Protect your data and professional assets separately
Connecting to the platform does not transfer your rights to data, core processes, recipes or your own models. Before adoption, agree with us what may be read, who can access it, whether data leaves the site and what happens when use ends.
Data flows, authorized people, external services and exit arrangements are included in the proposal for your team to review.
Using Be Water Applications gives you the agreed usage rights, without transferring source code or proprietary models. Models, files and results provided by you and your experts also follow their respective licences for calls, deployment and exports.
Choose where data is processed and who can receive it
Self-service diagnosis is not open yet.It will not upload production data to Be Water automatically. Offline mode will process exported files on your computer; customer-environment mode will connect only to authorized data sources and models. You separately choose what to share for reports or remote support.
| Mode | Reading and computation | External transmission |
|---|---|---|
| Fully offline file diagnosis | Selected exports, local rules, calculations and models on your computer | Data stays on the computer; no intranet or internet access |
| Customer-environment diagnosis | Only authorized data sources and models | Only approved endpoints; hosting providers and processing regions are disclosed |
| Approved external-model diagnosis | Full records and specialist calculations stay on the customer side; only approved content is sent | Provider, content, purpose, region and retention are shown before authorized transmission |
You can choose not to use external AI. If you select an external model, you will first see the provider, endpoint, processing region, actual content to be sent, purpose and retention arrangements. Nothing outside the authorized scope will be sent.
You can stop further reading and calls, export the agreed configurations, reports and records, and check how copies are handled. Your data and models remain under your control; Be Water and partner models and software follow their licences. Data already sent follows the recipient's retention and deletion arrangements.
Give specialist computation and AI distinct roles
Use the appropriate models and programs for defined calculations, rules and operating constraints. AI interprets tasks, organizes information and uses authorized tools. A natural-language instruction must not bypass interlocks, data permissions or established operating limits.
Return execution results, exceptions, and model or configuration versions to the same task so customers can review the evidence, handle exceptions and manage ongoing operation.
Grant data and tool access for the task
Permissions define which assets and records users can see, which tools they can call and which actions require approval. An assistant investigating exceptions may read operating records, while dispatch and control require separate authorization. Update access and action records when people, assets or tasks change.
For the selected deployment, confirm identity integration, permitted model endpoints and the data those endpoints may receive. Under the target specifications, external model access can be disabled, MCP calls stay within the caller’s authority, and physical writes require an explicit allowlist and separate authorization. Customer material is not automatically used for general model training.
Attribute capability use and agree contributor rewards To be launched
Agree the license, attribution, maintenance responsibility and compensation before using a contributed capability. Usage evidence should let contributors check the recorded use against settlement. The partnership guidance explains the contributor’s concerns, the protection to agree and how discrepancies are reviewed.
The planned blockchain records would make jointly confirmed usage and settlement entries harder for any party, including Be Water, to alter quietly. They cover selected, authorized references rather than public copies of customer data or proprietary models. Deployment scope must define the recording and access arrangements; the shared recording capability has not yet launched.
Application or runtime evidence is still needed to establish actual usage; a shared entry alone does not establish a valid calculation or customer savings. Operating outcomes use site evidence and an agreed comparison. Licenses and access controls continue to govern source material, derivatives, training use and permitted exports.
For your deployment
Who may view data, change configuration, approve actions and carry out site work?
How may customer data, professional knowledge, site configuration and new deliverables be used and handed over?