You run it. / Tú lo diriges.
01
Sistemas de contenido IA
We train brand-voice language models on your existing content, style guides, and editorial history — then build production pipelines that generate, review, and publish on-brief content at scale across every channel and language you operate in.
What's included
02
Marketing de rendimiento
We rebuild paid media operations with ML-driven automation — from creative production and multi-variant testing to real-time budget allocation across all channels. The result: more spend decisions per day, fewer wasted impressions, and compounding ROAS over time.
What's included
03
Inteligencia de marketing
Most marketing teams are drowning in data they can't act on. We build unified intelligence layers — connecting your ad platforms, CRM, analytics, and data warehouse into a single source of truth, with AI-generated insights surfaced to the right people at the right time.
What's included
04
Arquitectura de crecimiento
For companies that need the full picture — we design end-to-end growth systems from acquisition through retention and expansion. We audit your current funnel, identify the highest-leverage AI interventions, and build the 12–24 month roadmap that compounds.
What's included
Every project was built from scratch — no off-the-shelf templates, no vendor lock-in. Custom AI systems designed around each client's specific business model, competitive context, and team structure.
Fashion Retail · 12 Brands · 4 Markets · Madrid
The challenge
Spain's third-largest fashion group was producing content for 12 distinct brands across Spain, Portugal, France, and Mexico — manually, with an 18-person editorial team that couldn't scale. Brand-voice consistency was degrading as the team grew. Turnaround time was 4–6 days per asset. SEO coverage was patchy across brands, and there was no way to localise content fast enough to react to seasonal trends.
What we built
12 individual brand-voice models, each fine-tuned on 6+ years of brand-specific editorial output, tone guides, and customer communications
Automated CMS pipeline: creative brief → AI generation → senior editor review → publish, integrated into their existing Contentful setup
Multilingual localisation layer covering ES, EN, FR, and PT-BR with automatic market-specific tone and cultural reference adjustment
SEO brief generator pulling from keyword research feeds, automatically creating structured briefs for each content type and brand
Performance feedback loop: engagement signals from GA4, email, and social flow back into quarterly model evaluation and retraining
We expected an efficiency gain. We didn't expect the AI to outperform our best writers on engagement metrics within 60 days. The brand-voice fidelity was the real surprise — our editorial director had to double-check who wrote what.
— CMO, Grupo Moda Norte · Madrid
Content performance at 12 months
B2B Software · Series A · Barcelona
The challenge
A Series-A SaaS company burning €80k/month on paid acquisition with a flat, declining ROAS. Their in-house team was manually managing creative rotation with no automated bidding. Creative fatigue was compounding every two weeks. The CAC had risen 40% in eight months despite no change in targeting strategy.
What we built
ML bidding automation on Meta and Google, using custom CRM signals (pipeline stage, ICP fit score) as optimisation targets rather than just ad-level conversions
Automated creative brief generation and variant production pipeline — 400+ live ad variants at any given time across all placements
Real-time budget reallocation engine that reallocates spend daily based on marginal ROAS by channel and audience segment
Salesforce integration feeding pipeline quality signals (deal velocity, contract value) back into audience targeting to optimise for revenue, not just leads
Private Equity · 7 Portfolio Companies · Valencia
The challenge
A PE holding with 7 portfolio companies across retail, hospitality, and services had zero unified marketing visibility. Each company used different tools, naming conventions, and reporting cadences. The group CMO was spending 12 hours every Friday compiling an executive summary in Excel — and it was always out of date by Monday morning.
What we built
Cross-portfolio data model normalising 14 source systems — GA4, Meta, Google Ads, HubSpot, Salesforce, and 4 custom in-house systems — into a single unified schema
Executive Looker dashboard showing consolidated group performance alongside per-company breakdowns, with drill-through to channel and campaign level
AI-generated group marketing brief auto-delivered every Monday at 7am, pulling the 8 most material developments from the prior week across all 7 companies
Anomaly detection with tiered Slack alerts: green (informational), amber (review required), red (immediate attention) based on materiality thresholds set per company
D2C E-Commerce · Outdoor & Adventure · Bilbao
The challenge
A fast-growing D2C outdoor brand had plateaued at €3.5M ARR. Acquisition channels were saturating. Repeat purchase rates were below category benchmarks. LTV was short and declining. The founding team had built a strong product and community but had no systematic approach to lifecycle marketing, and no clear picture of which customers were actually profitable.
What we built
Full marketing audit — 47 improvement opportunities identified across acquisition, conversion, and retention, each with an estimated revenue impact and implementation complexity
AI-driven email lifecycle system: 22 behavioural sequences triggered by purchase history, browsing signals, and predicted churn probability
Predictive LTV model integrated into Facebook and Google targeting, shifting acquisition budget toward audiences predicted to have 3× average 12-month value
UGC content engine: systematised collection via post-purchase flows, AI curation for quality and brand fit, and programmatic distribution to paid and organic channels
Subscription programme design and tech build, with AI-powered churn prediction triggering personalised retention interventions 14 days before predicted cancellation
They gave us a proper roadmap for the first time — not a slide deck, but an actual system. 14 months later we've nearly doubled ARR and our retention metrics are the best in our category. We're now on a retainer to keep building.
— Co-founder & CEO, Taiga Outdoor · Bilbao
Every engagement follows a structured, four-phase process engineered to move fast without sacrificing precision. We embed with your team, learn the business deeply, and deliver systems that compound over time — not black-box tools that create dependency.
Cada proyecto sigue un proceso estructurado de cuatro fases. Nos integramos con tu equipo, aprendemos el negocio a fondo y entregamos sistemas que escalan con el tiempo, no herramientas de caja negra que generan dependencia.
8 – 14 weeks
from first call to live production system
Descubrimiento y auditoría
We spend the first two weeks inside your business. We map your current marketing infrastructure, data flows, team workflows, and technology stack. We review existing content, creative assets, and attribution models. We interview key stakeholders across marketing, data, and revenue functions to understand not just the tools, but the decisions being made and the information gaps that get in the way.
Full martech stack audit and integration dependency mapping
Data quality assessment: completeness, consistency, and latency analysis
AI opportunity identification — each ranked by estimated ROI and implementation complexity
Stakeholder interviews and team capability assessment across marketing, data, and engineering
Arquitectura y diseño
We design the full system blueprint before writing a line of production code. This means selecting the right models for each task, defining data pipelines and transformation logic, structuring automation workflows, and designing the interfaces your team will actually operate. We present the architecture to your stakeholders and iterate before building anything.
Model selection rationale and architecture decision records (ADRs)
Data pipeline schemas, transformation logic, and orchestration design
Automation workflow diagrams with defined approval flows and fallback logic
Integration specs for every third-party tool in your existing stack
Construcción y entrenamiento
We build and deploy the AI systems. For content projects this means fine-tuning language models on your data and building the surrounding editorial pipeline. For performance marketing it means ML bidding logic and creative automation infrastructure. For intelligence it means data pipelines, predictive models, and dashboard layers. All built to spec, all tested in staging, all owned entirely by you.
Model fine-tuning and rigorous evaluation against held-out test sets with agreed quality benchmarks
End-to-end pipeline development with full test coverage and QA protocols
Staged environment testing, stakeholder UAT, and sign-off before production deployment
Complete documentation: architecture diagrams, runbooks, data dictionaries, model cards
Lanzamiento y optimización
Go-live isn't the end — it's the beginning. We monitor every system through the critical first weeks, tuning parameters and catching edge cases in production. We run structured team training sessions until your team can operate everything independently. We run 30/60/90-day performance reviews against the KPIs defined in Discovery. Most clients move to a lighter ongoing retainer to extend the system and keep optimising.
Monitored production launch with daily stand-ups for the first two weeks post-go-live
Team training programme: recorded walkthroughs, written playbooks, and live Q&A sessions
30/60/90-day performance review against baseline KPIs agreed during Discovery
Optional ongoing retainer: model updates, pipeline extensions, new features, and quarterly roadmap reviews
We're a deliberately small, senior team. Every person who works on your project has shipped AI systems in production — no juniors on delivery, no outsourced builds, no account managers between you and the work. You deal directly with the people who designed the architecture and wrote the code.
Somos un equipo pequeño y senior. Cada persona que trabaja en tu proyecto ha desplegado sistemas de IA en producción. Sin juniors en entrega, sin subcontratación. Trabajarás directamente con quienes diseñaron la arquitectura y escribieron el código.
No juniors on client work. Every engagement is staffed by people who have built and shipped AI systems at scale before — and who understand the commercial pressures you're under.
We work inside your tools, alongside your team. We're in your Slack, in your Notion, in your data warehouse. Not emailing decks from the outside.
Everything we build, you own completely. We document obsessively, train your team rigorously, and consider a project complete only when your team can run it without us.
We take on 3–4 new projects per quarter. If we're not the right fit, we'll tell you directly and point you toward someone who is. We'd rather lose a deal than deliver poor work.
Founder & AI Systems Lead
Criteo · Glovo · Universidad Politécnica de Madrid (MSc)
10 years building performance marketing systems at scale before founding Forja Labs. At Criteo he led the Southern Europe ML team responsible for bid optimisation across €200M+ in annual ad spend, managing a team of 12 engineers. At Glovo he was Head of Performance Marketing during the Series C–D growth period, scaling paid acquisition from €2M to €40M annual spend across 25 markets. He leads AI architecture, model selection, and technical delivery on every Forja Labs engagement and holds a master's degree in machine learning from the Universidad Politécnica de Madrid.
Head of Strategy & Client Lead
Cabify · Factorial HR · IESE Business School (MBA)
Former Head of Growth at Cabify (pre-IPO) and VP Marketing at Factorial HR, where she built the growth function from zero to $100M ARR across Spain, Germany, and the UK. She has run marketing organisations of up to 40 people and managed agency rosters of 8+ vendors simultaneously. She understands the organisational and commercial realities that determine whether AI projects succeed or fail. She leads all client Discovery and Growth Architecture engagements, serves as the primary relationship lead throughout delivery, and translates technical outputs into business language that executive teams can act on.
Lead Data & Automation Engineer
Typeform · Adevinta · Universitat Pompeu Fabra (PhD NLP)
ML engineer with a PhD in computational linguistics and natural language processing from Universitat Pompeu Fabra. His doctoral research focused on low-resource multilingual fine-tuning — directly applicable to brand-voice model training. Spent four years at Typeform building their internal ML platform and data infrastructure before moving to Adevinta, where he led data engineering for Southern European marketplaces serving 40M monthly users. He designs all data pipelines, fine-tunes the language models, and architects the automation systems that run client marketing operations end-to-end. His two open-source data tools are used by 2,000+ ML engineers.
These are the questions that come up most often in our discovery calls. If you don't find what you need, the fastest route is to book a call — we'll give you a straight answer in under 30 minutes.
Have a question that isn't here? Send it directly.
Email us →Book a free 30-minute strategy call. We'll assess your current marketing infrastructure and identify exactly where AI can have the fastest, most measurable impact — no pitch, no pressure.
Fill in the form and we'll send a calendar link within one business day with a proposed agenda. We do our homework before every call — expect specific, pointed questions about your current setup, not a generic discovery process.
What to expect on the call
Expect a calendar link from hola@forjalabs.com within one business day.