I founded sifamo GmbH in 2019 with two people, €15,000 in fixed costs in the first month and not a single euro of outside capital. Seven years later we are 28 people in Berlin and Frankfurt, with €2.5–2.8m in revenue in 2026. Here are the numbers, the decisions behind them and what I do differently as a founder.
Revenue 2019 to 2026
in €m per year, 2026 as forecast. FTE = employees at year end.
Team 2019 to 2026
employees at year end: consultants, architects and engineers across both offices.
The timeline: seven years in numbers and decisions
- 2015–18Before founding
From developer to freelancer
Master of Science in Computer Science (2015), then a permanent job. After eight months I quit; the first contract of my own comes through an agency at an insurer in Hanover. Two years as a freelance developer and architect, until it is clear this carries more than one person.
- 2019Founding€0.3m2 employees
sifamo GmbH, self-funded
Founded with my co-founder Ayman Kalaie, registered in Berlin. €15,000 in fixed costs in the first month, no loan, no investor. The first year closes with €0.3m in revenue.
- 2020€0.4m3 employees
Growing through the pandemic year
Projects continue remotely and the company pays for itself from its own contracts. Every euro of profit stays in the company, which becomes the rule for the next seven years.
- 2021€0.5m4 employees
Enterprise architecture for regulated clients
The focus shifts to banks, insurers and public administration: large platforms where every decision has to be traceable. The foundation for everything that later comes with AI.
- 2022€0.8m7 employees
Doubling pace, then ChatGPT arrives
The team grows from four to seven, revenue by 60 percent. In November ChatGPT is released and suddenly everyone is an AI expert. We decide against the hype and for rebuilding our own company.
- 2023€1.2m10 employees
The first million and the rebuild around AI
Two years of rebuilding begin: SIFAMO itself is switched to AI, from recruiting to internal processes to consulting. Since then AI in regulated environments has been our core business.
- 2024€1.6m13 employees
AI goes into production
Our AI-assisted recruiting recognises hard skills from CVs with around 90 percent precision. For clients, the first systems with registry, logging and human in the loop go live: the core of the later governance layer.
- 2025€1.8m18 employees
SIFAMO AI Summit at the Burj Khalifa
In February 2025 we host the SIFAMO AI Summit at the Burj Khalifa in Dubai with a team of 15, our biggest stage so far. The ERP works with an AI board, internal workflows are AI-assisted, and I am on stage more often, including at the Business Forum in Frankfurt. The team grows to 18.
- 2026Forecast€2.5–2.8m28 employees
The Gulf, Paris, seven-figure proposals
Business trips to Dubai, Doha, Abu Dhabi and Paris: investments assessed, AI partnerships initiated. Over €1m in personnel costs and €150,000 in monthly fixed costs, a seven-figure proposal for a security platform. 28 people across two offices.
- 2027Outlook
SIFAMO AI Summit in Doha
The next summit is being planned with Visit Qatar, with an investment of over €1m. The goal remains: the first port of call for AI governance in Germany and Europe.
What I do differently as a founder
For seven years I did not pay myself a profit. Liquidity matters more than my own pocket, because it is what makes growth without outside capital possible.
We rebuilt SIFAMO around AI ourselves before recommending it to clients. Whoever has transformed their own company knows where the pitfalls are.
Those who only stay on the safe path never grow. Full but considered risk: our own summit at the Burj Khalifa with 15 people, Gulf trips in uncertain times.
We never start with the model but with registry, boundaries, escalation and human in the loop. Only then come the use cases. That survives audits.
Technical founder: what I have built myself
I come from architecture, not from sales. Master of Science in Computer Science, more than ten years of enterprise systems for some of the largest platforms in Germany. This is the part of my work that still interests me most:
- Governance layer for AI in regulated environments: every model registers, every step is logged, boundaries and escalations are defined, and decisions remain traceable for BaFin, DORA and MaRisk.
- Training without personal data: pseudonymisation with IDs across three data-protection zones, so models learn without ever having seen a real name.
- LLM, RAG and agent architectures inside existing banking and insurance landscapes, embedded rather than bolted on.
- AI recruiting that recognises hard skills from engineering CVs with around 90 percent precision, and an ERP with an AI board for our own operations.
The question that drives me: how do capable AI systems stay controllable in production when they influence decisions about credit, job applications or insurance? I have written up the architecture in detail: The governance layer that keeps AI controllable in production. More under AI Governance and EU AI Act.
For keynotes, consulting or founder questions: contact, contact@sifamo.com or LinkedIn. The stories behind the numbers are in the playlist Unternehmertum & Gründung on YouTube (German).
More: AI Expert · AI Speaker · AI Consultant · AI Governance · EU AI Act