Bank of Georgia’s Data Analytics and AI team has increased the value of non-risk AI initiatives tenfold while reducing its own headcount by 10%. Its transformation rests on a fundamental change in what the team considers successful: not delivering models, but delivering measurable business outcomes.
A technically successful AI model that never changes the business is still a failed project. That principle has driven a three-year transformation of Bank of Georgia’s Data Analytics and AI team, changing its role from specialist capability to one of the engines of the bank’s commercial performance.
The team has redesigned how it works with business units, changed how its own people are measured and rewarded, introduced AI agents into delivery, and given employees across the bank the ability to create their own AI solutions.
From Model Delivery to Business Impact
Previously, data scientists built models while business teams handled process change, leaving neither side fully accountable for the eventual outcome. Bank of Georgia replaced that separation with joint ownership.
Business units now dedicate a senior translator full-time to AI initiatives, while leaders participate in weekly decision meetings. Data and AI professionals are expected to co-design the process change surrounding their models and remain accountable until solutions are operating in the business.
The team reinforced the change through its capability and reward structures. Data scientists spend time in branches, shadow frontline employees, and participate in senior reviews to understand the commercial context behind their work. Business consulting skills have been incorporated into the team’s capability matrix, while incentives are linked to business outcomes.
The results demonstrate the scale of that shift:
- Around 80% of AI projects reach live business use and generate value.
- Non-risk AI project value has increased tenfold.
- AI projects contribute approximately 5% of the bank’s operating profit.
- Stakeholder NPS has reached 86.
- These results were achieved while team headcount fell by 10%.
Operational applications include automating 98% of retail unsecured loan underwriting and reducing SME low-limit underwriting turnaround from four days to one minute.
“Exceptional profit-driving AI team.”
“Outstanding. The numbers speak for themselves: the DS Agent innovation halving delivery timelines is genuinely forward-thinking.” – Judges’ comments
Using AI to Redesign the Team Itself
The team has applied the transformation philosophy internally. Its DS-Agent multi-agent system supports investigation, modelling, and deployment under human guidance. In pilots, work previously requiring five people can be delivered by a senior data scientist and delivery manager, while project timelines have reduced by approximately 50% without a reported loss of quality.
This is changing the role of the data scientist as much as the underlying technology with greater emphasis placed on directing, validating, and applying analytical work rather than executing every stage manually.
Making AI Everyone’s Capability
Greater productivity within the central team did not solve every capacity constraint as smaller AI opportunities continued accumulating. The response was to remove it as the gatekeeper.
An internal self-service platform now enables non-technical colleagues to create assistants for tasks including document summarisation, spreadsheet analysis, sales pitches, customer experience assessment, and knowledge retrieval.
The rollout was supported by a GenAI Centre of Excellence and around 100 AI ambassadors embedded throughout the organisation. Data masking, human review of customer-facing outputs, and ambassador validation provide controls around employee-created applications.
Within six months, weekly back-office AI use increased from 10% to 80%, while employees created almost 400 AI assistants now in daily use. The bank reports productivity improvements of up to 40% in some teams, with zero material incidents during the rollout.
As Deputy CEO David Davitashvili puts it: “The Data Analytics and AI team isn’t a support function; it’s one of the main drivers of our success.”
Bank of Georgia’s team has increased its impact by distributing capability while retaining accountability for the outcomes that matter. Data and AI have consequently moved beyond something a specialist team delivers for the bank. They are increasingly capabilities that the team enables the whole bank to use, govern, and turn into measurable value.
