Welcome to the Agentic Era: Meet Ada — The AI Underwriter Built for Today’s World

Generative AI isn’t the future. It’s already here. And it’s not just changing how we write emails or summarise meetings, it’s fundamentally reshaping how complex decisions get made in business. At Liberis, that means using AI to completely transform how we underwrite funding.

Introducing Ada, our in-house AI credit decisioning agent that’s already making thousands of funding decisions in production. Fast, fair, explainable, and trained on real data, Ada represents the third pillar of our embedded finance evolution.

Even with years of investment in automation, there’s always been a subset of funding decisions that required manual review — higher-value deals and complex documentation. Before Ada, these cases often sat in underwriting queues, waiting for a human to comb through PDF bank statements, scanned VAT returns, or grainy receipt images.

It was slow. It was resource-intensive. And it wasn’t scalable.

In fact, manual underwriters were spending up to 90% of their time just searching through documents to locate relevant data points. The decision itself was often quick, it was the digging that took forever.

Ada was built to fix that. It’s powered by Google’s Gemini model and trained on tens of thousands of historic Liberis deals. It knows our credit policy. It understands what a healthy deal looks like. And it’s designed to surface decisions instantly, with complete transparency and traceability.

Here’s how it works:

• Speed: Ada reviews all submitted documentation in minutes, not hours, and delivers a clear funding recommendation straight into the underwriter’s dashboard.
• Clarity: Every risk factor and positive signal is listed and clickable. Underwriters can trace exactly where each data point came from.
• Conversation-ready: Underwriters can ask Ada direct questions about a case. If context is missing, Ada can even search Google in real time, summarising external data alongside internal insights.
• Human in the loop: Ada doesn’t operate in a black box. Underwriters review, validate, and provide feedback back into the system, helping Ada learn and improve with every decision.

This loop of continuous improvement is key, and it’s already paying off.

Since launching Ada into live production in May, thousands of funding decisions have been made using her recommendations. So far:

• There’s a 68% alignment rate between Ada’s suggested decisions and final human decisions.
• Approval decisions are even more closely aligned, with Ada showing strong risk calibration.
• Underwriters report a massive reduction in time spent processing each case, freeing them up to focus on edge cases, support, and strategy.
• And importantly, humans always have the final say. Ada supports; it doesn’t replace.

As Google continues to improve its foundational models and as Ada continues to learn from real-world results, this alignment is only getting better and faster.

For small businesses, this means faster access to funding, often the same day.

For underwriters, it means less admin and more meaningful decision-making.

For partners, it means more consistent approvals and better outcomes across the board.

Ada isn’t just an AI tool. She’s a core part of how Liberis is building the next generation of business finance, one that’s fast, personalised, and radically efficient.

In the penultimate session in the series, we look at insights from Google, exploring how generative AI is evolving across industries, how Gemini is setting the pace, and what the next 12 months will bring for businesses ready to embrace intelligent systems.