No. 001 — Spring 2026
Thesis Loop Work Team Contact
Delft, NL — 52°N An applied AI lab for consumer commerce Vol. I

Stop hiring.
Start deploying.

A modern consumer brand still requires a team of thirty. Almost none of that work genuinely needs a human anymore. We build the systems that close the gap.

The company itself is the wrong unit. Headcount is not a moat. It is a tax.

For most of the last decade, scaling a consumer brand meant hiring. A creative team. A media buyer. A CFO. A customer service manager. A growth lead. Each function defended its existence with the language of craft.

The work those teams did was always the same loop: hypothesize, test, read, iterate, anticipate. The loop is what built the brand. The humans were the bottleneck.

That bottleneck is gone. The loop runs faster, cheaper, and more consistently when an AI stack runs it. The operator who understands the loop and the stack now has the leverage of a department of thirty.

We are building for that operator.

Operating principles
i.
We test on our own P&L first.
Every system we build runs against our own brands and our own money before any partner sees it. Theory is cheap. The market is the only audit that counts.
ii.
Skill compounds. Capital does not.
A brand can be built from ten thousand euros if the operator has the right loop and the right stack. We will not build a company that requires venture funding to function.
iii.
Headcount is a constraint, not a goal.
We will stay small on purpose. Distributed by default. No office cosplay. No nine-to-five performance art. People who think; agents that execute.
iv.
We deploy. We do not consult.
Selected partners receive working systems and the operating model that runs them. Not slides. Not workshops. Not retainers for advice that AI already gives away for free.
The loop

A team runs the loop once a quarter.
The system runs it once a second.

A team of thirty.
Then
i ii iii iv v vi
cycles run
0001
Bottlenecked
by humans
A loop that runs itself.
Now
cycles run
000000
Bottlenecked
by API quotas
Six steps: scrape, hypothesize, generate, deploy, read, decide. The loop is what builds a brand. The team was always the bottleneck.
The work
A platform, an intelligence layer, a creative engine, an import pipeline. Built and run from Delft.
Platform — Live
The autonomous commerce backend.
A continuously running system that operates the daily mechanics of a consumer brand: ads, inventory, copy, pricing, support. Currently powering brands under our own roof.
Module — Live
Product intake.
From source image to storefront listing in hours. Try-on, product photography, copy, sizing, classification, inventory. Built for operators who refuse to wait on agencies.
Module — Internal
Creative engine.
Continuous extraction of visual and narrative playbooks from the open ad market, rendered into testable creative on demand. The research department, automated.
Module — Internal
Competitive intelligence.
Continuous scraping, classification, and ranking of every relevant brand in a category. Daily briefings on what is being launched, by whom, with what claims.
The team

Three engineers from TU Delft. Robotics, aerodynamics & control, mechanical engineering. We have shipped consumer brands, advised operating ones, and written the systems that run both. We are building DLFT because no one else is building the company we want to work at.

Talk to us. contact@dlft.ai
Deployment partnerships are limited and selected. We work with operators who already know what they are doing and want the leverage to do it without the team.