That principle shapes every layer of BizApps360 — from the tools the AI can call to the ledger that records why it called them.
Architecture
Seven layers, one execution foundation
1. Experience
Agent Employee dashboard, Action Center, approvals, explanations.
2. Agent orchestration
Goals, tasks, planning, state, memory and the execution loop.
3. AI / reasoning
Model router, LLMs, vision, classification and extraction.
4. Policy engine
Limits, permissions, deterministic business rules and approvals.
5. Action engine
Controlled tools for ERP, email, documents, CRM and other systems.
6. BizApps360 platform
AP, AR, workflows, exceptions, reporting and integrations — the proven execution layer.
7. Audit ledger
Immutable, traceable record of evidence, policy evaluation, action and approval.
Observe
AI reads and analyses. No action.
Recommend
AI proposes an action with evidence.
Assisted
AI executes after human approval.
Controlled autonomous
AI executes defined low-risk actions within policy.
Autonomous employee
AI runs the full defined workflow within strict financial and security boundaries.
Action engine
Existing BizApps360 APIs become strictly-schematised agent tools. Every call is authenticated, tenant-isolated, permission-checked, idempotent, validated and logged. The LLM never receives unrestricted database or ERP access.
AI decision ledger
Each entry captures the full chain from input to result, timestamped and versioned.
| # | Phase | Typical duration | Outcome |
|---|---|---|---|
| 1 | Discovery | 1–2 days | Workflow, policies, ERP and success criteria documented |
| 2 | Secure connection | 2–5 days | SAP B1, email and document sources connected |
| 3 | Historical learning | Several days | Past invoices and exceptions used to validate behaviour |
| 4 | Configure employee | 1–2 days | Role, authority, policies, tools and escalation rules |
| 5 | Shadow mode | 1–2 weeks | AI decisions compared with human decisions |
| 6 | Assisted mode | 1–2 weeks | Humans approve AI recommendations and actions |
| 7 | Controlled autonomy | Progressive | Low-risk work executed automatically |
| 8 | Go-live | Ongoing | Production Agent Employee with monitoring |
| 9 | Monthly value review | Monthly | ROI, accuracy, interventions and expansion opportunities |
> 98%
Invoice extraction accuracy
> 97%
PO matching accuracy
> 95%
Exception classification
0
Critical financial errors
0
Policy violations
100%
Audit coverage
Illustrative targets. Thresholds are validated by risk category and customer workflow during the pilot; we never claim a universal accuracy guarantee.
Onboarding data checklist
The workflow assessment maps your approval matrix, tolerances and exception causes to the autonomy model — before any pilot begins.