Source layer
Workbooks, CSV, JSON, Parquet, documents, databases, APIs, and connected systems enter as approved project context.
ONE GOVERNED CORE ENGINE
Source-neutral by design. Controlled by architecture.
XLpilot separates source access, understanding, governance, execution, validation, and delivery—so the operating boundary stays visible from first input to final artifact.
ONE GOVERNED CORE ENGINELIVE MODEL
ExcelCSVDocumentsDatabasesAPIsCloud
XLpilot Core EngineUnderstand → Govern → Validate
01Plan
02Approve
03Run
04Validate
RELEASED OUTPUTWorkbook · Report · Dashboard · Evidence
ARCHITECTURE LAYERS
Every boundary has a defined responsibility.
The system is designed around controlled transitions—not a single opaque prompt-to-output jump.
Understanding layer
Structure, relationships, revisions, semantics, and material uncertainty are profiled before execution.
Governance layer
Plans, permissions, policy boundaries, model routing, and user approvals define what may run.
Execution layer
Bounded operations run locally, in configured cloud environments, or through visible hybrid routing.
Validation layer
Reconciliation, formula safety, data quality, lineage, limitations, and reopenability are tested.
Artifact layer
Workbooks, reports, dashboards, and evidence are released as professional, reviewable work.
CONTROLLED ROUTING
Use the right boundary for the work.
Deployment and model selection remain explicit, policy-aware, and visible to the user.
Keep sensitive work on-device.
Approved local models and local data paths support zero-cloud-egress workflows.
Route by operation and policy.
Combine local and configured cloud execution without hiding where each step runs.
Use configured provider capacity.
Operate within approved provider boundaries using user-controlled credentials.
See the governed execution path in action.
Follow a workflow from source context to validated artifact.