Guru · An operating system for business

The AI Operating System
for Business.

Describe the outcome. Guru understands the objective, plans the work, activates the right agents and systems, builds the required interface and executes through governed controls.

Build a merchant onboarding platform.
Interactive product concept·Not connected to a live backend
Your business no longer operates application by application. It operates through intent.
The Guru sequence

From intent to outcome.

A single progression runs through every request — from natural-language goal to governed execution.

  1. 01

    Understand

    Interpret intent in context.

    Guru interprets natural-language objectives using user identity, organisational context and authorised company knowledge.

  2. 02

    Plan

    Decompose the work.

    Guru breaks the objective into steps, identifies dependencies, selects the right agents and determines required approvals.

  3. 03

    Build

    Assemble what is needed.

    Guru generates the required workflow, workspace, application, service or integration on demand.

  4. 04

    Connect

    Work across the estate.

    Guru operates across authorised enterprise systems, data sources, tools and external agents.

  5. 05

    Execute

    Deliver the outcome.

    Guru completes the workflow within configured permissions, financial limits and human-approval controls.

The seven roles

One Platform. Seven Transformative Roles.

Seven capabilities emerging from a single shared architecture — not seven disconnected products.

01

AI Operating System

The coordinating layer across people, agents, applications, data, workflows and commerce.

  • Natural-language command layer
  • Multi-step planning and orchestration
  • Tool selection and approval management
  • Execution monitoring and audit trail
02

Company Brain

Governed institutional context, not a document chatbot.

  • Knowledge graph and document intelligence
  • Source attribution and verification status
  • Access permissions and data lineage
  • Conflict detection and retention controls
03

AI-Native Service

Not another tool to operate. A service that performs the work.

  • Describe the outcome, review the plan
  • Approve where required
  • Receive the completed result
  • Configurable autonomy per task
04

Dynamic Interface

Software that forms around the work.

  • Generated dashboards and workspaces
  • Forms, tables, case-management views
  • Persistent applications and headless APIs
  • Reusable, saveable, governable
05

SaaS Consolidation

Replace the complexity, not necessarily every system.

  • Consolidate fragmented internal tooling
  • Retain specialist and regulated systems
  • One conversational and operational surface
  • Progressive replacement over time
06

Software for Agents

Built for people. Native to agents.

  • Governed tool gateway and capability registry
  • Scoped agent identity and permissions
  • Agent-to-agent collaboration
  • Execution sandbox and human escalation
07

Agentic Commerce Engine

Make every business transactable by humans and AI.

  • Sell to agents through structured catalogues
  • Buy through agents within policy
  • Tokenised credentials and spending controls
  • End-to-end audit and reconciliation
01 · AI Operating System

Turn business intent into coordinated action.

Guru interprets objectives, plans multi-step work, selects the correct capabilities, manages approvals and delivers completed outcomes across connected systems.

Example request — Review yesterday's failed settlements, identify material discrepancies, prepare investigation cases, assign them to the appropriate team and send me an executive summary.

Guru performing
  • 1
    Interpret
  • 2
    Plan
  • 3
    Delegate
  • 4
    Analyse
  • 5
    Request Approval
  • 6
    Execute
  • 7
    Report
PoliciesPeopleProductsCustomersMerchantsSuppliersSystemsWorkflowsTransactionsDecisionsContractsEventsBRAIN
02 · Company Brain

Every company has knowledge.
Guru makes it operational.

Governed institutional context — a living graph across people, policies, systems, customers, products, transactions and workflows. Every answer can display its authorised source and context, where technically available.

Knowledge graph
Source attribution
Verification status
Access permissions
Effective dates
Conflict detection
Data lineage
Retention controls
03 · AI-Native Service

Not another tool to operate. A service that performs the work.

Traditional Model
  • Open software
  • Navigate menus
  • Search data
  • Complete forms
  • Coordinate teams
  • Create the result
Guru Model
  • Describe the outcome
  • Review the plan
  • Approve where required
  • Receive completed result
Configurable autonomy
Mode 1
Answer Only
Guru responds. No systems are touched.
Mode 2
Analyse
Guru investigates and returns findings.
Mode 3
Recommend
Guru proposes actions for a human to take.
Mode 4
Prepare
Guru drafts the work; a human dispatches it.
Mode 5
Execute With Approval
Guru executes after named approvers sign off.
Mode 6
Automate Within Controls
Guru operates inside defined limits and policies.
04 · Dynamic Interface

Software that forms around the work.

Guru can design, generate and configure applications — subject to technical validation, security review and deployment approval.

Prompt · Build a merchant portfolio review workspace showing volume changes, profitability, open support cases, terminal health and risk alerts.
  1. 01Data model
  2. 02Dashboard layout
  3. 03Filters
  4. 04Workflow actions
  5. 05Permissions
  6. 06Export & reporting
  7. 07Reusable saved application
Mode
Conversational Interface
Interact through natural language.
Mode
Generated Workspace
Task-specific screens built on demand.
Mode
Persistent Application
Save, share, iterate and govern.
Mode
Headless API Service
Expose the outcome to other systems.
05 · SaaS Consolidation

Replace the complexity, not necessarily every system.

Consolidate fragmented interfaces, workflows and point solutions while continuing to integrate with specialist and mission-critical systems.

STEP 1
Connect
Operate across the existing software estate.
STEP 2
Unify
One conversational and operational interface.
STEP 3
Replace Selectively
Retire redundant dashboards, forms and internal apps.
STEP 4
Become the Operating Layer
Work through Guru; specialist systems remain underneath.
Potentially consolidated
Internal dashboardsWorkflow toolsForm buildersBasic CRM processesTask coordinationReporting toolsApproval systemsInternal portalsKnowledge searchSelected low-code apps
Continues to integrate with
Core bankingAccountingERPPayment processorsScreening providersSpecialist risk systemsRegulated infrastructureSystem-of-record databases
06 · Software for Agents

Built for people. Native to agents.

Guru exposes authorised business capabilities as structured, discoverable tools — with identity, scope and audit.

Core components
  • Capability registry
  • Governed tool gateway
  • Agent identity
  • Scoped permissions
  • A2A collaboration
  • Delegated workflows
  • Execution sandbox
  • Human escalation
  • Observability
  • Audit trail
  • Reusable skills
  • Event-driven actions
Sample multi-agent workflow
  1. 01Merchant Onboarding AgentOrchestrator
  2. 02Document AgentDelegated
  3. 03Screening AgentDelegated
  4. 04Risk AgentDelegated
  5. 05Contract AgentDelegated
  6. 06Human ApprovalOversight
  7. 07Activation AgentExecution
  8. 08Audit RecordRecord
Multi-agent orchestration

Agents coordinated. Humans in command.

Guru coordinates specialised agents through defined hand-offs — with human approval gates before execution.

Step 1
Merchant Agent
Onboarding, KYB, profile
Step 2
Risk Agent
Exposure and limits
Step 3
Compliance Agent
Sanctions, AML, policy
Step 4
Fraud Agent
Signals and interventions
Step 5
Settlement Agent
Payouts and reconciliation
Step 6
Reporting Agent
Institutional reporting
Step 7
Human Approval
Gated decisions
Step 8
Execution
Committed to systems of record
07 · Agentic Commerce Engine

Make every business transactable by humans and AI.

Structure your catalogue, controls and checkout so both human buyers and enterprise agents can transact within policy.

Sell to Agents
  • Product & service catalogue
  • Pricing & inventory
  • Availability & delivery
  • Purchase conditions
  • Checkout capabilities
Buy Through Agents
  • Research approved suppliers
  • Compare products & quotes
  • Procurement recommendations
  • Renewals & reorders
  • Purchases within policy
Operate Commerce
  • Discovery & recommendation
  • Quotation & negotiation
  • Authorisation & checkout
  • Payment & fulfilment
  • Refunds, service, loyalty
Commerce journey
Discover
Evaluate
Authorise
Pay
Fulfil
Support
Spending limits
Approved merchants
Category restrictions
Tokenised credentials
Transaction approval
Geographic limits
Expiry
Revocation
Full audit trail
Purpose restrictions

Agentic Commerce Protocol support is offered as an integration / standards-readiness option, not as verified compliance.

Platform architecture

Seven layers. One governed system.

Governance surrounds every layer — identity, access, approvals, model and data governance, security, observability, resilience and human oversight.

Layer 1

Experience

Conversation, voice, generated workspaces and embedded surfaces.

ConversationVoiceGenerated workspacesWeb & mobileEmbedded interfacesExternal AI channels
Layer 2

Intelligence

Intent, planning, reasoning, memory and decision support.

Intent understandingPlanningReasoningModel routingMemoryRetrievalPrediction
Layer 3

Company Brain

Institutional knowledge, ontology and verified sources.

Knowledge graphDocumentsPoliciesOperational dataOntologyPermissionsInstitutional memory
Layer 4

Agents

Specialised agents, orchestration, delegation and handoff.

Specialised agentsOrchestrationRegistryDelegationCollaborationEvaluationHuman handoff
Layer 5

Software Creation

Interfaces, workflows, data models and APIs generated on demand.

Interface generationWorkflow generationData-model generationAPI generationTest generationReusable componentsDeployment pipeline
Layer 6

Integrations

Standards-based connectivity across the enterprise estate.

MCPA2AAPIsEventsWebhooksDatabasesFilesThird-party platforms
Layer 7

Commerce & Payments

Catalogue, checkout, orchestration, settlement and loyalty.

CataloguesInventoryBookingsCheckoutTokenisationPayment orchestrationSettlementReconciliationRefundsLoyalty
Governance envelope
IdentityAccess controlApprovalsData governanceModel governanceSecurityObservabilityAuditabilityResilienceHuman oversight
Enterprise governance

Intelligence with institutional control.

Guru is designed for organisations where auditability, oversight and revocable authority are non-negotiable.

Human oversight
Approval gates for sensitive actions.
Role-based access
Minimum required permissions per agent and user.
Source-grounded outputs
Attribution to authorised sources where available.
Model & tool governance
Approved models and tools per capability.
Simulation
Dry-run sensitive execution before dispatch.
Kill switch
Revoke agent authority and running workflows.
Governance control panel
Concept
Permitted Data3 domains
Permitted Tools12 allowed
Autonomy LevelRecommend
Financial Limit≤ USD 25,000
Approval ChainL2 → L3
Geographic ScopeSG · EU
Validity Period90 days
Execution History1959 events
Industry use cases

Deployed as a service, shaped by sector.

Each engagement follows the same pattern — business challenge, connected systems, activated agents, generated interface, controls applied, outcome delivered.

Financial Institutions

Investigations, reconciliations and portfolio reviews performed as governed services.

Payment Companies

Operational intelligence across acquiring, issuing and switching estates.

Retail & Commerce

Make merchandise and inventory transactable by human and AI channels.

Service Businesses

Book, quote and coordinate services through conversational interfaces.

Enterprise Platforms

Consolidate internal tooling behind one intent-driven operating layer.

Government

Case handling and citizen services with source-grounded, auditable answers.

Telecommunications

Provisioning, care and revenue operations with delegated agents.

Multi-Entity Groups

One operating layer across subsidiaries, regions and product lines.

Developer platform

Built to be integrated, extended and audited.

Sample structures shown below. Endpoints and standards support are labelled with a current status until formally released.

example · agent capability schema (illustrative)
{
  "capability": "settlement.reconcile",
  "owner": "guru.finance",
  "inputs": {
    "date": "iso-date",
    "scheme": "string"
  },
  "authz": {
    "scope": ["finance.read", "finance.reconcile"],
    "autonomy": "execute_with_approval",
    "limits": { "amount_usd": 25000 }
  },
  "outputs": { "report_id": "uuid" },
  "audit": true
}

Illustrative structure. Final specifications may differ; do not integrate against this as a stable schema.

Capability status
  • REST APIsIn Development
  • SDKsPlanned
  • MCP-compatible tool connectivityPlanned
  • A2A agent collaborationPlanned
  • WebhooksIn Development
  • Event streamsPlanned
  • Agent registryIn Development
  • Tool schemasIn Development
  • AuthenticationIn Development
  • Sandbox environmentPlanned
  • ObservabilityPlanned
  • VersioningPlanned
Guru · Part of the Takara Solutions ecosystem

Guru is not another application.
It is the operating layer.

The governed intelligence and execution layer through which a company can understand itself, create software, coordinate agents, operate systems and participate in the emerging agentic economy.