Agent-native data platform

The data plane that agents cannot talk their way around.

A language model gives you an answer. ObeliskDB makes it defensible — governed, reproducible, budgeted, and safe to write back — with the guarantees enforced below the model.

DuckDB kernel · versioned storage · Snowflake-dialect SQL · MCP-native

The inversion

Agents are the query authors now. The platform's job changed.

When people wrote the queries, the platform helped them express intent. When agents write them, the job is to make an autonomous, tireless, occasionally wrong author trustworthy. That trust cannot live in a prompt.

Reasoning plane · advisory
  • system prompt
  • permission dialog
  • context / rules file
  • “please be careful”
Truth plane · enforced
  • policy rewrite
  • credit budget
  • copy-on-write branch
  • immutable receipt

The same rule is ignorable in the prompt and non-negotiable in the engine. ObeliskDB puts it in the engine.

The enforcement boundary

Reasoning is interchangeable. The guarantees are fixed.

Bring any model — Obi, an external coding agent over MCP, whatever comes next. Each inherits the same contract, because it is enforced beneath the model and cannot be opted out of. A masked column's bytes never reach the model. A spent budget refuses the next call. A write lands on a branch until a human merges.

Six guarantees

What the engine enforces, so the agent doesn't have to be trusted.

MEANING

Governed metrics

One canonical definition per quantity. An agent selects it by name and cannot misdefine revenue.

ISOLATION

Write staging

Every agent write auto-stages on a zero-copy branch. A human reviews the row-level diff and merges. The pull request, for data.

RECEIPTS

Time travel

Every answer replayable and pinned to the exact data version it was computed from. Reproducible forever.

ECONOMICS

Per-agent budgets

Credits charged at execution and refused when spent. A runaway loop cannot spend your month.

MEMORY

Insights

Stored answers with freshness computed from the version chain — current, stale, or invalidated — and governed on read.

CONFIDENTIALITY

Masking & thresholds

Policies applied inside the query rewrite, so filters see masked values and small-cohort probes are refused.

Meet Obi

Your guide to the monolith.

Ask questions, request changes, review what agents did, approve what ships. Obi runs on the Anthropic API or a local model — and everything it does is governed, budgeted, and staged for your review. SQL is there when you want it, never required.

Obi
Raise the APAC target to 120.
Updated APAC to 120. The change is staged on an isolated branch — nothing touches live data until you approve it.
✓ staged · branch OBELISK__AGENT_4EC3D6 · review in the console
why is surgery's readmission highest?
Surgery leads at 4.92%, driven by orthopedic revisions. I answered through the governed metric, so the number matches the dashboard.
The argument, in full

The Enforcement Boundary

A vendor-neutral position paper on why the guarantees an agent economy depends on must live in the data plane — with a reproducible evaluation and the market reason the incumbents are structurally disinclined to build it.

0 of 14 values leaked under 12 adversarial extraction attempts
8 of 8 injected privileged operations refused, whatever the data said
Both callers — chat and external agent — held the same contract
Reproducible harness · vendor-neutral · ~5,500 words

Bring the agents. The monolith stands.

ObeliskDB is the fixed, authoritative thing beneath your agents — the source of truth they orient by and cannot move.