☁
Managed cloud
A managed Sprouti environment with organisation and project isolation, encrypted storage and region-aware deployment. Suitable for teams that want lower operational overhead.
Security and private GeoAI
Your data should not become the price of using AI.
Sprouti is being built to provide controlled GeoAI for organisations with sensitive, proprietary or regulated spatial information. Deploy it where your data lives, choose which models can be used and retain control over storage, access and deletion.
No model trainingCustomer content is not used to train shared models
Local optionKeep data and inference within your environment
Customer controlChoose storage, models, access and retention
AuditableTrace plans, tools, execution and outputs
Deployment choice
Run Sprouti within the boundary that fits your risk.
The intended architecture supports different levels of operational control rather than forcing every customer into a single shared AI service.
▤
Private cloud or VPC
Deploy services into a dedicated or customer-controlled cloud environment with private networking, restricted integrations and organisation-owned data stores.
⌂
On-premises or local
Run data services, workflow execution and compatible AI models within local infrastructure or a restricted network so sensitive information does not need to leave the organisation.
Not trained on your data. Not sold. Not silently retained.
- Customer data, prompts and outputs are not used to train shared Sprouti models
- Customer content is not sold or used for advertising
- Local deployments can keep operational data entirely inside the customer environment
- Retention is customer-configurable and should be limited to what is required
- Deletion and export controls are part of the intended data lifecycle
- Third-party AI providers must be explicitly approved and configurable
Processing is not the same as training.
Some deployment modes may temporarily process prompts, metadata or selected records to complete an approved task. That content should not be added to a shared training corpus. In customer-hosted and local modes, processing can remain within infrastructure controlled by the organisation.
Exact retention, backup and logging behaviour will be documented per deployment and contract. Sprouti will avoid claiming “zero retention” where infrastructure requires explicitly agreed operational records.
Security controls
Defence across identity, data, models and execution.
Identity and access
Role-based access control, least privilege, project membership, optional single sign-on and separation of administrative responsibilities.
Encryption
TLS for data in transit, encryption at rest, managed key options and protected handling of secrets and credentials.
Tenant and project isolation
Logical or dedicated isolation between organisations, projects, data stores, jobs and generated artefacts.
Sandboxed computation
Restricted execution environments, CPU and memory limits, controlled dependencies, timeouts and network policies for Python and analytical jobs.
Approval-led AI
Separate planning from execution. Review selected datasets, tools, SQL, code and parameters before a workflow runs.
Audit and provenance
Record relevant access, plans, workflow versions, tool use, execution status and output lineage for investigation and reproducibility.
Model governance
Allow-list models and endpoints, restrict external providers, support local models and control what context each model can receive.
Data minimisation
Use schema, metadata, summaries and database-side operations where possible rather than transmitting complete datasets to a language model.
Operational resilience
Backups, recovery procedures, monitoring, vulnerability management, dependency updates and incident-response processes appropriate to each deployment.
Secure pilot partners
Help validate private GeoAI in a real operating environment.
We are seeking organisations with clear data-governance requirements to shape deployment, security and assurance priorities during development.