500+ legacy systems kept Greece's largest bank branch-bound and exposed. A €1.3B rebuild made Sofia AI the face of digital banking.
A Greek bank serving 6 million customers across 12 million accounts and 1.5 million loans, operating on a fragmented technology estate of over 500 legacy systems accumulated over decades.
The legacy infrastructure had accumulated significant technology debt and cybersecurity exposure, including vulnerability to third-party lockout of...
“Ο Παύλος Μυλωνάς περιέγραψε μια πλήρη ανακατασκευή της τεχνολογικής βάσης, από την αποθήκευση δεδομένων μέχρι τα ψηφιακά κανάλια και το core banking.”
Also reported by
Commercial bank providing retail, corporate, and investment banking services.
National Bank of Greece's Virtual banking assistant is part of this use case:
Related implementations across industries and use cases
Agents dug through a vast catalog for answers. Now, AI preps calls and surfaces instant insights for 300 remote employees.
Treated as a plug-in, not a platform. Now 370+ scenarios handle 10M+ daily AI calls — and new models ship in a day, not two weeks.
Each payroll meant up to 20 manual transfers; each invoice re-keyed from a spreadsheet. Agents now prepare both; owners approve once.
Customers waited in call center queues; documents took 3-5 minutes each. Adele now fields 40% of conversations and clears docs in seconds.
Complex ML overhead held back personalized financial insights. Now, AI feeds real-time tailored recommendations to clients and agents.
Quarterly compliance updates meant costly reshoots. Now scripts regenerate instantly — no cameras, no vendors, no delays.
Engineers capped at basic chat, complex work out of reach. Now 700 tackle cross-jurisdiction compliance reviews in hours, not weeks.
Tournaments running simultaneously meant an hour of manual checks each. AI agents now run them in minutes, freeing the team to be proactive.
Large AI training jobs meant fighting for preemptible slots or leaving campus. Marlowe gave any lab guaranteed multi-node access on demand.
500+ legacy systems kept Greece's largest bank branch-bound and exposed. A €1.3B rebuild made Sofia AI the face of digital banking.
A Greek bank serving 6 million customers across 12 million accounts and 1.5 million loans, operating on a fragmented technology estate of over 500 legacy systems accumulated over decades.
The legacy infrastructure had accumulated significant technology debt and cybersecurity exposure, including vulnerability to third-party lockout of...
“Ο Παύλος Μυλωνάς περιέγραψε μια πλήρη ανακατασκευή της τεχνολογικής βάσης, από την αποθήκευση δεδομένων μέχρι τα ψηφιακά κανάλια και το core banking.”
Also reported by
Commercial bank providing retail, corporate, and investment banking services.
National Bank of Greece's Virtual banking assistant is part of this use case:
Related implementations across industries and use cases
Agents dug through a vast catalog for answers. Now, AI preps calls and surfaces instant insights for 300 remote employees.
Treated as a plug-in, not a platform. Now 370+ scenarios handle 10M+ daily AI calls — and new models ship in a day, not two weeks.
Each payroll meant up to 20 manual transfers; each invoice re-keyed from a spreadsheet. Agents now prepare both; owners approve once.
Customers waited in call center queues; documents took 3-5 minutes each. Adele now fields 40% of conversations and clears docs in seconds.
Complex ML overhead held back personalized financial insights. Now, AI feeds real-time tailored recommendations to clients and agents.
Quarterly compliance updates meant costly reshoots. Now scripts regenerate instantly — no cameras, no vendors, no delays.
Engineers capped at basic chat, complex work out of reach. Now 700 tackle cross-jurisdiction compliance reviews in hours, not weeks.
Tournaments running simultaneously meant an hour of manual checks each. AI agents now run them in minutes, freeing the team to be proactive.
Large AI training jobs meant fighting for preemptible slots or leaving campus. Marlowe gave any lab guaranteed multi-node access on demand.