Why We Invested in Querio

Why We Invested in Querio‍

Every company runs on business logic: how revenue is recognized, which customers count as active, how a supply chain KPI is actually calculated. Almost none of that logic lives in one place. It sits in dashboards, in notebooks, in transformation models, and in the heads of the analysts who wrote the original queries. Ask three teams for the same number and you will often get three answers.

That fragmentation was tolerable when humans were the only consumers of data. It breaks down when AI agents become a primary interface to it. An agent writing SQL against a warehouse without the company's definitions produces answers that are fluent, confident, and frequently wrong. The bottleneck has shifted from access to data toward access to the context that makes data mean something.

The modern data stack has a system of record for raw data in the warehouse and a system of record for transformations in tools like dbt. It has no system of record for business logic that both humans and agents can create, reuse, and govern. Querio is building that layer.

We're excited to back Rami Abi Habib, Javier Bonilla, and the Querio team as they build the context layer for company data.

A Codebase for Data Logic

Querio's design is intentionally simple at its foundation. Business logic is stored as files (SQL, Python, YAML) in Git and retrieved through a combination of deterministic and semantic search. Humans work with it through a notebook environment and plain-language exploration. Agents work with it through APIs, an MCP server, and embedded use cases. Both operate on the same repository, the way engineers operate on a shared codebase.

The important difference from traditional semantic layers is where the logic comes from. Querio does not ask a data team to model the business upfront. Logic is generated as a byproduct of everyday analysis, persists as versioned context, and is reused the next time anyone, human or agent, asks a related question. Over time the repository becomes the system of record for how the company defines its own numbers.

Because the context lives in files rather than a proprietary UI, it is portable across warehouses, auditable in version control, and agnostic to which model is doing the reasoning. Those three properties turned out to matter a great deal to the customers we spoke with.

Why Now

Three forces are converging.

Architectural shift. The agent ecosystem is settling on a pattern where context is an external, governed layer that models retrieve rather than knowledge baked into weights or hidden behind a single vendor's API. Agent harness architectures, MCP, and file-based tool access all point the same direction. A context layer built as a Git repository fits that pattern natively instead of fighting it.

Market timing. AI is becoming the primary interface to company data, and the people asking questions are increasingly outside the data team. That only works if the definitions behind the answers are shared, governed, and reusable. Data teams are moving from writing every query to curating the context that lets everyone else write queries safely.

Competitive gap. Incumbent BI and notebook tools store business logic inside proprietary UI layers, which ties a company's definitions to a single platform. Customers told us plainly that they do not want to lock their business logic into one more vendor, whether a data platform or a model provider. A warehouse-agnostic, model-agnostic, file-based layer is the opposite of that lock-in.

Querio is building for this gap: a context layer that compounds with use, travels with the company, and serves humans and agents from the same source of truth.

Built by Founders Who Lived the Problem

Our conviction starts with the team.

Rami Abi Habib (CEO) spent his career at Amazon using SQL and Python to drive roughly $150M in supply chain impact. He watched business logic fracture at scale: the same metric defined three ways across three teams, with no shared place to reconcile them. That is the problem Querio exists to solve.

Javier Bonilla (CTO) built and operated data-critical systems at Glovo and Friday Finance, environments where a wrong definition was not an inconvenience but an operating failure. He has shipped production systems where accuracy and consistency of metrics were the product.

Both have repeatedly been the person responsible for debugging and standardizing metrics across an organization. They also demonstrated something we weight heavily at seed: the ability to change course. Querio launched in 2023 as a SQL-only product for product managers and found the ceiling quickly. In mid-2025 the team rebuilt the product around a full notebook environment and a dedicated context layer, shifted the customer to data teams, and relaunched in November 2025. The current trajectory followed directly from that decision.

Our Conviction

We invested because Querio is addressing a structural gap in the data stack with an architecture that aligns with where AI systems are heading, and because the early evidence of compounding is real.

In mature accounts, roughly 75% of queries now draw on context that already exists in the repository. Context creation is heavily front-loaded in the first month and tapers as the repository matures, which is exactly the shape you would expect from a system of record forming. Customers describe connecting a warehouse in minutes and reaching first outputs in under an hour, and the majority of Querio's users today sit outside the data team. The ROI is concrete: analyst time reclaimed, decisions unblocked for business users, and hiring avoided.

As Kasper Sage, Managing Partner at BMW i Ventures, put it: "The data stack has a system of record for raw data and for transformations, but not for business logic. That gap was not as apparent when analysts were the only ones writing queries. It now matters enormously in the age of AI as agents are driving a lot of the requests. Querio is providing that missing layer, and building it as open files rather than another proprietary silo."

The defensibility is cumulative. Every analysis adds logic to the repository, every reused definition makes the next answer more reliable, and the more of a company's business logic lives in Querio, the harder it becomes to recreate elsewhere. The enterprise foundation is already in place with SOC 2 Type II, GDPR and CCPA compliance, and IAM-style permissions, with SSO, self-hosting, and deeper governance on the roadmap in response to direct enterprise demand.

What's Next

We're backing Querio as the data stack reorganizes around AI as its primary interface. The companies that own the context layer will set the terms for how agents interact with enterprise data.

We're proud to partner with Rami, Javier, and the Querio team as they build the system of record for business logic.

About Querio: Querio is a London-based data infrastructure company building a code-native context layer for company data, where humans and AI agents create, reuse, and govern shared business logic. Learn more at querio.ai.

About BMW i Ventures: BMW i Ventures is a corporate venture fund backed by BMW AG, investing in early-stage companies at the intersection of AI infrastructure, agentic systems, physical AI, and industrial automation.

‍