About

A venture builder that writes its own software.

LucidMaze starts companies rather than backing them. We take no client work and sell no consulting: everything we make, we own and operate.

What a venture builder is

A venture capital firm picks companies. An agency builds what a client asks for. A venture builder does neither: it starts companies itself, with its own people and its own money, and keeps running them.

That distinction decides almost everything about how we spend a week. There is no pipeline to source and no pitch to win. There is a portfolio of live products with real users, real invoices and real outages, and the work is making each of them better than it was.

It also means we are exposed to our own judgement in a way an investor is not. If a product is badly designed, we do not lose a position — we spend the next year maintaining it.

In practice

Founded
Portugal, remote-first
Portfolio
10 companies
Client work
None
Outside capital
None raised to date
Infrastructure
Self-hosted, on our own network

The model

How a venture gets built here.

01

A problem we actually have

Every venture here started as an obstacle in our own way of working. We needed to reach servers with no public address, so we built the network. We needed an agent that could be trusted to work unattended, so we built the limits first. Starting from a problem you own is the cheapest form of market research there is.

02

Build the smallest real version

Not a prototype and not a pitch — a working product, deployed, used daily by us before anyone else is asked to care. The goal at this stage is to find out what the problem actually is, which is almost never what it looked like at the start.

03

Operate it

Then the long part: running it. Reliability, billing that reconciles, support, the unglamorous edges. This is where a venture builder differs most from an investor — we do not hand the company to someone else at this point, we staff it and keep going.

What connects the portfolio

On the surface these are unrelated businesses: a networking product, an agent platform, a growth tool and a chess site do not obviously belong in the same holding company.

What they share is a technical centre of gravity. All of them are infrastructure problems wearing different clothes — moving traffic to a machine that cannot be addressed, constraining a process that runs without supervision, deciding what is worth surfacing out of a firehose, settling money in a way that can be audited afterwards. The domain changes; the engineering does not.

They also share a network. Every one of these products is served to the internet through offcloud, which is both convenient and a useful discipline: our own uptime depends on a product we sell.

See the portfolio