews every change adversarially. Then I review every screen by hand before anything reaches production.
05 · Partners on Álvaro
He’s the rare designer who diagnoses the business problem first, the kind of partner any founder needs.
What I remember most, though, is his workshops. They never ended with vague alignment or nice ideas. They ended with real decisions and clear commitments from the team.
06 · The route
Architecture, games, regulated gaming at FTSE 100 scale, consumer AI, B2B SaaS, and now my own company. Hover the line.
Founder
vecinOS
Governance software for homeowner communities. 64 communities, 72 meetings, 10,975 owners so far.


Graphic design at Abalos & Herreros, then video games and freelance work. Everything you learn when nobody’s watching.
Regulated gaming at scale. Reusable components before they had a name, specs shipped as PDFs.
Built an 8-person UX team and one system for about 15 brands. The cashier redesign lifted deposit value 4x in tested markets.
An AI guitar coach. Its Sign in with Apple sign-up redesign put it on the WWDC19 stage. App of the Day in 127 countries.
Design partner to the CEO. 360 → 23,770 monthly contributions and 3x MRR on flat headcount.Team up
I’m the designer you put on it when you’re not sure you’re solving the right thing. A short call is usually enough to find out.