Does Nexus replace any of these systems?

No. Each system keeps its own records. Nexus runs the process between them and records each decision.

Does the copilot assume weather or market feeds are already available?

No; any driver must be supplied through an approved source, and missing inputs remain visible rather than being fabricated by the model.

Does the language model itself have to generate the numerical forecast?

No; the design selects and evaluates a suitable predictive model, while a language model can support explanation and planner interaction.

Critical infrastructure use case · Snowflake + OpenAI + SAP

Load forecasting copilot

Forecast revisions are difficult to challenge when the input snapshot, assumptions and planner interventions are scattered across tools. Energy planners receive a reviewable load forecast with stated inputs, changed drivers and a recorded final planning decision.

Discuss this use case Critical infrastructure overview

Systems involved

What each system contributes

Snowflake

35 actions

Tables6
Views6
Warehouse3

OpenAI

25 actions

Chat Completions8

SAP

97 actions

Core SAP7
Products10

The model step runs against the AI provider you choose, including Anthropic, Google Gemini or a model you host yourself. Anything outside a pre-built action runs through an API, protocol or database call, or as Rules + Code in Python, Bash, Go, Java, .NET, PHP, PowerShell or JavaScript/TypeScript, with the same approvals and audit.

Architecture

One governed run

Trigger: schedule / planner releaseIntegration: Snowflake + SAPRules + CodeAI step: selected forecasting modelHuman ApprovalSystem Action: SnowflakeAudit

How it starts: A load-planning cycle opens or the planner approves a revised input set for forecasting.

Step by step

How the process runs

01
Assemble the forecast basis
Read the approved Snowflake load history and the permitted SAP planning context for the defined assets, horizon and units.
02
Check the input window
Rules validate timestamps, measurement units, data completeness and the availability of any approved external drivers.
03
Generate the load proposal
The selected forecasting model estimates the agreed horizon and retains the input and model version behind the result.
04
Explain the revision
The model-assisted workflow compares the new proposal with the previous version and identifies supported input or driver changes, leaving uncertain causes explicit.
05
Capture the planner's decision
The planner accepts or overrides the proposal with a reason before a separate step publishes the approved version to the configured Snowflake or SAP planning interface.

Control

Human decision and exception path

Human decision

The named energy planner approves the planning version before publication. Dispatch, procurement, infrastructure control or any financial commitment requires its own authorized process; a forecast proposal does not actuate equipment or move money.

Exception path

Missing measurements, timezone inconsistencies or unavailable drivers hold the affected forecast scope.
Unexplained jumps or outputs outside configured checks require planner review rather than automatic release.
A revised input snapshot creates a new proposal without overwriting the prior approved forecast or override history.

Evidence

What gets audited

Snowflake and SAP input references, units and forecast horizon.
Model version, driver inputs and validation results.
Original forecast, change explanation and planner override reason.
Approved planning version and publication responses.

Nexus runs on premises, in private or sovereign cloud, or air-gapped. Credentials stay in the vault, roles decide who can approve, and the execution record stays inside your deployment boundary.

Sovereignty and deployment

Snowflake-SAP integration
Integration pattern · SAP data reaches Snowflake as validated, repeatable loads with a record of completeness.
Snowflake connector
35 pre-built actions
OpenAI connector
25 pre-built actions
SAP connector
97 pre-built actions
Enterprise automation
Governed processes across the systems you already run.
Infrastructure alert to incident and remediation
Another critical infrastructure use case.

FAQ

Questions about this use case

Does Nexus replace any of these systems?: No. Each system keeps its own records. Nexus runs the process between them and records each decision.
How does the run start?: A load-planning cycle opens or the planner approves a revised input set for forecasting.
What if a step is not covered by a pre-built action?: The model step runs against the AI provider you choose, including Anthropic, Google Gemini or a model you host yourself. Anything outside a pre-built action runs through an API, protocol or database call, or as Rules + Code in Python, Bash, Go, Java, .NET, PHP, PowerShell or JavaScript/TypeScript, with the same approvals and audit.
Does the copilot assume weather or market feeds are already available?: No; any driver must be supplied through an approved source, and missing inputs remain visible rather than being fabricated by the model.
Does the language model itself have to generate the numerical forecast?: No; the design selects and evaluates a suitable predictive model, while a language model can support explanation and planner interaction.

Use case review

Discuss this use case

Bring your version of the process. We will map the systems, approvals and evidence it needs.

Discuss this use case Book a demo