About

Research-ledAI productstudio.

We research, design, and build AI products that ship.

Est. 2019 Valencia · CET Research, design, build Accepting inquiries
About the studio

THEFT is an AI product studio based in Valencia, Spain. Founded 2019. We run three engagement shapes, UX research, AI product build, and retained partnership, and the studio carries whichever phases are scoped end to end, so findings, design, and code never get relayed to a different vendor mid-flight. The studio has shipped product for Apple, EA Games, Equinix, FedEx, Google, ManyChat, and YouTube.

7Yrs
Studio operating, enterprise product work
39Brands
11 Fortune 100, plus startups
10+Industries
Tech, finance, auto, media, logistics, travel, retail, defense, gaming, telecom
3Offerings
Research, build, retained

Whichever phases are scoped, the studio carries them. Research that holds up in session. Design that survives the build. Code that ships into production. No relay between disciplines.

SECTION.01

Origin

Seven years of research-led product work. The studio has sharpened as it has grown: fewer engagements, more depth, published scope.

  1. 2019

    Studio founded. First engagements: product discovery and design for enterprise platforms.

  2. 2021

    First generative-AI product engagement shipped.

  3. 2023

    Research and build unified under a single studio engagement model.

  4. 2024

    Compliance-aware AI build shipped inside a regulated-industry perimeter.

  5. 2025

    Retained behavioral research program begins across US and UK markets.

  6. 2026

    Studio based in Valencia. Three engagement shapes.

SECTION.02

What we do

01
Research

Behavioral research with the people who will use the product, before a line of code is written.

02
Design

Interfaces, flows, and data visualization that carry the research finding into production.

03
Build

Production code with evals in the loop from day one. The studio that ran the research ships it.

SECTION.03

How we work

01

Research and build on one engagement

No handoff, no translation layer. Architectural decisions trace back to the interviews that produced them.

02

Ship, do not deck

When the engagement includes build, the output is working software, not slideware. When the engagement is research-only, the report is decision-ready and routes straight into a build plan.

03

Evals are in the build from day one

Behavioral findings become eval cases before the model is wired in. The product is judged against them at every commit.

04

Compliance-aware from the start

Governance, audit trails, and responsible AI sit in the architecture. We have shipped inside regulated perimeters where that is the only way through.

SECTION.04

Why not a research firm or a dev agency

Research firms hand you a deck. Dev agencies hand you a codebase. Nobody owns the stretch between them, which is where AI features actually break.

THEFT
Research firm
Dev agency
Behavioral research with real users
yes
yes
no
Ships production code
yes
no
yes
Same studio for discovery and build
yes
no
no
Responsible AI practice
yes
yes
no
One studio across discovery and ship
yes
no
no
Fortune 100 engagements
yes
yes
yes

Recognition

  • Fortune 100 brands shipped to11
  • Industries shipped to10+
  • Years operating7

Among clients

  • Apple
  • Google
  • Google Cloud
  • YouTube
  • Waymo
  • Booking.com
  • Ford
  • FedEx
  • Mercedes
  • Microsoft
  • Volkswagen
  • American Express

Newengagementpleasecontact.

Office
Valencia, Spain · CET
Remote-first engagements · US + EU time zones
Availability
Accepting Q3 inquiries
About · research-led AI product studio · THEFT Studio