The data platform that answers back

Metrix is a fully managed data and AI platform. Load every source, query it at any scale, and let built-in agents tell you what changed.
Built for teams like these
Stripe

Easy. Fast. Answers built in.

Metrix runs the infrastructure, so your team runs the questions. Storage, compute and AI agents share one engine and one set of rules.
AI in the engine

Metrix Agents run next to your data, not on a copy of it. Ask why revenue dipped and an agent breaks the change down by segment, region and release, then shows the SQL behind every step.

Fast at any scale

Storage and compute scale separately, so a petabyte query and a dashboard refresh never fight for the same machines. Billing is per second, and every heavy query shows its cost before it runs.

One copy of your data

Stream, batch or federate every source into one governed copy. Metric definitions live next to the data, so every chart, alert and agent answers from the same number.

Everything your data needs, built in

Ingest anything
Stream, batch or change-data-capture from 200+ sources into one table. No pipelines to babysit.
Governance built in
Policies follow the data, not the tool. Row-level access, masking and audit logs for people and agents.
Machine learning in SQL
Train, forecast and detect anomalies with a SQL function. No exports, no separate stack.
Vector search
Store embeddings next to your tables and search by meaning, not keywords.
Live data sharing
Share live tables with partners and teams without copying a row. They query, you stay in control.
Geospatial analytics
Map any metric by country, region or store, with native geography types and functions.

The latest from Metrix

Engineering notes, benchmarks and product updates from the team that builds the platform.
Picking an anomaly model

Seasonality, sparse data and sudden launches each break a different detector. How our ML functions pick a model per metric, and why the default is the boring one.

One customer, two definitions

Sales and Product both define “customer.” How a shared metric layer inside the platform settles it without another meeting.

Separating storage from compute

Why we bill storage and compute apart, what it does to cost at 10 TB and at 1 PB, and where per-second billing pays off.

Five fixes for slow queries

Most slow queries share five causes, from scans that miss the partition key to joins that fan out. Before-and-after plans for each.

Lineage from source to agent

When an agent answers a question, every number traces back to a column. How we built lineage that covers people and agents alike.

Sharing data without copying it

Live data sharing between companies sounds risky. How policies travel with a shared table, and what the receiving side can and can’t see.

Leading teams run on Metrix

How teams use Metrix every day.
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab
“We moved 400 TB and twelve years of orders onto Metrix in a quarter. On a Sunday, an agent flagged a checkout drop, narrowed it to one payment method in two markets, and put the query in the thread before on-call opened it.”
Sarah Chen
Head of Data, Shopfront
“Our dbt models came over as they were, and the numbers matched Finance from week one. Query costs went down 38% because heavy jobs now show their price before they run.”
Marcus Rodriguez
Analytics Engineering Lead, TechFlow
“We don’t have a data team yet. Our one engineer connected six sources in an afternoon, and now 40 people ask Metrix questions in plain English. When a number looks off, the SQL is right there.”
Henry Lubowitz
Founder, GrowthLab

Under the hood

Built for the questions your security team asks first.
  • Separate storage and compute
    Storage grows on its own, compute spins up per query and stops when it’s done. You can see and cap every second of it.
  • Governed by default
    Row-level policies, masking and lineage apply to people and agents alike. Your data never trains a model, and every query is logged.
  • Always on
    Cross-region replication, managed failover and point-in-time recovery. Every number shows when its source last loaded.

Put your data to work

This site never went live

Metrix was designed and built, then shelved before launch. What’s left is the page itself, from the type scale to the motion.

Designed & Developed by Bram van der Laar

Built with
See my work
Email me