Neural Data Intelligence

Your AI is only as smart
as the data it sees.

A unified business knowledge layer connecting your ERP, banking, and operations data. Not a data warehouse. Not a dashboard. The live context that makes your AI agents actually useful — before they act.

BJMRE
Trusted by 200+ businesses
Neural Business Graph — Live context query● Indexing
Agent query"What is the credit risk for Acme Corp right now?"
Customer: Acme CorpPartner ID #1042
Open invoices3 overdue · €14,200 total
Payment history (12 mo)Avg 18 days late
Current credit limit€30,000
Current exposure€28,500 (95%)
Bank transactions (last 90d)14 payments matched
Context assembled in1.2s· 6 data nodes · 4 sources
1.2sContext assembly
4Knowledge levels
EUData residency
100%Isolated per client

An agent without context is just an expensive chatbot.

The difference between an AI that hallucinates and an AI that's useful is the data layer underneath it. PLANA's Neural Data Intelligence is the foundation that makes everything else work — the reason agents give accurate answers, make sound decisions, and execute reliably.

Without NDI

Agent asked: "Should we extend credit to Acme Corp?"

"Based on general business principles, extending credit involves risk assessment and should consider payment history…"

Generic. No real data. Useless for a decision.

With NDI

Agent asked: "Should we extend credit to Acme Corp?"

"Acme has 3 overdue invoices totalling €14,200 (avg 31 days late). Their last 6 payments averaged 18 days late. Current exposure: €28,500. Credit limit: €30,000. Recommend: do not extend — currently at 95% utilisation."

Specific. Data-backed. Actionable.

ERP Integration

One connection to Odoo. Every business record, live.

PLANA connects directly to your Odoo instance and maintains a continuously updated knowledge graph of your business entities: customers, vendors, invoices, purchase orders, employees, projects, and inventory. Not a nightly sync — a live read.

When your Finance Agent asks about an open invoice, it reads the current state in Odoo — not yesterday's export. This makes the difference between advice that is useful now and advice that was accurate 18 hours ago.

  • Reads all Odoo modules: Accounting, Sales, Purchasing, Inventory, HR, Projects
  • Near-real-time sync via Odoo event bus — changes reflect within seconds
  • Full entity graph: customers linked to invoices linked to payments linked to contacts
  • Supports Odoo 16 and 17 Community and Enterprise
Data sources → Knowledge graph
🏢Odoo ERP18,400 records● Live
🏦DSK Bank API2,847 transactions● Live
🏦UniCredit API1,204 transactions● Live
👥HR module47 employees● Live
📦Inventory1,203 products● Live
📊CRM pipeline340 opportunities◌ Pending
↓ unified into knowledge graph ↓
6 sources·24,847 indexed nodes·pgvector embeddings
Banking Context

Banking data meets business context — linked, not siloed.

Bank transactions don't mean much on their own. A €12,400 credit from "ACME CORP SRL" only becomes useful when the system knows it maps to invoice #1047, that Acme usually pays on day 32, and that this payment arrived 4 days early — a signal worth noting.

PLANA links every bank transaction to its corresponding ERP record — the invoice, the vendor payment, the payroll run. The reconciliation context is built automatically, so agents can detect anomalies and explain discrepancies with full business context.

  • Automatic bank transaction → GL entry linking
  • Counterparty recognition — maps bank sender names to Odoo partners
  • Payment pattern learning per customer and vendor
  • Multi-bank support: DSK, UniCredit, and standard OFX/MT940 imports
Banking data integrated with ERP context
Knowledge Architecture

Four levels of business knowledge — structured for AI retrieval.

Most data layers store records. PLANA's Neural Data Intelligence organises knowledge into a four-level hierarchy designed for AI agents to retrieve exactly the right context at the right level of granularity — without over-fetching or missing relevant signals.

L1
DomainBusiness domain — the top-level industry and company context
Retail / Manufacturing / Services
L2
CollectionGroups of related entities with shared business rules
Accounting, HR, Inventory
L3
DocumentIndividual business records and their full metadata
Invoice #1047, Employee record, PO #302
L4
EntryAtomic data points, transaction lines, and field values
Line item, payment amount, due date
L1: DomainRetail / Manufacturing / Services
L2: CollectionAccounting, HR, Inventory
L3: DocumentInvoice #1047, Employee record, PO #302
L4: EntryLine item, payment amount, due date
Each level inherits context from the one above →
Privacy & Compliance

GDPR-first. Your data stays yours. Always.

Every PLANA Pulse deployment is isolated per client. Your data never touches another client's knowledge graph. Embeddings and indexed nodes are stored in your dedicated PostgreSQL instance, not in a shared vector database.

Data residency is in the EU (Sofia, Bulgaria — Exoscale bg-sof-1). No data is sent to third-party AI providers without explicit configuration — AI inference runs through PLANA's controlled API layer, not directly from your raw business data.

  • Dedicated PostgreSQL + pgvector per client — no shared storage
  • EU data residency (Exoscale bg-sof-1, Sofia)
  • Encrypted at rest and in transit — AES-256 / TLS 1.3
  • Right-to-erasure support — data deletion propagates to knowledge graph
  • Access control: agents see only data their role permits
Data isolation model
Data isolationDedicated PostgreSQL per client
Data residencyEU — Exoscale bg-sof-1, Sofia
Encryption at restAES-256
Encryption in transitTLS 1.3
Right to erasurePropagates to graph automatically
Access controlRole-scoped — agents see only permitted data

A data layer built for the long term.

🔢

pgvector embeddings

All business entities are embedded into semantic vectors. Agents can find relevant context through meaning, not just exact field matches.

🔄

Near-real-time sync

Changes in Odoo propagate to the knowledge graph within seconds via the Odoo event bus — no nightly batch jobs.

🌐

Multi-source fusion

ERP, banking, HR, and CRM data are merged into a single unified entity graph. One query resolves across all sources.

📈

Pattern learning

The system recognises behavioral patterns per customer, vendor, and workflow — and surfaces deviations automatically.

🔍

Audit-grade provenance

Every data node in the knowledge graph traces back to its source record and sync timestamp. Agents cite sources.

🗺

Industry knowledge models

Pre-loaded business rules and patterns for 8 industries. The data layer understands what "overdue" means in your sector.