Bad data costs the global economy an estimated $3 trillion every year. That’s more than the GDP of France. And as companies rush to embed generative AI into customer service, product recommendations, forecasting, and pricing, the risk of making decisions on flawed data is rising fast.
Sifflet, a startup based in New York, has raised $18 million to address that problem. Its AI-native data observability platform is designed to ensure the reliability of data pipelines before they cause real damage. Founded in 2021 and launched in 2023, Sifflet has since tripled both its customer base and its revenue. Its clients now include Penguin Random House, Euronext, and Saint-Gobain.
Pete Williams, Chief Data Officer at Penguin Random House, says the company uses Sifflet “to ensure data reliability across the business, not just within engineering.”
Sifflet’s model challenges how most organizations think about data quality. Instead of treating it purely as an engineering problem, Sifflet says their making it a shared responsibility. It delivers tooling for both technical and non-technical teams, aiming to embed data reliability into decision-making processes across the enterprise.
No AI Without Good Data
The company’s CEO and co-founder, Salma Bakouk, argues that businesses can’t unlock the value of AI until they first address the integrity of their data. Along with co-founders Wissem and Wajdi Fathallah, she built Sifflet to provide both visibility and control, without requiring teams to rewrite their workflows.
The platform includes standard capabilities like anomaly detection and data lineage, but also introduces AI agents that assist with incident response and resolution. Sentinel monitors metadata and recommends what to watch. Sage traces the origins of problems using lineage and query history. Forge suggests fixes based on past resolutions.
These features aim to reduce manual triage and alert fatigue, two growing problems as companies scale their data infrastructure. The goal is not just to surface problems but to provide usable paths to resolution.
Investors backing the $18 million round include EQT Ventures, Mangrove Capital Partners, and Capmont Technology. The raise brings Sifflet’s total funding to nearly $36 million.
Yannick Oswald of Mangrove Capital Partners says, “Backing a company once is a bet. Backing them three times is conviction. [Sifflet is] redefining how data-driven organizations operate.”
The company is now expanding its presence in North America and investing in product development. It has also brought on two new executives: Rémi Bastien, formerly of Content Square, and Romain Doutriaux, formerly of Pigment, to lead operations and sales strategy.
Built for Accessibility
Sifflet is often compared to Monte Carlo, a longer-standing player in the observability space. While both platforms offer anomaly detection and monitoring, Monte Carlo is built primarily for engineers. Sifflet focuses on making observability accessible across analysts, data scientists, operations teams, and executives. The platform includes a natural-language interface, no-code monitor creation, and tiered alerting based on user role.
As AI systems move further into production and data volumes continue to grow, companies are facing a scaling challenge that observability alone won’t solve. They need systems that can adapt to complexity and deliver answers without delay.
Sifflet sees this as a turning point. Data observability is becoming part of core business infrastructure. The company’s roadmap includes autonomous capabilities that go beyond alerting and analysis to active optimization. The aim is to create a system that not only surfaces problems but prevents them.
AI, according to Bakouk, is only as good as the data behind it. Sifflet wants to make sure that data is usable, trustworthy, and ready to drive real decisions.
