Limosaidentity

SOLUTION ARCHITECTURE

Connected systems.
Explicit boundaries.
Traceable outcomes.

See how AI assistants and the web workspace use the same governed operations to connect inventory, decisions and durable execution.

AI ACCESS ARCHITECTUREA full interface. The same governance.
YOU + YOUR ASSISTANT

Describe the work

A compatible AI assistant discovers available operations and acts on your instructions.

AUTHENTICATED MCP CONNECTION

Connect with consent

Your tenant and current permissions determine what the assistant can do.

SHARED APPLICATION API

The same operations as the web workspace

ValidationPermissionsTenant isolationAudit attribution
Configure

Connections, mappings, roles and workflows

Operate

Requests, reviews, tasks and fulfillment

Investigate

Inventory, reports and operation history

Conceptual architecture. Bulk work runs through background jobs; approvals, execution results and observed access remain separate.

01 / SYSTEM CONTEXTA connected governance architecture
Authoritative sourcesHR records · identity attributes
Connected applicationsAccounts · entitlements · memberships
PeopleEmployees · approvers · administrators
↓ Import and correlate inventory · Submit requests and decisions ↓
LIMOSA · TENANT CONTEXT & SERVER-SIDE PERMISSIONS
Identity & inventoryMappings, correlation and ownership
Policy & decisionsRoles, requests and certifications
Execution & evidenceFulfillment, histories and reporting
↕ Durable state and background work
PostgreSQLTenant-scoped records · Row-level security · Durable job queue
↓ Configured operations, with tracked results ↓
Application connectorsRead inventory and execute supported writes
Manual fulfillmentAssigned work for operations that need people
Isolated transformsSeparate runtime for customer JavaScript

Logical architecture. Connections and execution capabilities depend on configuration; this is not a deployment topology.

A MODEL THAT PRESERVES CONTEXT

Know what every status
actually means.

Identity governance spans several systems and several kinds of truth. Limosa Identity keeps those records connected while preserving what each one establishes.

02 / THE EVIDENCE MODELFour questions. Four distinct records.
01

Intended access

What should this person have?

Role assignments and configured policies define the desired state.

02

Approval decision

Who authorized the change?

Requests and review decisions capture the human or policy decision.

03

Execution result

What action was attempted?

Provisioning operations and manual tasks record progress and outcomes.

04

Observed access

What does the application report?

Aggregation and read-back provide evidence of the target state.

A recorded approval does not establish that provisioning succeeded. Compare the execution result with observed access.

BUILT FOR OPERATIONAL WORK

Let people decide.
Let durable jobs execute.

Aggregation imports accounts and entitlements. Identity refresh evaluates identity data and configured policy effects. Provisioning attempts changes in target systems. Each has its own trigger, progress and history.

Background processing keeps bulk operations outside page requests. Operators can monitor work in Tasks and investigate individual execution outcomes when an application is unavailable or a change needs manual completion.

Custom JavaScript runs in the separate transform runtime with execution limits. Connector access uses configured credentials and approved secret references.

Explore security boundaries →
03 / EXECUTION BOUNDARIESInteractive work and background processing
Web workspace, AI assistants & API clients

People and authorized AI clients search, configure, approve and review through authenticated interfaces.

Application API

Validate requests, enforce permissions and persist tenant-scoped decisions.

PostgreSQL-backed jobs → Workers

Process aggregation, refresh and provisioning outside the interactive request.

Connectors · Isolated transform runner

Call configured target operations and evaluate customer JavaScript within its separate execution boundary.

Conceptual processing path. Worker progress and operation histories remain separate from the original user decision.

FROM THE MODEL TO THE WORKSPACE

Operational visibility in context.

The administration dashboard brings together current inventory and governance activity. It gives administrators a starting point for investigation, while task and object histories retain the underlying detail.

Example dashboards contain seeded activity. Your own deployment’s records and metric collection determine the values you see.

Administration dashboard in the populated John Doe example workspace
Administration dashboard · Actual product interface with synthetic data. Select to enlarge.

Plan an implementation around your environment.

Start with sources

Identify authoritative attributes, account schemas and matching rules. Decide how ambiguous or unmatched accounts will be resolved.

Define responsibility

Establish role ownership, requester eligibility, approval paths and review responsibility before enabling governance workflows.

Validate execution

Test supported connector operations, credential access, transforms and manual fulfillment. Confirm the target result before expanding scope.

FULL AI SUPPORT

Your workflow.
Your AI assistant.

Operate and configure Limosa through an MCP-compatible AI assistant. The full management interface is available to AI, with the same permissions, validation and audit trail as the web workspace.

Start with https://identity.limosa.io/mcp. No existing Limosa account needed: sign in and create your workspace when prompted.

Connect your assistant ↗

YOUR NEXT STEP

See governance in practice.

Explore the public demo or bring your applications and governance questions to a guided conversation.