gHawk AnalyticsPublic demonstration
Use your own data
GARNET HAWK SOLUTIONS

Your data and AI

Where data comes from, where it goes and who can use it.

AI expands what your team can do.
Your people retain the judgment.

gHawk does the calculations. AI helps your team explore the results. You can check the data and challenge an explanation before making a decision.

Know the origin of every finding.

Each public collection identifies its publishers, original feeds, licences, reference dates, retrieval time and join method. Missing values stay missing. If a refresh fails, the last verified snapshot is labelled as saved data.

Data versions identify the records used. Charts and briefings carry their calculation and filters. The public demo combines Canadian economic, mobility and weather datasets; publication frequency varies by source.

Public exploration. Private uploads.

The public interface has no file upload endpoint accessible to anonymous users. A verified sign-in and workspace acceptance are required before CSV or JSON files can be saved. Every dataset, saved briefing and connection is restricted to its owner.

Files are limited to 10 MB, 10,000 records and 80 columns. Uploads are parsed as data, never executed. You can delete active files and their saved briefings. Platform and provider backups follow their respective retention terms.

See what AI receives.

Before saving a file, you can ask an approved AI connection to review field names, types and quality counts. No row values are sent in that review. Connection advice uses only the description you choose to send. New saved datasets have ongoing AI disabled. You must approve a tested connection for each dataset. A request sends column definitions and your question, followed by bounded statistics and up to 12 group labels. Source files and raw record arrays are not sent. Aggregates, labels and questions can still contain sensitive information.

The processing receipt shows those inputs, the provider, exact model and effort. AI drafts can be wrong; review their evidence. The calculation engine performs arithmetic and validates every suggested view.

Choose the right deployment.

Files and workspace records are stored on the gHawk server. The hosting region has not been confirmed for this preview. AI processing, logs and backups need their own location and retention agreement. Access is granted after review. Signing in does not send your files to AI.

Client deployments can be designed around customer-hosted models, approved private AI providers or cloud APIs. Network boundaries, storage region, inference location, logs, backups, support access and contractual commitments must all be agreed before sensitive data is connected.

Respect First Nations data governance.

Garnet Hawk is rooted in Garden River First Nation. That relationship does not confer authority over another Nation’s data. The First Nations principles of OCAP®—ownership, control, access and possession—require governance determined by the appropriate First Nation, including its rules for collection, use, sharing, access and custody.

This demo makes no blanket claim of OCAP® compliance. Nation-governed data needs a community-approved purpose, permissions, stewardship arrangements and deployment before use. Indigenous data governance is not one uniform framework.

Read the First Nations Information Governance Centre’s guidance. OCAP® is a registered trademark of the First Nations Information Governance Centre.

Be precise about provider terms.

OpenAI API data is not used for model training by default, but retention, abuse monitoring and residency depend on the service and account configuration. Other providers have their own contracts and controls. A private endpoint alone does not establish compliance or eliminate provider access.

gHawk requests disabled response storage on its native OpenAI connection. This is not a promise of zero retention. API keys are encrypted with a server-held key, connections are tested, custom hosts require administrator approval and there is no silent fallback to another provider.

OpenAI API data controls · Anthropic retention guidance

Before you bring sensitive data

Use suitable evaluation data in this review environment. A production engagement should agree data ownership, approved users, hosting and inference regions, retention and deletion, provider terms, incident handling and any Nation-specific requirements. Signing up is an access control, not a certification of security or regulatory compliance.

Discuss your deployment with Garnet Hawk