01Shared business context
Customer, project, account, document, employee and workflow context is assembled from authorised sources before an agent acts.
SN Brain · Intelligence & Orchestration Layer
Business intelligence that is not tied to one AI model.
SN Brain is the model-independent intelligence layer inside SN Insight. It combines shared business context, multiple AI models, specialised agents, deterministic rules, controlled memory and human approval into one governed operating pattern.
Model-independentRoute work to the right AI capability
Context-awareReuse approved business context across agents
ControlledRules, approvals and audit remain outside prompts
BUSINESS CONTEXTERP · CRM · Accounts · Documents · Drawings · Workforce
SN BRAIN
Shared Intelligence Layer
Context · Routing · Agents · Rules · Memory · Evaluation
MODEL GATEWAYModel AModel BSpecialised AIFallback
AGENT COORDINATIONCollectionsConstructionFinanceWorkforce
CONTROL LAYERRules → Approval → Execute → Audit
01Understand the businessShared context before model prompts.
02Choose the intelligenceRoute by task, policy and evaluation.
03Coordinate the workAgents collaborate around defined jobs.
04Keep control outside AIRules, approval and execution stay governed.
Architecture
The layer between business systems and AI models.
SN Brain is designed so the business does not have to rebuild its workflows whenever a model changes. Context, policy and execution boundaries remain stable while the model layer can evolve.
02SN Brain intelligence layer
Normalises context, selects tools and agents, coordinates reasoning steps, applies memory boundaries and evaluates outputs.
03Multi-AI gateway
Requests can be routed to different configured models or specialised services based on task capability, policy, cost, latency and evaluation results.
04Deterministic control layer
Permissions, thresholds, business rules and approvals are checked outside model output before any sensitive production action is allowed.
Core capabilities
Eight responsibilities inside SN Brain.
Each responsibility is separated so intelligence can evolve without weakening business control.
Model-independent intelligence layer
Business workflows depend on SN Brain contracts and context—not on one model vendor or one prompt format.
Shared business context
Approved context can be reused across agents so collections, finance, construction and workforce work from the same operating picture.
Multi-AI gateway
Route generation, verification, extraction or specialised tasks to the configured AI capability best suited to the job.
Agent coordination
Specialised agents can hand work to one another through explicit tasks, inputs, outputs and approval checkpoints.
Deterministic business rules
Thresholds, permissions, due dates, approval counts and posting conditions remain code- or configuration-driven.
Evaluation & observability
Track request paths, model selection, latency, cost, verification outcomes, exceptions and quality evaluation results.
Controlled memory
Keep only scoped, approved context for the right user, tenant, workflow or task instead of uncontrolled conversational memory.
Approval, execution & audit trail
Recommendations remain separate from production actions. Approved tools execute, and the source, decision and outcome are recorded.
Controlled reasoning path
One request. A visible path from context to action.
The model can help understand and recommend. The enterprise control path determines what is allowed to happen next.
01ContextRetrieve authorised business facts
→
02PlanSelect agent, model and tools
→
03ReasonGenerate findings or proposed work
→
04VerifyEvaluate output or use a second AI
→
05RulesApply deterministic policies
→
06ApproveHuman or policy checkpoint
→
07ExecuteUse approved production tools
→
08AuditRecord source, decision and outcome
Business examples
The same brain pattern across different domains.
SN Brain reuses the same orchestration and control concepts while each domain keeps its own data, rules and specialised agents.
Which payments need attention today?
Combine ledger context, due dates, previous follow-up and customer history, then prepare the next reviewed collection action.
Receivables→Priority→Draft follow-up→Approval
What changed, and what does it affect?
Compare drawing or BOQ revisions, surface quantity and cost impact, identify affected packages and route the change for review.
Revision→QTO impact→Cost→Review
Can this transaction proceed?
Read voucher and account context, evaluate policy, check limits and approvals, then execute only through an authorised finance tool.
Voucher→Rules→Approval→Post
What needs HR or manager attention?
Bring together attendance, leave, payroll, performance or employee context and surface exceptions without bypassing HR policy.
Employee→Exception→Recommendation→Manager
Remember only what the workflow is allowed to remember.
SN Brain separates useful business context from unrestricted conversational memory. Memory can be scoped by tenant, user, customer, project, workflow and retention policy.
- Tenant and permission boundaries
- Task-specific and workflow-specific context
- Explicit source references where available
- Configurable retention and deletion policy
Know what the AI did, which model was used and how it performed.
Observability is designed around the full orchestration path, not only the final answer.
- Model and route selected
- Latency, usage and operational cost
- Verification and evaluation result
- Exceptions, approvals and execution outcome
SN Brain
Models can change. Your business logic should not have to.
Use SN Brain as the intelligence and orchestration layer that keeps shared context, agent coordination and enterprise control stable while the AI model layer evolves.