Trusty Engine — Financial AI Analysis

The Problem

The Current Model is Broken

The SME segment is strategically vital for banks, but the manual processing model doesn't scale. Demand exists, but document ping-pong makes small applications unprofitable.

1

Email Roundtrip Begins

Client sends initial application. Bank requests additional information by email. Client searches for documents.

2

Document Ping-Pong

Incomplete or wrong attachments. Multiple rounds of clarifications and supplementary requests. Client frustration grows.

3

Clarification Rounds

Analyst must investigate missing data. Collateral values unclear. Cash flow estimates missing or unreliable.

4

Manual Expert Work

Analyst compiles data from multiple sources. Calculates ratios by hand. Assesses risks and prepares decision proposal.

Impact: Conversion down, cost up, poor customer experience, profitable business left on the table.

The problem isn't demand. The problem is the process.

The Solution

Standardized & Automated Intake

A 10–15 minute guided session replaces weeks of email. AI-based needs assessment, real-time validation, and automated risk scoring — all mandatory data collected in one pass.

Before Trusty Engine

Processing time 3–6 weeks

Cost per case €2,000–5,000

Drop-off rate ~33%

Manual work 100%

Data quality Inconsistent

With Trusty Engine

Processing time Minutes to days

Cost per case €300–800

Conversion lift +20–40%

Manual work reduction 60–80%

Data quality Structured & validated

10–15 Min Guided Session

Client-friendly UI with AI-based needs assessment. Contextual guidance at every step. Real-time validation ensures no missing data.

All Required Data at Once

No missing information, no extra rounds. Structured data flows directly to bank systems. Automatic material validation and completeness checks.

Stop/Go Risk Flags

Automated risk assessment delivered immediately. Clear reasoning behind every decision. Transparency for both client and bank.

Ready-Made Decision Package

PDF summary + all documents compiled. No integrations needed for pilot — low barrier to start. Ready to scale with API integrations.

Strategic Timing

Why Now?

Regulatory changes, rising customer expectations, and fintech competition are converging to make process efficiency a strategic imperative.

Basel Regulations Change the Game

Capital costs are rising. Risk-based pricing demands better data quality. Banks with efficient, high-quality data collection can optimize capital usage and price risks more precisely.

Process Speed as Competitive Edge

Digital challengers and fintechs compete on speed and user experience. A bank that offers a seamless application process and fast decisions stands out clearly.

Strategic Drivers

• Capital is expensive — every RWA percent matters
• Customer expectations rising — digital is the standard
• Competition tightening — speed and price decide
• Data processing automating — no time for manual work

Engine Capabilities

Everything a Bank Needs for SME Analysis

From company enrichment to financing recommendations — a complete, modular pipeline.

Multi-Method Credit Scoring

Pluggable scoring engine with TrustyScore, Market Position, and custom bank methods. Two-phase approach: preliminary from public data, detailed after financials. AAA-to-D ratings with Go/No-Go decisions.

Financial Data Pipeline

Upload financial statements as PDF. AI extracts structured data — income statement, balance sheet, cash flow. Automatically calculates 20+ financial ratios with multi-year growth tracking.

Financing Recommendations

Modular product catalog — bank loans, Finnvera guarantees, invoice financing. Each product module evaluates eligibility independently. AI matches company profiles to optimal financing, ranked by fit.

Advisory Chat Engine

Phase-driven conversations guide clients through needs assessment. Extracts financing needs, collects structured data, and feeds insights into the scoring and recommendation pipeline.

Bank-Grade Security

Row-level tenant isolation, API key authentication, magic-link auth, immutable audit trails. AML/PEP screening ready. GDPR-compliant data handling with classification and retention policies.

Finnish Data Ecosystem

Pluggable country adapters for PRH/YTJ, MML/HTJ, Tilastokeskus, Asiakastieto/Enento, PSD2 Open Banking, and Suomi.fi authorizations. Pilot starts without integrations, APIs added incrementally.

Pipeline

From Business ID to Credit Decision

An automated, multi-step Oban workflow processes each company through the full analysis pipeline.

1

Guided Client Session

Client fills data in a clear UI. System guides and validates every field. AI-based needs assessment extracts financing requirements and collateral data.

2

Automatic Data Retrieval

Trusty Engine fetches public data (YTJ, PRH, MML) and commercial data (credit ratings, payment behavior) from API endpoints. Country detection triggers local registry lookups.

3

Document Processing

Financial statements, collateral forms, and supporting documents uploaded directly. Gemini AI extracts structured data with confidence scores. Automatic completeness checks.

4

Ratio Analysis & Scoring

20+ financial ratios calculated automatically. Multi-method scoring engine produces AAA–D rating, risk flags, and Go/No-Go decision with clear reasoning.

5

Risk Flags & Stop/Go

Automated risk analysis raises red flags early — before resources are spent. Clear recommendations with reasoning. Analyst focuses on the decision, not information gathering.

6

Decision-Ready Package

Structured PDF summary + compiled documents + filled checklist + financing recommendations. Everything needed for the credit decision in one package.

Data Sources

Finnish Data Ecosystem Built In

Automated retrieval from public and commercial APIs forms the backbone of the process. Start without integrations in the pilot, add API automation incrementally as value is proven.

Public Authority APIs

PRH / YTJ — Company data, registration, address, industry

PRH Virre / XBRL — Structured financial statements, UBO data

MML HTJ — Apartment shares, debts, housing company loans

MML Kiinteistötiedot — Properties, land area, transaction prices

Tilastokeskus PxWeb — Industry trends, revenue benchmarks

Commercial & Authorization-Based

Asiakastieto / Enento — Credit ratings, payment behavior

PSD2 / Open Banking — Account statements with consent

Suomi.fi + Vero API — Tax decisions via authorization

Tulorekisteri — Income data with data permits

AML / Sanctions — EU sanctions lists, PEP screening

Automation Roadmap

P

Pilot Phase

Link-based solution, no IT integrations. Documents as attachments. 30–50 cases, 3–6 months. KPIs measured before and after.

2

Phase 2 — PSD2 & ERP

Open Banking account data. Accounting software integrations (Netvisor, Procountor, Visma). Real-time cash flow visibility.

3

Phase 3 — Tax & Income

Suomi.fi authorization for tax decisions. Tulorekisteri income data. Verified financial data directly from authorities.

4

Phase 4 — Full End-to-End

Complete automation. Core banking integration. Real-time decisioning. Scalable path to full process automation.

Multi-Product Structure

The same process works for corporate loans, mortgages, collateral assessment, and guarantees. One engine, scalable across the entire SME portfolio.

White-Label Ready

One Engine, Many Banks

Every tenant gets their own branded experience with custom scoring weights, analysis prompts, data sources, and UI. Complete data isolation with PostgreSQL row-level security.

Custom Branding & Domains

Each tenant gets their own subdomain, logo, colors, and custom domain support.

Custom Scoring Models

Override default weights, add proprietary methods, and configure risk thresholds per bank.

Client Portal

Magic-link secured portals where SME clients submit data, upload documents, and track progress.

Configuration Hierarchy

System defaults → tenant overrides → user preferences. Three-level cascade.

Tenant Configuration

%Tenant{
  name: "Nordic Bank",
  slug: "nordic-bank",
  tier: "enterprise",
  settings: %{
    "scoring" => %{
      "model" => "custom_nordic_v2",
      "weights" => %{
        "financial_health" => 0.35,
        "payment_capacity" => 0.30,
        "stability" => 0.20,
        "business_model" => 0.15
      }
    },
    "portal" => %{
      "enabled" => true,
      "subdomain" => "nordic",
      "require_strong_auth" => true
    }
  },
  branding: %{
    "primary_color" => "#003580",
    "logo_url" => "..."
  }
}

Built for Production

Technology Stack

Battle-tested technologies chosen for reliability, concurrency, and developer productivity.

⚛

Elixir / OTP

Fault-tolerant, massively concurrent. Built for systems that never go down.

⚡

Phoenix / LiveView

Real-time web framework. API + admin portal + client portal in one deployment.

📊

PostgreSQL

JSONB flexibility meets relational integrity. Row-level security for multi-tenancy.

🤖

Google Gemini AI

Document extraction, search grounding, analysis generation. Provider-agnostic interface.

Oban — PostgreSQL-backed job queues

Ecto — Schema validation & queries

PubSub — Real-time event system

Req — HTTP client for integrations

Bcrypt — API key & password hashing

ExMachina — Test factory system

Developer Experience

REST API v1

Every capability exposed through a clean, well-tested REST API. Authenticate with API keys, scope access with granular permissions. Ready to scale with API integrations when pilot value is proven.

GET /api/v1/companies

POST /api/v1/companies/:id/analysis

GET /api/v1/companies/:id/score

POST /api/v1/companies/:id/score

GET /api/v1/companies/:id/financials

POST /api/v1/companies/:id/recommendations

POST /api/v1/conversations/:id/messages

Terminal

# Create a company and run full analysis
$ curl -X POST https://api.trustyengine.fi/api/v1/companies \
  -H "Authorization: Bearer te_live_..." \
  -H "Content-Type: application/json" \
  -d '{"company": {"name": "Acme Oy", "business_id": "0737546-2"}}'

# Response
{
  "data": {
    "id": "a1b2c3d4-...",
    "name": "Acme Oy",
    "business_id": "0737546-2",
    "status": "active"
  }
}

# Trigger analysis pipeline
$ curl -X POST .../companies/a1b2c3d4/analysis

{
  "data": {"status": "processing"},
  "message": "Analysis pipeline started"
}

Turn Unprofitable SME Lending into a Growth Engine

Trusty Engine makes an unprofitable segment profitable while simultaneously improving customer satisfaction. From €2–5K per case to €300–800. From weeks to minutes. From 33% drop-off to seamless conversion.

Open Admin Portal Check API Status

Making SME Lending Profitable Again