Skip to content
oozmi

Stop reconciling the business after it happened

Five system decisions, five vendor contracts, five auditors asking the same question five different ways — every executive layer pays a tax for the stack beneath it. oozmi puts the close, the operations, and the infrastructure on one database. One platform decision, one audit trail, one number all three executives agree on.

  1. 01 CFO

    Close on the same database the business wrote.

    The CFO opens the trial balance. AR debits, AP pays, and the GL read against the same rows OMS routed, Warehouse received, and Sales closed. There is no overnight batch posting the storefront to finance, no reconciliation queue between the order tool and the ledger. SEF e‑Faktura submits from the source invoice. The auditor's questions land on a config screen, not a forensics expedition across five vendor logs. The boundary: multi-subsidiary consolidation at NetSuite depth is not yet matched — single-entity production at Laguna is the envelope today.

  2. Sub-ledger collapse

    AR, AP, and GL share the row OMS routed. Sub-ledger reconciliation collapses to a join, not an overnight batch.

    SEF from the invoice

    e‑Faktura submits from the source invoice. No nightly export to a localization partner, no separate sync.

    One trial balance

    FX revaluation, accruals, intercompany eliminations all post against the trial balance the consolidator reads.

    Auditor reads a config

    Retention, lawful basis, consent versions, agent_id on every AI write — queryable records, not a vendor-log expedition.

    Approvals on the row

    PO, journal, expense — each approval writes to the record being approved. Amount cannot drift between approval and posting.

  3. 02 COO

    Change a rule without filing a ticket.

    A COO sees the receiving cycle-count cadence drift. On the old stack, changing it was a developer ticket and a sprint. On oozmi, the ops lead opens the admin, edits the cadence as a record, and ratifies. The new cadence is live on the next receive. Reorder points, approval thresholds, project templates, carrier routes — every operating rule is a record the operations team owns. Configuration replaced customization; the audit log writes a row for every change with actor, before, after. The org chart is no longer the bottleneck for how fast a process changes.

  4. Rules as records

    Cadences, thresholds, routes, templates are queryable rows the ops team edits. No partner-repo config, no developer ticket.

    AI proposes the diff

    AI Builder proposes reorder points, cycle counts, project timelines. The ops lead reviews the diff and ratifies.

    One-click revert

    Every applied proposal is a single transaction with a one-click revert. Risk-free experimentation on the floor.

    Audit on every change

    Each rule edit, threshold change, template tweak writes a row — actor, target, before, after, agent_id.

    Org chart unblocked

    Ops moves at the speed of decisions, not the speed of developer queues. The bottleneck moves to the lead, not the ticket.

  5. 03 CIO

    Govern the AI surface like the rest of the platform.

    A CIO who's been asked to approve five point-AI vendors stops. On oozmi, the agent runtime is part of the platform: agents authenticate against the same identity, write through the same permissions, and land in the same audit_events table as every human action. Each AI-driven change carries agent_id, the prompt context, and a before/after the auditor can read. Model keys are yours; the runtime runs inside your tenant. The shadow-AI estate across vendors collapses into one governed surface — one identity, one ACL, one log.

  6. One identity for humans and agents

    Agents authenticate against the same IdP and bind to the same roles as users. No service-account sprawl, no parallel identity graph.

    Shared permission model

    An agent cannot read or write a row a human in its role could not. The ACL is shared, not duplicated.

    Every AI write in the log

    agent_id, prompt context, before and after on every change. The same query that audits people audits agents.

    BYOK and tenant isolation

    Model keys are yours. Prompts and outputs stay in your tenant. Swapping the underlying model is a config change, not a re-procurement cycle.

KPI prompt Production today

Ask the platform a question in plain English — gross margin by category last quarter, AR aging over 60 days, on-time-in-full by warehouse — and read the answer from the same rows finance closed against.

The substrate is the unified database; the natural-language surface reads against the same rows finance closed against.

Production vs Pilot

Production today
  • Finance close on the same database as orders, invoices, contracts
  • Operations rule changes by ops lead, no developer ticket
  • KPI prompt — real-time business KPIs from a single natural-language prompt
  • audit_events as a first-class table across every module
  • Configuration over customization — partner repo replaced by admin
  • One vendor surface, one upgrade cycle, one auditor relationship
In pilot
  • Multi-subsidiary consolidation at NetSuite depth
  • NIST AI RMF, ISO 42001, SOC 2 coverage for the AI surface
  • Multi-country statutory reporting
  • Board-pack generation against the live close

Where it fits

Executives review across ERP , Ticketing , Workflows , AI Builder , OMS , CRM , and Channels — the same row across every module that wrote it. No second copy, no reconciliation step.

Often read alongside: Analysts , Finance & Compliance , and Managers .

Next step — 25 minutes

Bring one number your three executives can't agree on — we'll show why.

Revenue, headcount, on-time fulfilment, AR aging, AI spend — pick the one the CEO asks about and the three answers your CFO, COO, and CIO give. Twenty-five minutes. We open the admin against a sandbox of your data, run the query each system answered with, and show where the rows disagree. Then we show what changes when those rows live on one database.