Velum Labs
Product & Competitive Intelligence
Automated privacy-first data quality OS with ML anomaly detection and encryption.
Company Overview
Builds an automated, privacy-first data quality operating system that continuously monitors, diagnoses, and remediates data issues across any enterprise data stack using ML-driven anomaly detection, automated root cause analysis, and ontology-powered data contracts.
Competitive Advantage & Moat
Product Roadmap & Public Announcements
Velum Labs has publicly described building ontology-powered data contracts, a content-level firewall for granular access control, and the "Ontology Engine for Enterprise AI" as the missing semantic infrastructure layer between raw data and AI. Their website positions the platform as automatically deriving contracts from real data traffic, tracing lineage across any stack, and enforcing integrity from ingestion to executive dashboards with no manual rules or migration required. They are seeking design partners.
Signals & Private Analysis
GitHub activity on their Thor conversational LLM framework (Rust/Go) and Kit function-calling toolkit signals investment in agentic AI orchestration and developer tooling. The YC profile lists them as YC F25 (Fall 2025) batch, though they appear in W26 materials as well. Benjamin Muñoz-Cerro's Stanford quantum computing and Harvard physics background and Alen Rubilar-Muñoz's ML engineering and mathematics expertise point toward next-gen encrypted compute capabilities. Strong indicators of a compliance-first GTM targeting finance, healthcare, and legal verticals where data trust commands premium pricing.
Product Roadmap Priorities
Automated ML-driven monitoring that continuously detects data drift, schema changes, null spikes, and silent data loss across enterprise pipelines in real time.
It's like having a tireless security guard watching every data pipeline 24/7, instantly flagging anything that looks wrong before it causes damage.
It's like replacing a smoke detector that only beeps when the house is already on fire with an AI nose that smells the wiring getting warm.
Automated root cause tracing and remediation engine that uses ML-powered lineage analysis to pinpoint the source of data quality issues and deploy fixes through existing workflows.
Instead of engineers spending hours tracing a data bug through dozens of tables, the system instantly finds the broken link and fixes it.
It's like having a plumber who not only finds the exact leaky pipe in your walls using X-ray vision but also patches it and installs a sensor so it never leaks again.
Privacy-preserving machine learning pipeline that enables data quality monitoring, anomaly detection, and analytics to run directly on encrypted data using fully homomorphic encryption (FHE), eliminating the need to decrypt sensitive information.
Your data stays locked in a vault the entire time, but the AI can still read it through the walls and tell you if something's wrong.
It's like a doctor diagnosing your X-ray while it's still sealed in a lead envelope — they give you the results, but they never actually see the image.
Company Overview
Key Team Members
- Benjamin Muñoz-Cerro, Co-Founder & CEO
- Alen Rubilar-Muñoz, Co-Founder & CTO
Founder Force Multiplier
Velum Labs combines Benjamin Muñoz-Cerro's rare expertise in quantum computing (Stanford) and physics (Harvard) with Alen Rubilar-Muñoz's ML engineering and mathematics background, allowing them to offer automated data quality with ontology-powered semantic understanding, a capability that positions them as the infrastructure layer between raw data and AI that most competitors lack.
Funding History
- 2025 | Benjamin Muñoz-Cerro and Alen Rubilar-Muñoz co-found Velum Labs.
- 2025 | Open-source Thor LLM framework and Kit toolkit released on GitHub.
- 2025-2026 | Accepted into Y Combinator (F25/W26 batch).
Competitors
- Data Observability: Monte Carlo, Bigeye, Anomalo, Metaplane (monitoring-focused).
- Data Quality Platforms: Ataccama, Talend, Informatica (legacy enterprise).
- dbt-native Testing: Elementary Data, dbt built-in tests (developer-focused).
- AI-Native Data Quality: Soda.io, Validio, Great Expectations (open-source).
- Privacy-Preserving Compute: Duality Technologies, Enveil, Zama (FHE-focused, not data quality).
