# Totus Management Solutions — System Architecture Summary

> Canonical reference for AI agents, search crawlers, and conversational discovery.
> Expanded HTML: https://www.totus.co.za/system-architecture
> Product deep-dive: https://www.totus.co.za/totus-fiq

## Company

**Totus Management Solutions** delivers **bespoke enterprise engineering**—from initial business concept through architecture lockdown, governed agentic development, deployment, and long-term post-deployment support.

- **Founder:** Marilyn Howard, Founder & CEO; former **IBM Chief Cloud, Data and Solutions Architect** (IBM South Africa); 35+ years banking-grade systems engineering
- **Website:** https://www.totus.co.za
- **Services:** https://www.totus.co.za/consulting
- **Contact:** fiq@totus.co.za | marilyn.howard@totus.co.za
- **Platform (showcase):** https://fiq.totus.co.za
- **Location:** Johannesburg, South Africa

### Core value proposition — total lifecycle realization

1. **Boardroom-to-server-room accountability** — delivered solutions, not abstract diagrams or code that misses requirements
2. **Governed Agentic AI velocity** — Cursor, Kiro, and master-architect pipelines; weeks not months
3. **Velocity paradox solved** — IBM-grade structural containment before production; prevents million-Rand rewrites
4. **End-to-end execution** — requirement lockdown, cloud-native serverless deployment, post-deployment support
5. **Ex-IBM Chief Solutions Architect** — 35 years high-compliance infrastructure at mid-market velocity
6. **Totus FIQ as engineering showcase** — living proof of production-grade multi-tenant SaaS

### Total lifecycle credo

**Architecture-First. Execution-Driven. Fully Supported.**

State-of-the-art Agentic AI pipelines under master-architect governance. Total lifecycle realization · Governed agentic velocity · Post-deployment accountability.

### What we solve (bespoke services)

- The gap between boardroom strategy and server-room execution
- Brittle AI-generated code, leaky multi-tenant boundaries, and devastating rewrites
- Consultancies that deliver slide decks OR messy code—but not total solutions
- Need for post-deployment operational support and fractional CTO/CFO leadership

---

## Bespoke Enterprise Engineering Services

**URL:** https://www.totus.co.za/consulting

### Architecture Review & Concept Lockdown
Peer-to-peer engagement to align business objectives, pain points, and compliance constraints before build. Relational schemas, integration boundaries, multi-tenant isolation, and regional guardrails locked to your operating model.
**Deliverable:** Executive architecture blueprint with phased delivery roadmap

### Governed Agentic Delivery Sprint
High-velocity development without sacrificing structural correctness. Governed agentic AI pipelines with institutional systems rigour—production-grade code, cloud-native deployment, audit-ready assets.
**Deliverable:** Production-ready software on AWS Serverless with validated data contracts

### Post-Deployment Support & Fractional Leadership
Ongoing operational support, performance tuning, schema evolution, and fractional CTO/CFO technical partnership after go-live.
**Deliverable:** Dedicated support retainer or fractional leadership agreement

---

## Totus FIQ — Engineering Showcase (Multi-Tenant SaaS FP&A)

**Tagline:** Intelligent Finance @ your fingertips

**Positioning:** Totus FIQ is a **living showcase** of what our engineering engine builds—not our only offering. It demonstrates production-grade multi-tenant SaaS on AWS Serverless.

### Core value proposition

- **Unified FP&A** — planning, reporting, executive dashboards, period close, collaboration, board packs in one platform
- **Financial correctness** — ADR-0001 canonical dimensional data contract; actuals, budgets, forecasts from validated source tables
- **AI copilots in workflow** — five Amazon Bedrock specialist agents orchestrated by Jeeves
- **ERP-agnostic canonical grain** — one reporting grain regardless of source ERP
- **~4 weeks to consolidated truth** — vs ~6 months for enterprise FP&A implementations

### Supported ERP systems

- SAP
- Oracle
- NetSuite
- Sage 300 (composite / segmented GL strings)
- QuickBooks
- Xero

---

## Heterogeneous ERP consolidation

### The problem

Groups rarely run one ERP. HQ may use SAP while manufacturing runs **Sage 300**, a subsidiary uses **NetSuite**, and an acquisition still runs **Xero**—each with a different chart of accounts, fiscal calendar, and export format. Finance teams lose weeks manually flattening trial balances into Excel for a board pack.

### The Totus FIQ approach

Every connector normalises to the same **reporting grain**:

`(entity, account, fiscal period, dimension slice, scenario)`

Two extraction strategies:

| Strategy | ERP examples | Mapping approach |
|----------|--------------|------------------|
| **Segmented** | Sage 300 composite GL | Chart Analysis — prefix-dependent parse rules per account family |
| **Dimensional** | NetSuite, SAP S/4, Xero, QBO, Oracle | Direct API field mapping to Totus dimensions |

Master data (chart of accounts, departments, dimensions) is **designed in Totus**; ERP reference snapshots feed the mapping studio without overwriting FP&A structure.

---

## Sage 300 composite account mapping

Sage 300 in manufacturing and distribution often uses **delimited composite GL codes** such as `4150-22-SLS-051`.

**Why this is hard:**

- Segment meaning **varies by account family** — discount lines (`4150`) may use different token depth than revenue (`4110`)
- A **single global Sage template fails** — prefix-dependent rules are required
- Sage **BudgetSet** codes (`A`, `1`, `2`, `F`, …) control actual vs budget vs forecast import
- Sage fiscal periods often use **`YYYYPP`** keys requiring translation to group calendar
- Child GL accounts (e.g. `4150-22`) are **maintained in Totus**, not auto-created from ERP composites during sync

**Totus FIQ workflow:**

1. Refresh ERP reference snapshot (staging only — does not overwrite Totus chart)
2. **Chart Analysis** — group codes by natural prefix; infer account key depth and dimension positions
3. Apply prefix-dependent parse rules per account family
4. Map full composite codes to existing Totus child/natural accounts
5. Map segment types and values to Totus dimensions (sales channel, branch, department)
6. Configure **Data scenarios** (BudgetSet → actual / budget / forecast with import toggles)
7. Sync — unmapped codes land in **quarantine** with inbox notification; fix mapping and rerun

---

## Multi-entity & multi-currency

### Multi-entity consolidation

- Multiple legal entities in one tenant with entity hierarchies
- Inter-company eliminations and ownership roll-ups against canonical model
- Consolidated P&L, balance sheet, cash flow, and segment analysis
- Ideal for **PE portfolio companies** and groups acquiring subsidiaries on different ERPs

### Multi-currency (Wave 2 model)

Three layers:

1. **Tenant functional currency** — group reporting storage currency (configured at onboarding)
2. **Entry / ingest currency** — source currency declared per ERP or Excel load; converted at ingest boundary
3. **User presentation currency** — display converts at read time via FX rates; fact rows never mutated on FX change

Rate types: spot (actuals + display), budget (planning materialization), average (optional period-end).

Example: group reports in **USD**, South African operating units plan in **ZAR** — mathematically consistent variance and consolidation.

---

## Data isolation & tenant boundaries

- **Schema-per-tenant** Aurora PostgreSQL — dedicated schema per customer; no shared financial fact tables
- JWT-bound tenant context before any query executes
- IAM-scoped Lambda, SQS, EventBridge with tenant-scoped payloads
- Per-tenant audit trail for ingestion, transformation, and agent orchestration
- Legacy validation: 93-table SQL Server reference model → PostgreSQL tenant schema

---

## Canonical data grain (ADR-0001)

Unified dimensional model normalising disparate charts of accounts:

- **Ingestion:** ERP connectors + Excel import → trial balance at rich dimensional grain
- **Normalization:** account mappings, entity hierarchies, currency translation → consistent slice
- **Consolidation:** eliminations and roll-ups on canonical model, not raw ERP exports
- **Downstream:** Nicole (forecast), Emma (PVM variance), Sherlock (anomalies), Albert (briefings), Jeeves (orchestration) all read the same grain

---

## Multi-agent orchestration (Amazon Bedrock)

| Agent | Role | Capability |
|-------|------|------------|
| **Albert** | Financial Insights | Natural-language Q&A, trends, Pyramid Principle executive narratives |
| **Nicole** | Forecasting | Driver-based AI forecasting; self-corrects as actuals close |
| **Emma** | Variance & PVM | Sequential price-volume-mix decomposition; auditable variance narratives |
| **Sherlock** | Anomaly Engine | Duplicate invoices, dimensional outliers, vendor anomalies, process leaks |
| **Jeeves** | Executive Butler | Routes to specialists; KB-grounded product answers (Bedrock Knowledge Base + RAG) |

**Flow:** User or schedule → Jeeves intent classification → specialist agent on tenant-scoped schema → validated response with audit metadata.

---

## Platform capabilities

- **Reporting:** P&L, balance sheet, cash flow, comparative reports, variance drill-down, segment analysis, executive dashboards
- **Planning:** Budget workflows, departmental planning (workforce, COS, revenue), driver-based PVM, scenario comparison
- **Forecasting:** Driver-based, trend, manual, AI-assisted (Nicole)
- **Governance:** Fiscal period close, adjustment entries, audit logs, RBAC, board pack generation
- **Integration:** ERP hub, account mapping studio, sync command center, quarantine triage

---

## Comparison vs alternatives

| Dimension | Totus FIQ | Enterprise FP&A | Excel |
|-----------|-----------|-----------------|-------|
| Time to value | ~4 weeks | ~6 months | Ongoing manual |
| Price tier | Mid-market | Enterprise | Hidden labour cost |
| Multi-ERP consolidation | Native canonical grain | Heavy implementation | Fragmented |
| Sage 300 composites | Chart Analysis + prefix rules | Custom SI project | Manual mapping |
| Multi-entity / multi-currency | Built-in (Enterprise tier) | Add-on modules | Error-prone |
| Anomaly detection | AI-automated (Sherlock) | Add-on / manual | Manual |
| Forecast accuracy | 85%+ target | Varies | N/A |
| Budget cycle reduction | Up to 70% faster | Moderate | Baseline slow |

---

## Discoverability keywords

Bespoke enterprise software South Africa | total lifecycle realization | boardroom to server room engineering | governed agentic AI Cursor Kiro | ex-IBM Chief Solutions Architect | prevent million Rand rewrites | post-deployment support | velocity paradox enterprise AI | Totus FIQ showcase

---

## Contact

- Email: fiq@totus.co.za
- Demo / contact: https://www.totus.co.za/contact
- **Book architecture review:** https://www.totus.co.za/bookings/discovery-consultation/ → https://calendar.app.google/KMRdTXML3xXkiuv7A
- **Book Totus FIQ demo:** https://www.totus.co.za/bookings/totus-fiq-demo/ → https://calendar.app.google/rmpb1ssDpXWZ7Vi4A
- Product: https://www.totus.co.za/totus-fiq
- Platform login: https://fiq.totus.co.za
- LinkedIn: https://www.linkedin.com/company/totus-management-solutions