Tech • IA • Crypto
Customer Ignite Talk: Ravneet Shah (CTO, Allica Bank) & OpenAI
8/10TL;DR
Allica Bank is rapidly scaling AI across its operations, using agent-based systems and new team structures to accelerate lending decisions and product development while maintaining human-led relationship banking.
KEY POINTS
AI Adoption Across the Organization
Allica Bank, a UK-based SME lender founded in 2019, began its AI journey in 2023 through experimentation before defining a clearer strategy. Adoption has since expanded significantly, with internal usage rising from about 25% to a 77% median workday rate. The bank emphasized cultural change, requiring employees across operations, product, finance, and distribution to adapt how they work with AI tools.
Shift to “Squadlets” and New Operating Models
The bank restructured its product engineering teams, moving from traditional cross-functional squads to smaller units called “squadlets.” These teams are more flexible and tailored to product complexity, reducing handoffs and enabling faster delivery. Roles have been merged, combining backend, frontend, testing, and even product responsibilities into broader “product engineer” functions.
Blurring of Technical and Product Roles
Allica is encouraging T-shaped and multi-skilled roles, where employees develop expertise beyond their core specialization. In some teams, product managers and analysts have been consolidated, and designers and product staff are expected to contribute directly to production code. The goal is for non-engineering roles to deploy code independently by the end of the year.
High Deployment Velocity
The new structure enabled more than 3,700 deployments in a single year, a high figure for a relatively small team of roughly 100 engineers within a 200-person product organization. The bank aims to double this output while focusing on meaningful product improvements rather than raw deployment counts.
Agent-Based Lending Automation
Lending, the bank’s core business, has been a primary focus for AI. Instead of forcing brokers and customers to adopt rigid digital workflows, Allica introduced AI agents that process unstructured inputs such as emails. These agents extract information, identify missing data, and request clarifications before feeding applications into internal systems.
Faster Credit Decisions
By combining deterministic and non-deterministic AI agents, the bank has reduced some lending decision times to under 7 to 12 minutes. This represents a major improvement in a traditionally manual and complex process, particularly in asset finance where incomplete applications are common.
Customer-Centric Approach to AI
Rather than forcing behavioral change on customers, the bank has focused on adapting its systems to existing habits. This includes accepting email-based applications and enhancing them with AI, reflecting a strategy of meeting customers “where they are” instead of requiring full digital adoption.
Preserving Relationship Banking
Despite widespread AI integration, Allica has deliberately avoided replacing relationship managers with chatbots. Relationship banking remains a core differentiator, and AI is used instead to augment human staff with better insights and contextual information about clients.
AI as a Context Engine for Staff
AI tools provide relationship managers with synthesized customer data and insights, reducing time spent gathering information. This allows staff to focus on higher-value interactions and more informed conversations with clients.
Focus on Incremental Product Value
The bank’s forward strategy emphasizes increasing both customer-facing and internal product increments, including improvements in risk, compliance, and security. The goal is to enhance service quality while maintaining speed and efficiency.
CONCLUSION
Allica Bank’s approach illustrates how regulated financial institutions can scale AI by reshaping teams, embedding agents into workflows, and augmenting rather than replacing human roles, with a focus on faster decisions and improved customer outcomes.