Banking Platform 6.0

AI-powered banking platform for SME credit growth

Algabas connects your systems and the public and closed registries you can't reach today, then puts agents on top that enrich, advise and show their work — all on your own infrastructure.

Live demo runs on real platform data. Examples shown use Uzbekistan SME portfolio.
Live platform data — refreshed every minute
SME clients
Active bankers
Transactions
Revenue UZS
AML cases
Deals closed
Proven results in similar banks
What banks like yours achieved with digital SME platforms
ABN AMRO · 6 months
€1B
SME loans issued through the digital platform
ABN AMRO New10
5,500+
Active SME clients in the first half year
ABN AMRO group
65%
New customers brought to the bank group
DBS Digibank · Asia
6M+
Retail and SME customers in 18 months
Algabas platform
+18%
Sales lift from AI-driven recommendations
Algabas delivery
16w
From kick-off to a working pilot
Our clients in the financial sector
SQB
Aloqa Bank
Ipoteka Bank
Access Bank
Asseco
Regulatory readiness

Governance built to the EU AI Act standard

Algabas's core is not a black box. It is a human-in-the-loop compliance assistant that sits alongside your core banking system — every AI recommendation is logged, and a person makes the final call. That maps directly onto the two requirements regulators care about most.

Article 12 — Record-keeping

Every AI decision is logged

Each recommendation the AI makes — its inputs, confidence, rationale and the model version behind it — is written to an immutable decision log. When a regulator or auditor asks "why did the system flag this?", the answer is one query away.

Article 14 — Human oversight

AI recommends, a human decides

The AI never acts on its own. An officer accepts, modifies or rejects every recommendation, and the reason for any override is captured. Meaningful human control is built into the workflow, not bolted on afterwards.

Timeline
2 August 2026

The EU AI Act becomes fully applicable, including the high-risk requirements for record-keeping (Art. 12) and human oversight (Art. 14). Some obligations for stand-alone high-risk systems are being phased under the 2026 simplification package — but logging and oversight are foundational either way.

Why it matters in Uzbekistan & CIS

A trust benchmark, not just an EU rule

The EU AI Act has become the de-facto global reference for trustworthy AI governance. Building to its logging and human-oversight standard is a signal your supervisors, correspondent banks and partners recognise — well beyond the EU's borders.

Talk to us about an audit-ready pilot
Agentic AI Studio

A team of AI agents — under human control

Algabas's Agent Studio runs banking AI agents on top of your core.

💬

🧰

👤

📋

💰

🛡️

Explore the platform
The gap

What changes with Algabas

Today, in most banks

Systems are siloed — banker, ops and client each see a different fragment.
Registry checks are manual and take days.
No single verified profile of the client.
AI is absent — or it sends sensitive data outside the bank.
Decisions are made without a traceable evidence chain.

With Algabas

One data fabric unifies your systems and registries in real time.
Agents enrich from seven registries in seconds — with a full trace.
A golden 360° profile — single source of truth.
AI runs on-premise — no data leaves your perimeter.
Every step is recorded; the human decides, the audit holds.
Modules

Six modules — one ecosystem

Each module solves a concrete entrepreneur task and triggers the right bank product at the right moment.

📈

Cashflow forecast

AI looks at 90 days of transactions and tells the owner about runway, gaps, and concentration risk. With recommendations.

🧾

Tax advisor

Simplified vs general regime — concrete figures of savings. Built-in payment calendar.

💸

Smart payments

Pre-flight compliance check before sending money: sanctions, AML, beneficiary verification.

☀️

Banker AI Copilot

Morning brief: today's priority, risk focus, deal opportunities. By email at 09:00 and in the cabinet.

⚖️

Compliance fabric

AML / KYC / SWIFT / Sanctions / Smart Onboarding — one engine, shared audit trail.

🏆

Banker effectiveness

Composite score 0–100 per banker. Transparent ROI of every RM. Leaderboard with medals.

Inside the platform

How a request travels through Algabas

What happens when an owner asks the AI for a cashflow brief. Eleven orchestration nodes pulling data from internal systems, the tax authority and the AI engine — composed into one response.

Live flow ·POST /v2/owner/cashflow/forecast· 11 nodes
HTTP request Owner cabinet Auth & RBAC JWT + audit Cache check Redis · 5 min TTL PostgreSQL 90-day aggregate Tax authority API e-soliq / e-gov gateway External APIs Uzum · 1C · ERP Build LLM prompt tone-aware AI engine OpenAI · Claude Parse & validate + fallback rules Compose response merge + format JSON Cache write 5 min TTL Brief delivered ~1.5s · HTTP 200
Cabinets

Five cabinets — one ecosystem

Without separate systems for compliance, sales and self-service. Everything integrated with real-time cross-cabinet data.

💼

Owner — SME entrepreneur

Personal cabinet where the owner sees the whole business: sales, finance, taxes, competitors — in one window.

Open Owner Portal
🏦

Banker — relationship manager

Portal that automatically highlights hot clients ready to buy. Manager works only with prepared offers.

Open Banker Portal
📡

Operations — compliance & audit

Compliance Center for AML/KYC/Sanctions, audit dashboard, banker effectiveness leaderboard, support inbox.

Open Ops Portal
🎛️

Agent Studio — AI control tower

Frontline AI agents under control: tools, autonomy, human-in-the-loop approvals and full audit. Five agents for prospecting, nurturing, meeting prep, deal-scoring and coaching.

Open Agent Studio
☪️

Islamic finance — Shariah-compliant

Separate cabinet for Shariah-compliant financing: Murabaha and Ijara pipeline, AAOIFI-aligned profit logic, Shariah-board approval and audit. No interest, full transparency.

Open Islamic Portal
🧾

Lending — credit conveyor

Dedicated banker cabinet for the whole loan lifecycle: queue, application, underwriting decisions, credit approval (4-eyes / committee by limit), disbursement, portfolio monitoring and leasing — retail, SME and leasing in one place.

Open Lending Portal
Impact

What changes after Algabas implementation

+18%
Sales growth
Automatic recommendations of bank and non-bank products at the right moment
Faster time-to-market
Test and launch new services 4× faster
Cheaper development
Cost of creating new products drops 6×
For DevOps

How Algabas deploys inside your bank

Containerised stack. On-premise. Three security zones — DMZ, application, data — with one controlled outbound channel for the AI engine. No bank data leaves the perimeter.

🏛️ Bank perimeter — your data centre USERS Bank staff browser · mobile SME owners browser · mobile Compliance ops browser DMZ ZONE · public access nginx reverse proxy · TLS WAF · rate limit SSO · MFA · JWT issue APPLICATION ZONE · Docker / Kubernetes Banker portal nginx + Vue separate container Owner portal nginx + Vue separate container Ops portal nginx + Vue separate container Node-RED orchestrator 50+ flows · low-code · audit Agent Studio nginx + Vue separate container Islamic portal nginx + Vue separate container Local AI model trained on your data stays in your perimeter RabbitMQ async broker · workers DATA ZONE · encrypted at rest Database roles + RLS · backup Redis cache session · brief cache 5 min TTL Audit log store append-only · 7 yrs retention Object store PDF exports MinIO / S3 compatible INTEGRATION (Node-RED connectors over VPN / internal API) Core Banking CRM ERP Legacy DBs ☁️ External (controlled outbound) OpenAI / Anthropic LLM · anonymised only 8 state registries e-soliq · KATM · sanctions · courts… Payment systems + many more via policy via outbound proxy + policy egress only DLP · egress log 🔒 Security model → TLS 1.3 everywhere → mTLS inter-service → JWT short-lived → audit trail 7 years → no PII to LLM → secrets in KMS / HSM → backup & DR (offsite)
Stack
Docker Compose or Kubernetes · 8 services · 12 vCPU recommended
Storage
≈ 500 GB in the first year
Monitoring
Prometheus · Grafana · structured logs to your SIEM
Backup
WAL-based PITR · daily snapshots · 7-year retention for audit data
Integration

Works on top of your core banking system — any vendor

Algabas does not replace your ABS. It sits alongside it as the AI + human-in-the-loop compliance layer. The exchange is the same regardless of which core system you run.

🏦
Your ABS
any core banking system
standard connector
REST · ISO 20022 · batch
🛡️
Algabas
AI recommends · human decides
⬇️ From ABS into Algabas
  • Clients & KYC profiles
  • Accounts & balances
  • Payments & transactions
  • Counterparties / beneficiaries
  • Financial-monitoring & sanctions lists
⬆️ From Algabas back into ABS
  • Compliance decisions (accept / hold / reject)
  • Risk flags & scoring
  • Immutable audit log (EU AI Act Art. 12)
  • STR / SAR drafts for the FIU

Connecting a specific core system is a configuration step — the connector adapts to its API, the logic above stays the same. No rip-and-replace of your existing banking platform.

Recognition

Global recognition

Algabas is part of the Asseco group — a banking-technology leader with €4B annual revenue and presence in dozens of countries.

🏆

Best AI startup, DIFC Dubai

One of the most prestigious financial innovation competitions in the world.

🇺🇿

IT Park Uzbekistan resident

Government-backed tech ecosystem resident with full state-supported infrastructure.

🌐

Part of Asseco Group

€4B annual revenue, dozens of countries, deep banking-tech expertise.

Try the platform on your data

Pilot in 4 weeks · with your bank's real clients · with your LLM API key

Request a pilot Open live demo
Lead → Credit → Decision

From first signal to funded deal

One pipeline turns a market signal into a funded deal — lead generation, qualification, the credit conveyor and the decision, all under human-in-the-loop control.

01

Lead generation

AI triggers from your existing base plus public landing forms.

02

Qualification

Sanctions / PEP screening, AI scoring and dedupe run automatically.

03

Credit conveyor

Six guided steps for legal entities and individuals.

04

Decision & funding

Auto-approve within limits, route to committee, or reject.

Credit conveyor — 6 steps
Applicant type Profile autofill Parameters Financial data Scoring & stop-factors Decision & docs

☪️Islamic track: the same flow runs as a Murabaha / Ijara conveyor — AAOIFI-aligned, Shariah-board approved, no interest.

Algabas at a glance

€4B
Asseco group revenue
80+
Orchestration flows
6
Production modules
8
Supported languages
100%
On-premise control
6–10
Weeks to pilot
ROI & TCO

Estimate your savings

Adjust the inputs to your bank's scale and see how AI-assisted underwriting reduces manual workload. Figures are indicative.

Your numbers
Loan applications / month600
Manual time per application (min)40
Loaded staff cost / hour$12
Share automated by AI70%
Staff hours saved / month
280
Estimated cost saved / year
$40,320
Decision time
40 → ~3 min
AI-handled / month
420
Trust & Security

Bank-grade by design

Security, governance and auditability are built into the platform — not bolted on.

🏛️

On-premise, three zones

Containerised stack on your own infrastructure. DMZ, application and data zones with a single controlled boundary.

👥

RBAC & four-eyes

Role-based access with maker-checker approval. Configuration changes need a second authorised reviewer.

📜

Immutable audit trail

Every configuration change and AI decision is logged with actor, time and reason — exportable, SIEM-ready.

⚖️

EU AI Act aligned

Model cards, reason codes and human-override logs — governance designed to the de-facto global reference.

🔐

Encryption & residency

Data stays in your jurisdiction. Encryption in transit and at rest; secrets managed outside the codebase.

🧑‍⚖️

Human-in-the-loop

AI proposes; people decide. No autonomous credit decisions — every recommendation is reviewed and approved.