After decades of investing across emerging economies, the European Bank for Reconstruction and Development is using Microsoft AI to make its knowledge easier to use – helping teams learn faster, avoid repeated mistakes, and make better investment decisions that benefit more people.
The European Bank for Reconstruction and Development (EBRD) was founded in 1991, in the aftermath of the Cold War, with a mission to support countries moving toward open, market-orientated economies.
Since then, it has invested more than €220 billion (£191 billion) across nearly 8,000 projects, supporting businesses, energy systems, financial institutions, and infrastructure in more than 30 economies around the world.
Projects – from financing solar farms in Egypt to assessing private sector reforms in Central Asia – have been evaluated using interviews, data, and on-the-ground evidence to understand what worked, what didn’t, and why.
That’s a lot of useful knowledge buried in the archives. But there was a problem, says Véronique Salze‑Lozac’h, former Chief Evaluator at the EBRD.
“Very often people don’t read the reports,” she says. “We have been publishing for decades – but we were battling to make sure all of this wealth of information was not lost.”
Reports can run to scores of pages (and longer with annexes), and finding relevant insights meant searching across dozens – sometimes hundreds – of documents. Previous attempts to capture lessons in databases had limited success: they were difficult to maintain and rarely used in practice, she says.
AI to the rescue
To solve this, the EBRD brought together its independent evaluation team, information management specialists, and its AI Centre of Excellence.
The result was ‘LessonsBot’ – a tool built on Microsoft Azure that helps staff find and use insights from across the bank’s vast library of evaluation reports.
“It’s a great example of cross‑departmental collaboration,” says Tamara Karam, Head of Information Management. “Without that, it wouldn’t have been possible.”
When a user asks a question about a country, sector, or type of investment, the tool searches a curated set of more than 300 reports from EBRD and other development banks, such as the World Bank and Asian Development Bank.
“From a simple prompt, the user gets a response with inline references,” says Abantika Chatterjee, Associate Director leading the EBRD’s AI Centre of Excellence. “You can click through to the exact page in the underlying report.”
The tool finds relevant material and summarises it so users can see both the conclusion and the evidence behind it.
“The responsibility lies with the human,” adds Gabrielle Bex, who leads adoption and training. “The AI can guide you, but you need to fact‑check and follow the references.”
To support this, staff are trained to treat the tool like a new colleague: knowledgeable, but needing clear direction and careful interpretation.
From months to days
Lessons Bot’s biggest impact has been speed.
“If we’d gone through all of our reports, it would have taken months,” says Salze‑Lozac’h. “With the Lessons Bot, it takes us just a few days.”
In one case, the evaluation team analysed more than 80 reports to examine how the EBRD ensures its investments add value without competing with commercial banks.
In another, the bank used Lessons Bot to support its expansion into Sub‑Saharan Africa. By drawing on evaluations from other development institutions, the team quickly identified critical success factors – such as the importance of pairing investment with support and training for local people, and recognising the sometimes-fragmented nature of local markets.
When you’re operating across diverse regions – from Central Asia to North Africa, where political, economic, and institutional contexts can vary dramatically – every past project contains lessons that can help make future ones more effective.
The challenge has never been creating that knowledge – it has been using it.
Lessons Bot changes that, says Salze‑Lozac’h. It turns institutional memory into something accessible, searchable and practical – helping teams move from “we think this might work” to “we know what has worked before, and why”.
LessonsBot users have certainly appreciated the easier access to knowledge.
One researcher commented: “There’s such a rich database of knowledge going back through the years… it’s never been so accessible or even usable.”
And better decisions at the EBRD translate into more effective investments – supporting stronger businesses, more resilient infrastructure, and more inclusive economic growth in the countries it serves.
“The work of the EBRD is as complex as it is important,” says Samer Abu-Ltaif, who oversees the Microsoft business across Europe, the Middle East and Africa.
“By using AI to put trusted insights at their teams’ fingertips, EBRD is helping people make better informed investment decisions and direct capital to projects that deliver real impact for communities all around the world.”
Around 68% of users are now frontline bankers, Salze‑Lozac’h says, and internal survey feedback suggests around three quarters report increased productivity, reduced manual effort, and improved quality of insight.
Deeper integration
Since its launch in 2024, the tool has expanded beyond the EBRD and is now used by more than 10 other international financial institutions, with interest continuing to grow.
Inside the bank it sits alongside other AI-powered services, such as Microsoft 365 Copilot, which helps staff draft documents, summarise information, and turn insights into reports and presentations. Lessons Bot searches for evaluation evidence, while Copilot supports everyday work.
The next step is deeper integration – developing more specialised tools that can support different teams and more complex decision‑making.
For the EBRD, the ambition is clear: making knowledge easier to find, easier to trust, and easier to use will lead to better outcomes for the people it serves.
The collaboration with the EBRD extends beyond LessonsBot.
Both organisations recently signed a Memorandum of Understanding establishing a strategic partnership to accelerate Africa’s start-up ecosystems through AI adoption.
Microsoft, through the Africa Transformation Office, will provide start-ups supported by EBRD with access to Azure AI, advanced AI frameworks, and global subject matter experts to help drive their growth, with a view to delivering measurable economic impact and positioning Africa as a frontier AI innovation hub.