Key concepts at a glance

Overview

SmixAI has two small models worth learning up front: how it connects to a system, and how it turns a problem into a fix. Everything else builds on these.

The connection model: System → Environment → Credential

SmixAI organizes every connection in three levels:

• System — a named instance you own, e.g. "Acme Salesforce".
• Environment — a specific deployment inside that system: production, sandbox, developer, staging, etc. Scans run per environment.
• Credential — the encrypted token/keys that grant access to an environment. One is marked primary.

This lets you, for example, scan a sandbox safely before touching production, each with its own credentials.

The value flow: Finding → Recommendation → Agent → ROI

• Finding — something SmixAI discovered (e.g. "storage at 94% of limit"). Each has a severity, a confidence score, and an evidence list.
• Recommendation — the proposed fix, often with an estimated ROI.
• Agent — an automated fix you can deploy to the right runtime (preview).
• ROI / Org health — the rolled-up value of what you've found and fixed, and an A–F grade.

Specifications

Term	One-line meaning
Severity	How urgent: low · medium · high · critical
Confidence	How sure the rule is (0–1; e.g. 0.9 = rock-solid)
Evidence	The data points behind a finding
Status (finding)	open · triaged · dismissed · resolved
Status (recommendation)	proposed · accepted · rejected · implemented
Deployment target	Where a fix runs: Agentforce · LangGraph · (more planned)
Role	viewer · member · admin · owner

Use cases

• Multiple Salesforce orgs: model each as its own System, with production and sandbox Environments under it.
• Safe rollout: validate a fix in a sandbox environment before applying it anywhere else.

Next steps

• Full definitions: Glossary.
• See the model in action: How discovery works.