AI-first data governance, security & workflow

Govern your data. Then let AI work on it.

HASH builds governed data foundations for spatial and non-spatial data — and the AI-assisted workflows, security layer, privacy controls and certifiable management systems that run on top of them.

Data Lake · Catalogue
Synthetic data
CatalogueQualityLineageAccess logAI-ready
142BronzeRaw, as received
96SilverStandardised, matched
41GoldAnalysis-ready
Catalogue 8 of 279 datasets
DatasetLayerOwnerClassQuality
● Asset registerGoldEngineeringInternal
98
● Field survey recordsSilverOperationsInternal
94
● Contract ledgerGoldProcurementConfidential
96
● Service requestsSilverCitizen servicesRestricted
89
◉ Land parcelsGoldRevenueInternal
93
◉ Utility networkSilverUtilitiesInternal
91
● Sensor readingsBronzeField ITInternal
78
◉ Imagery tilesBronzeGIS cellPublic
96
● Non-spatial   ◉ Spatial
Lineage Asset condition index
Field survey app
Bronze · raw records
Silver · standardised
Gold · condition index
Dashboard · API
Completeness 99Validity 96Uniqueness 98AI-ready
GIS Data Lake · Explorer
Synthetic data
LayersQueryAnalysisReadinessExport
Tile 14 · 2026-10200 m buffer · 37 assets
Layers 5 on
Ask across layers
◆Which utility assets sit inside flood-prone zones?
37 assets  ·  5 datasets joined
Lineage verified AI-ready
Data Governance · Overview
Synthetic data
OverviewAssetsGlossaryStewardshipClassificationQualityPolicySharing
Registered data assets1,284▲ 38 this quarter
Stewardship coverage96%▲ 4 pts
Quality score92▲ 3 pts
Open issues7▼ 5 closed
Classification column level
ColumnDatasetClassTagControl
mobile_numberService requestsRestrictedPIIMasked
vendor_tax_idContract ledgerRestrictedPIITokenised
contract_valueContract ledgerConfidential—Role-based
survey_chainageAsset registerInternal—None
ward_codeLand parcelsPublic—None
Stewardship RACI
DomainStewardOwnerDPOIT
AssetsRACI
ContractsARCI
Citizen dataCARI
GeospatialRAIC
Policy library 6 policies in force
ClassificationRetentionAccessSharingIncident handlingGeospatial data
DPDP Compliance · Workspace
Synthetic data
ReadinessProcessing recordConsentRightsBreachProcessorsImpact assessments
Readiness against the Act and Rules
Notice & consent
92
Processing record
85
Rights requests
78
Breach readiness
70
Processors
64
Impact assessments
55
Rights requests time remaining
RequestStatusDue in
AccessOpen—
CorrectionIn review—
ErasureAwaiting steward—
Consent ledger tamper-evident
TimePurposeStatusHash
10:42Service updatesGranteda91f…03c2
10:37Programme surveyGranted73be…e11a
10:21Service updatesWithdrawn5d08…9a47
09:58Programme surveyGrantedc2f4…18bd
Breach response simulation
Statutory clock—
✓  Detect and contain
✓  Assess personal data affected
●  Report to the Board
○  Notify data principals
Illustrative workflow. Legal timelines are configurable and should be confirmed with counsel.
AI Security Layer · Action queue
Synthetic data
Action queueAutomationKnowledge baseLocal AIComplianceAudit
Alerts received today10,482
Incidents after grouping214
Need a person now17
Suppressed, with a reason9,436
Action queue ranked by score
#PriIncidentAction bucketScoreSLA left
1P1Repeated failed logins, then a success: possible compromiseContain & investigate107—
2P1Backup volumes show an encryption patternRespond now101—
3P2Storage array has lost redundancyDispatch repair78—
4P2Certificate expiring on a core gatewayApprove auto-fix71—
5P3Personal data found in an application logCompliance review58—
6P3Disk latency trending upwardPlan & monitor46—
7P4Routine health-check messages · 412 groupedSuppress & close22routine
Why 107? explainable
Signature priority P194
Asset criticality+6
Blast radius+3
Security flag+4
Recurrence+0
Score107 · P1
Security🔒 Locked: cannot be lowered
Local AI model
Open-weight model on your own hardware. Knowledge base first; model only for unfamiliar messages; rules fallback if unreachable.
Auto-fix approvals
Only for unlocked incidents with a scripted runbook. Requester and approver must be different people.
Audit ledger
Hash-chained record of every decision and action. Integrity verified
One foundation for
Databases & applicationsExcel & filesDocuments & scans Maps & vector layersSatellite & drone imageryLiDAR & CAD
How it fits together

From raw sources to trusted action.

Data enters once, is governed once, and is reused everywhere — by people, by dashboards and by AI.

InputsSources
DatabasesEnterprise applicationsExcel & flat filesDocuments & scansGIS layersImagery & LiDARCAD drawingsField apps & sensors
FoundationData platform
Data Lake · bronze · silver · goldGIS Data Lake · vector · raster · point cloudCatalogue & lineage
TrustData governance
Asset registerGlossaryStewardshipClassificationQuality rulesPoliciesControlled sharing
ProtectionDPDP platform
Processing recordNotice & consentRights requestsBreach responseProcessorsRetention
AssuranceSecurity & audit
AI security layerISMS · ISO/IEC 27001PIMS · 27701AIMS · 42001SMS · 20000-1Audit & evidence
ActionAI & workflow
AI assistantsHuman-in-the-loop reviewStage-gated workflowsDashboardsAPIs
Security & assurance

AI watches your estate. Standards prove it works.

An air-gapped AI security layer turns thousands of alerts into a short, ranked list of actions. Management systems for security, privacy, AI and service delivery turn good practice into evidence you can certify.

Our principles

Built so that you can trust the answer.

Governed from day one

Every dataset has an owner, a steward, a classification and a quality score before anyone builds on it.

AI with people in charge

AI reads only governed data it is permitted to use. People approve what matters, and every prompt, output and action is logged.

Open and in your control

Open data formats, no lock-in, and deployment in India — on-premises, in a sovereign cloud or in your own cloud account.

Where it applies

For organisations that run on data they must protect.

Government & public sector

Programme, procurement and citizen-service data under one governance framework.

Infrastructure & utilities

Asset, survey and network data — spatial and tabular — in one place.

Healthcare

Sensitive records that are classified, consented and shared with purpose limits.

Financial services

Customer identity, onboarding and consent workflows with a complete audit trail.

How we work

Framework first, then platform.

We study before we build, so that every rupee of technology spend is justified and defensible.

01

Study

Map systems, data flows and risks. Classify what you hold.

02

Frame

Policies, ownership and a consent framework your board can approve.

03

Build

Stand up the lake, governance and workflows, one source at a time.

04

Operate

Monitor quality, access and compliance, with evidence ready for audit.

Start with a data readiness review.

A short, vendor-neutral look at what data you hold, how it is governed, and where AI can safely help.

Talk to usSee our approach