Nightshift Console
Warehouse-grade analytics for your team and your agents.
Connect any source and query it live, in a notebook or over MCP, at warehouse speed with no warehouse to run. Every read is governed by identity, so you can open your data without giving up control.
Data
select region,
sum(net) as revenue,
count(*) as orders
from orders
where status = 'paid'
group by region
order by revenue desc
| EMEA | €4.28M | 1,284 |
| AMER | €3.11M | 902 |
| APAC | €1.64M | 511 |
One notebook over the governed catalog. People run it here; agents run the same query over MCP.
A query engine your team and your agents share.
Connect your data, query it from anywhere, and open it up by identity. One console, no second system to stand up and nothing to keep in sync by hand.
Managed warehouse
A dedicated query engine on your own cloud, scaled automatically. No cluster to size, warm, or babysit. Query data where it lives in object storage, or ingest the hot paths.
Connectors and catalog
Register a bucket, a warehouse, a database, or an API and Nightshift introspects it into tables and views. One catalog everything reads, kept in sync with the source.
Notebooks
Write SQL against the catalog and get results inline, with the schema a click away. The same surface a person opens and an agent calls.
Remote MCP
Point any agent at your warehouse over MCP and the catalog shows up as tools. A scoped key decides exactly what it can see.
Data apps
Build a dashboard on the Nightshift SDK and it gets its own URL and auth, reading through the same catalog. Hosted, live, and shareable.
Access and policy
The trust layer underneath: allow, deny, or mask by row, column, and value, scoped to identity. Write it once and every read is shaped to who is asking.
The console
A home screen for everyone who touches your data.
Open the console and the day is right there: what people and agents queried, the outcomes policy shaped, your data apps, and anything that needs a person. A real workspace to sit in, not a config screen you visit once.
Reads · last 24h
3,204+8%
Shaped
142
Blocked
8
Reads by reader
| Opus 4.8 | 1.3k | ↑+12% |
| GPT-5 | 700 | ↓-3% |
| Sonnet 4.6 | 533 | ↑+8% |
| A human | 410 | ↑+6% |
Connect a source. Query it anywhere. Serve it to everything.
Your data takes one path through Nightshift, and every consumer, a person in a notebook or an agent over MCP, reads through the same live catalog.
Connect
Bring your sources into one catalog.
Register a bucket, a warehouse, a database, or a stream and Nightshift introspects it into tables and views. No pipelines to build, no second copy to keep in sync by hand.
How connect works →| sapSDK | 5.7k | Connected |
| snowflakeGuided | 14.2k | Connected |
| kafkaStream | 31.9k | Streaming |
| stripeSDK | 920 | Syncing |
Query
Warehouse-grade, without the warehouse bill.
Each query runs at the edge of your source on a dedicated engine, so you get warehouse speed with no cluster to provision and no metered per-query compute. The pattern agents hammer, thousands of iterative queries, stops being a cost problem.
How querying works →Serve
Hand any agent a live endpoint over MCP.
Point an agent at your warehouse and the catalog shows up as tools. A scoped token gives it exactly the columns it is allowed, shaped to its identity on every request.
How serving works →Usage · 24h
Build
Ship a data app on top of it.
Build on the SDK and it reads through the same catalog, with its own URL, auth, and audit. Live and shareable, with nothing leaking that policy would not allow.
See data apps →Net revenue · 24h
€4.28M
Customers
EMEA share
Avg contract
People and agents
Built for the people on your team and the agents they run.
Nightshift does not pick a side. A person opens a notebook or a data app; an agent calls the same catalog over MCP. Both go through one engine, so there is no second system to run and no gap between them.
For people
- SQL notebooks
- Explore the catalog, write queries, and read results inline. The schema is a click away.
- A shared console
- Browse data, manage access, and watch the day’s reads. Sign in with your own identity provider.
- Hosted data apps
- Turn a query into a dashboard with its own URL and auth, shareable with the team.
For agents and services
- Remote MCP
- Point any agent at your warehouse and the catalog shows up as tools, no custom integration.
- Scoped tokens
- Each agent or service gets a key that sees only the objects you grant it, nothing more.
- The catalog as tools
- Agents query the same views as people, shaped to their identity on every call.
The trust layer
Open your data up. Keep control.
Governance is what makes all of this safe to do. A grant binds an identity to a behavior, enforced on every read, so you can give people and agents real access to live data without ever handing over more than they should see.
- Allow, deny, or mask
- Shape access by row, column, and value. Hide a column from one caller and pass it to another, from the same table.
- Scoped to identity
- Every read is minted for who is asking, so the same query returns a different shape to a person, an agent, or a service.
- Monitor, then enforce
- Start a rule in monitor and watch what it would change against real traffic, then promote it to enforce when you are sure.
| Identity | orders | customers | events | arr_by_region |
|---|---|---|---|---|
| Ada OkaforOwner | rw | rw | rw | rw |
| Lena ParkAdmin | rw | rw | rw | rw |
| Marc DiazMember | r | masked | — | r |
| analytics-agentToken | r | — | — | r |
The rest of the product
Nightshift Fiber
The typed React API for building governed data apps on your catalog.
Learn more
Connectors
Register any source and expose it as a governed catalog.
Learn more
Policy
Allow, deny, and mask down to the row, scoped to identity.
Learn more
Query
Query your governed catalog from a notebook, over MCP, or in an app.
Learn more
See it with your own data.
Create a free workspace, connect a source, and query your first tables in minutes, from a notebook or straight over MCP.
Want to look first? Take the product tour