Ballina/ Studime rasti

Probleme reale.
Sisteme reale.

Emrat janë anonimizuar aty ku angazhimet janë aktive. Numrat të vërtetë. Sample from 52 live opportunities in our pipeline across banking, retail, construction, real estate, telecom, and insurance.

Filter by sector:
BANKING · KOSOVO CASE 01

Top-3 Commercial
Bank.

12-month AI transformation roadmap for a commercial bank with 500+ employees and 300,000+ retail customers. Every major operational friction point — mapped, prioritized, and solved.

INVESTMENT
€300,000
PROJECTED ROI
15–20×
PAYBACK
2.2 months
The problems
  • 01
    Call Center Overload

    60–70% of 15–25k monthly calls were simple questions stealing time from complex cases.

  • 02
    Loan Processing Bottleneck

    3–7 days per application. 25–40% of applicants dropped off to competitors.

  • 03
    Compliance Document Hell

    30–40% of staff time on manual KYC, AML, and regulatory reporting.

  • 04
    Reactive Fraud Detection

    €500k–2M in annual fraud losses caught after the fact.

The systems built
  • AI Banking Assistant

    Conversational AI across web, app, WhatsApp, and voice. Handles 60–70% of queries autonomously.

  • AI Loan Underwriting

    From 3–7 days to 15–30 minutes for 70% of cases. Analysts focus on edge cases only.

  • Document Intelligence

    KYC onboarding from 2 days to 30 minutes. Compliance reports from 5 days to 4 hours.

  • Real-Time Fraud ML

    Transactions scored in milliseconds. 70–80% fraud prevention.

€1.5M
Annual loan revenue recovered
€400k
Call center cost saved
70%
Fraud reduction
12×
Cross-sell conversion uplift
RETAIL · GROCERY CASE 02

National Supermarket
Group.

30+ store chain with €100M+ annual revenue. Food waste was eating 5–8% of revenue. Stockouts cost another 8–12%. We built the forecasting and inventory engines that ended both.

INVESTMENT
€380,000
PROJECTED ROI
15–20×
PAYBACK
2 months
The problems
  • 01
    Food Waste Bleeding Revenue

    5–8% of gross revenue lost to expiry. €5M+ annually.

  • 02
    Primitive Demand Forecasting

    Based on previous weeks only — ignoring weather, holidays, events.

  • 03
    Frequent Stockouts

    8–12% out-of-stock on high-demand products = €3M lost revenue.

  • 04
    Flat Loyalty Program

    Points-based, no personalization, no recommendations. 2–3% conversion.

The systems built
  • Demand Forecasting ML

    Per SKU × per store × per day predictions. Weather, events, holidays modeled.

  • Dynamic Inventory AI

    Automated re-order, warehouse-to-store distribution, stock rotation.

  • Personalized Loyalty AI

    Per-customer recommendations, targeted promotions, lifecycle triggers.

  • Dynamic Pricing AI

    Competitive pricing based on market, margin, elasticity. +1.5% margin lift.

€2.5M
Food waste reduced (year 1)
70%
Stockout reduction
92%
Forecast accuracy
50%
Loyalty redemption lift
CONSTRUCTION · INTERNATIONAL CASE 03

Mid-Cap Engineering
Firm.

500+ employees. Projects across multiple countries. Bid process took 2–3 weeks while competitors won projects in days. Cost overruns detected only after they'd blown budgets. We changed both.

INVESTMENT
€320,000
PROJECTED ROI
12–18×
BID WIN RATE
2× uplift
The problems
  • 01
    Slow Bid Estimation

    2–3 weeks per bid. Competitors turned around in days and won projects.

  • 02
    Blind Project Risk

    Delays and overruns discovered after the fact, costing millions.

  • 03
    Document Chaos

    10,000+ documents per project. Retrieval took hours. Things got lost.

  • 04
    Safety Reactive

    Manual audits, incidents handled after. Insurance costs climbing.

The systems built
  • Bid Estimation AI

    Trained on historical wins. Full estimate in 1–3 days. Win rate doubled.

  • Project Risk Monitor

    Real-time progress vs baseline. ML flags risk 60–90 days ahead.

  • Document Intelligence

    Semantic search across specs, contracts, RFIs. Seconds, not hours.

  • Safety AI (Computer Vision)

    PPE compliance monitored live. Unsafe behavior auto-flagged.

€1.5M
Additional bids won
60%
Cost overrun reduction
80%
Safety incidents prevented
10×
Bid turnaround speed
TELECOM · REGIONAL CASE 04

Dominant Mobile
Operator.

400+ employees, €80M+ revenue, 15–20% annual churn eating €2M in lost customer lifetime value. 50k+ monthly support calls, 70% of them trivial.

INVESTMENT
€380,000
PROJECTED ROI
8–12×
CHURN CUT
50%
€1M
Churn revenue recovered
68%
Support call deflection
40%
Truck rolls reduced
Upsell conversion
BANKË · MOBILE RASTI 05

Aplikacioni Mobile
i Bankës Rajonale.

Një bankë top-5 retail në Ballkan me 600k+ klientë aktivë kishte nevojë për një aplikacion mobile që të mos dukej si një projekt intern IT i vitit 2014. Dorëzuam native iOS dhe Android në 14 javë — autentifikim biometrik, transferta instant, kontrolle karte, dhe një asistent AI.

INVESTIMI
€140,000
ROI VITI 1
10–14×
KOHËZGJATJA
14 javë
Problemet
  • 01
    Përvojë mobile e vjetëruar

    Aplikacion hibrid nga 2018, rating 2.6★. 40% drop-off në login.

  • 02
    Varësi nga dega

    70% e operacioneve të zakonshme kërkonin vizitë në degë.

  • 03
    Volum thirrjesh support-i

    30k+ thirrje mujore, 65% për pyetje rutinë.

  • 04
    Kontrolle karte të ngadalta

    Ngrirja e kartës kërkonte telefon. Limitet — 48 orë.

Çfarë dorëzuam
  • Native iOS & Android

    SwiftUI + Jetpack Compose. Face ID / fingerprint. Cold start nën 200ms.

  • Transferta instant & SEPA

    Vendore + transferta EU, pagesa të planifikuara, pagesa QR.

  • Kontrolle karte në aplikacion

    Ngrirje, limite, bllokim rajoni, contactless. Të gjitha instant.

  • Asistent AI konversacional

    Shqip + anglisht native. Trajton 60% të pyetjeve rutinë nga fillimi në fund.

4.8★
Rating App Store në launch
62%
Reduktim thirrjesh support-i
Përdorues aktivë ditorë vs më parë
€2.3M
Kursime operacionale viti 1
RETAIL · MOBILE RASTI 06

Aplikacioni i Besnikërisë
për Zinxhirin Ushqimor.

Programi i besnikërisë i një zinxhiri me 30 dyqane ishte një relikt punch-card — 5% redemption, pa personalizim, pa të dhëna. Ndërtuam një aplikacion mobile që i ktheu 850k karta në 850k marrëdhënie të drejtuara nga AI, dhe një levë marzhi që ata mund ta përdornin vërtet.

INVESTIMI
€180,000
ROI I PROJEKTUAR
12–18×
DOWNLOAD-E · MUAJI 6
850k+
+22%
Madhësia e shportës për anëtarët
Norma e redemption të kuponëve
35%
Vizita të përsëritura në muaj
€4.1M
Të ardhura inkrementale (viti 1)
§ 46 më shumë në pipeline

We can't publish
all of them.

Active NDAs limit what we share publicly. In a discovery call we walk through detailed case analyses from your industry — including the failures and the hard lessons.

Request industry-specific cases