An AI-ready governance layer inside Snowflake

The practical

Airlock communicates and enforces the rules for data moving into and out of Snowflake. Apps, agents, and humans prepare and propose work through Airlock. Accepted submissions become governed records in Snowflake. Validated files can be prepared automatically for downstream tables and views, or enter workflows for further review, aggregation, and real-world business processes.

The profound

Airlock lets you move governance and system-of-record duties out of individual apps and into Snowflake, so AI can safely participate in real business work. When apps do not have to be the authority, they can be bespoke, ephemeral tools built for the exact task and person at hand. This is why Airlock also works well with Excel, the original flexible, ungoverned work surface. We call this an agent-oriented business architecture: governance and system-of-record duties at the center, so teams can move quickly, experiment, adapt at the edges, and deploy AI without fragmenting truth.

Use cases

Airlock is for real business workflows where humans, agents, and bespoke apps need to participate without becoming another place where infrastructure cost and complexity soar, and where truth must be reconciled.

AI adopter workflows

Snowflake intake and operations

🤖 Governed OpenClaw

OpenClaw-style agents can gather context, prepare submissions, attach evidence, and call documented Airlock procedures. In a reimbursement workflow, an employee can send a receipt to a named agent in Slack; the agent prepares the work, Airlock validates it, finance reviews it, and approved reimbursements can be batched for payment.

Read the reimbursement case study Back to use cases

🛠️ Agent-built apps

A team can build a focused internal app for a workflow, use it for as long as it is useful, and submit governed outputs through Airlock. The app stays lightweight. Snowflake keeps the durable record.

Back to use cases

📎 Governed file intake

Users and agents can attach PDFs, images, documents, receipts, and other evidence to governed submissions instead of passing files through email, shared drives, or ad hoc buckets. Airlock applies attachment policies, workflow state, permissions, and audit history while the files remain inside Snowflake-backed storage.

Back to use cases

🧩 Data gaps

Almost every business has data that does not fit neatly into an existing enterprise system. Airlock gives people and agents a governed way to submit that missing context into Snowflake without creating another unofficial system of record.

Back to use cases

🕸️ Data mesh

Domains can own the work they produce while central teams define the rules for what becomes official. Airlock lets local teams and agents submit governed outputs into Snowflake with centralized visibility, validation, and audit history.

Back to use cases

🔄 Reverse ETL

Prepared, approved outputs can be made available from Snowflake to the people, teams, and agents that need them. Airlock keeps the submission and governance path controlled before those outputs are reused downstream.

Back to use cases

🧰 Pipeline fallback

When an automated process is delayed, incomplete, or missing a one-off input, Airlock gives teams a governed manual or agent-assisted path to keep work moving while preserving validation, permissions, and history.

Back to use cases

And many more.

How Airlock works

Airlock sits inside your Snowflake environment as the governed boundary for AI agents, OpenClaw workflows, humans, scripts, and user-built apps.

Those tools can reason, draft, transform, calculate, collect evidence, and prepare work wherever they run. Airlock controls what they can read, what they can submit, which rules apply, and what becomes official in Snowflake.

Snowflake keeps the record: official records, shared context, accepted submissions, and historical outputs stay centralized.
Airlock governs the boundary: rules, permissions, validations, workflow state, and submission procedures are enforced in one place.
Agents, apps, and humans do the work: the cognition layer stays flexible, task-specific, and fast.

Key features

Airlock carries forward the useful Reseed pattern: local teams handle delivery while central rules govern admission. Airlock extends that pattern to AI agents, bespoke apps, and multi-step business workflows.

Governed submissions, not broad table access: agents and apps submit through documented procedure surfaces; Airlock validates schemas, files, attachments, permissions, and workflow rules before work becomes official.
Specs over one-size-fits-all apps: define the contract for reimbursements, budget requests, proposals, leads, or operational records, then let teams meet it with OpenClaw, agents, spreadsheets, scripts, or focused apps.
Workflow states that lock the record: submitted work can move into states where users and agents can no longer alter or delete it, preserving current state and the history needed for audit and recovery.
Governance between agents: one agent can prepare a submission while another reviews, reconciles, or carries processing forward against the governed Snowflake context you choose to expose.
Shared context without data sprawl: reference data and accepted outputs stay in Snowflake, so agents do not have to turn private files, prompts, or proprietary runtimes into business memory.
Snowflake-native operations: Airlock runs inside Snowflake, reducing separate auth flows, integration boundaries, data pipelines, cloud accounts, and infrastructure for teams to secure and maintain.

The result is distributed execution with centralized business truth: people and agents can build the tool that fits the moment, while Airlock governs what becomes official.

Who it is for

Teams bringing human-submitted files into Snowflake who want Excel, CSV, PDFs, images, documents, and attachments governed at the point of intake
Data teams tired of manual intake work who want non-technical users to get immediate data-quality feedback without waiting for an engineer to inspect a file downstream
Executives shaping an AI strategy around Snowflake, governed experimentation, fewer SaaS sync surfaces, and high-agency AI-focused builders driving outcomes
Leaders hiring AI-fluent builders who move fast, ship purpose-built tools, and still need enterprise-grade storage and policy
Teams integrating agents (including OpenClaw-ready workflows) against a stable, documented procedure surface
BI and SQL users who prefer predictable CALL shapes over ad hoc table access
Admins configuring roles, specs, templates, references, retention, and licensing via airlock.admin.*

Documentation

The Airlock docs include the full Stored Procedure Reference (admin and user schemas, compatibility rules, and common workflow examples) plus the Spec Library for reusable workflow patterns. Builders and agents can also use the public airlock-specs and airlock-tools repos.

Examples assume the installed app object is named airlock; if an account uses a different app name, substitute that name.

In Snowflake, the installed app can return the bundled contract as UTF-8 markdown (one line per row):

CALL airlock.user.documentation();

View Airlock on the Snowflake Marketplace

$9.99 per user (or agent) / month