Uncontrolled SQL generation
AI-generated queries need validation, limits, and reviewable execution paths.
AI SQL GatewaySelf-hostedValidated SQLAudit metadata
Governed AI Access to Your Data — without giving up control.
GAARD is a self-hosted AI SQL gateway for organizations that want natural-language access to relational data while keeping prompts, SQL generation, validation, execution, connectors, and auditability under governance.
Built for enterprises, healthcare teams, and public-sector organizations that need useful AI access without uncontrolled data exposure.
Self-hosted deployment
SQL validation before execution
Connector-based access
Reviewable request metadata
AI data access is useful. Ungoverned AI data access is risky.
The risk is rarely the question itself. It is the path from prompt to generated query to executed operation.
Fragmented data access
Teams need one governed way to ask questions across trusted data sources.
Low auditability
Regulated organizations need visibility into prompts, generated SQL, outputs, and operational metadata.
Sensitive operating environments
Healthcare, enterprise, and public-sector teams cannot treat data access as a black box.
Solution
GAARD puts a governed gateway between people, models, and data.
Instead of connecting every assistant directly to every database, GAARD centralizes natural-language data access behind a controlled SQL gateway. Configure data sources, models, prompts, validation rules, and execution policies in one place.
Governed NL-to-SQL
Natural-language requests move through an approved generation path instead of ad hoc database prompts.
Question -> policy -> SQL
Connector-based access
Approved connectors define which sources can be reached and how access is bounded.
Sources stay explicit
Prompt and policy control
Prompts, model choices, and generation rules live in one managed operating layer.
Policy before response
SQL validation before execution
Generated SQL is inspected before it can touch trusted production data.
Validate then execute
Execution metadata
Requests, queries, outputs, and runtime context remain reviewable after every run.
Trace every run
Self-hosted deployment
Keep the gateway, connectors, and execution surface inside your own environment.
Operate in your stack
How it works
A controlled path from natural language to governed SQL.
GAARD keeps the flow explicit: approved sources, selected models, validated SQL, and reviewable execution metadata.
1
Connect sources
Register relational databases and approved data endpoints.
2
Configure models
Choose which LLM providers and models can be used.
3
Ask questions
Users or systems ask natural-language questions instead of writing SQL manually.
4
Validate and audit
GAARD controls generated SQL, execution, responses, and traceability.
Sectors
Designed for organizations where data access must be controlled.
Enterprises
Give analysts and business teams faster access to trusted operational data while keeping SQL execution, data source access, and model usage governed.
- Executive reporting
- Operations analytics
- Internal data products
- Controlled AI-assisted analysis
Healthcare
Let authorized teams ask questions over approved datasets while keeping data access paths explicit and reviewable.
- Operational reporting
- Capacity and activity analysis
- Research support datasets
- Internal decision support
Public sector
Support transparent, controlled access to structured data for agencies and public institutions that require auditability and operational discipline.
- Administrative reporting
- Public service analytics
- Internal knowledge and data access
- Controlled AI adoption
Governance
Control the path from question to query to answer.
GAARD is built around operational control: source approval, model configuration, SQL validation, execution boundaries, and metadata that can be reviewed.
GAARD is not just a chat box over a database. It is a governed access layer for natural-language data operations.
- Configure approved data sources
- Control which LLM providers and models are available
- Keep SQL generation and execution inside your environment
- Validate generated queries before execution
- Preserve request and execution metadata
- Use self-hosted deployment patterns for stronger operational control
Pricing
Clear plans for open use, individual analysis, and governed teams.
Start with the community version today. Individual subscriptions are handled through Stripe; enterprise adoption starts with a focused pilot.
Community
Free
Start with the open community version and run GAARD with relational data sources.
- Run anywhere
- Relational data sources of any type
- SQL / Analysis mode
- Talk to one source at a time
- LLM On premises or Cloud
- Use any openai compatible model
- Use one LLM model at a time
- One human/machine user
- One dashboard
Data Analyst
PopularGood for individuals
$29 / month
Monthly subscription with a 7-day trial.
For individual analysts who need broader source access and more flexible model usage.
- Everything in Community
- Connect non SQL (like Spreadsheets or API) data sources
- Talk to multiple sources at once
- Use multiple LLM models at once
Pilot Plan
EnterpriseGuided pilot scope
GAARD does not replace BI. It complements BI dashboards and BI workflows as a governed natural-language access layer for executives and analysts.
Pilot scope
- Self-hosted deployment
- Up to 5 data sources
- 20-50 pilot users
- Model and policy configuration
- Query logs and audit trail
- Dashboard/adoption report
- Rollout plan recommendation for 50+ seats
- Everything in Data Analyst
- Extract data from unstructured notes
- Query unstructured data like SQL
- Identity management
- Multiple human accounts
- Multiple dashboards for humans
- Unlimited machine API consumers
Stripe hosts checkout, invoices, card updates, plan management, and cancellation. The website never stores card details.
Frequently asked questions
What is GAARD?
GAARD is a self-hosted AI SQL gateway that gives users natural-language access to governed data sources while keeping SQL generation, validation, execution, model usage, and auditability under control.
Is GAARD only for relational databases?
The Community plan focuses on relational data sources. Paid plans are available and expand source options, including Excel and API data sources. Enterprise adds extraction from unstructured notes so extracted information can be queried in a SQL-like way.
Does GAARD replace BI tools?
No. GAARD complements dashboards and BI workflows by providing a governed natural-language access layer for data questions, analysis, and controlled SQL execution.
Is GAARD cloud-only?
No. GAARD is designed as a self-hosted gateway, giving organizations more control over deployment, configuration, and data access paths.
Are paid plans available now?
Yes. Data Analyst and Enterprise subscriptions are available through Stripe checkout, while the Community plan remains available as the open version.
Is GAARD certified for healthcare or public-sector compliance?
Do not claim certification unless your organization has completed the relevant process. GAARD is designed for governed, auditable data-access patterns, but compliance depends on deployment, configuration, policies, and the customer environment.
Get started
Put a governed gateway between AI questions and operational data.
Start with the open community version today, or choose a paid subscription from the pricing section when you need broader access.