Customer Ignite Talk: Ravneet Shah (CTO, Allica Bank) & OpenAI

TL;DR: Allica Bank CTO Ravneet Shah shares how the bank scaled AI across the organization—from 25% to 77% mid-week adoption—by rethinking team structures, merging roles, and using AI agents to accelerate loan decisions from days to 7–12 minutes while enhancing, not replacing, their relationship banking model. ## Introduction Ravneet Shah, CTO of Allica Bank, joined Clem from OpenAI's market expansion team to discuss how Allica—a UK-based digital bank for SMEs—is embedding AI into every aspect of its business. Allica offers lending, business current accounts, and deposits, and differentiates through a combination of technology and relationship banking. As one of the fastest‑growing challenger banks in the UK, Allica has taken a deliberate approach to applying AI across product development, relationship management, and loan operations. ## How Allica Thinks About AI at a Growth Stage Allica began its AI journey in 2023, making many mistakes along the way. The initial challenge was figuring out which use cases were appropriate. After years of experimentation, the team now has a clear view on scaling. Three key lessons emerged: 1. **Organization‑wide adoption** – Before learning new tech, the team had to unlearn old habits. The first mantra: to scale AI, everyone in the organization must adopt it and change how they work—across operations, distribution, technology, product, and finance. Adoption rose from about 25% to a mid‑week median of 77%. 2. **Building products differently** – The product‑engineering organization needed its own operating model. What worked for the whole company didn't work for product‑engineering teams; a specific operating model was created. 3. **Product itself is changing** – AI enables things that traditional ML and software applications couldn't. This adds a new layer inside the business. The team repeated internally: "We need to think differently when building products that use agentic applications." ## Reorganizing Team Structure to Enable Fast Innovation Allica’s product‑engineering group is not large—roughly 100 engineers, or under 200 colleagues including product, data, and design. Previously they used the Spotify model with cross‑functional squads (product, data, backend, frontend, design). AI forced a change. **Shift to "small teams"** – Teams are now smaller, with structure varying by complexity and product nature. Hand‑offs that used to be necessary are no longer needed. Allica embraced the concept of T‑shaped talent: deep expertise combined with adjacent skills. **Role merging** – Backend, frontend, and dedicated testers were combined into a single role. Product roles also merged: where there used to be a product manager and a product analyst, now the two are one. Some teams have a "product engineer"—someone who can both do product and engineering. By the end of the year, the goal is that all product, design, and engineering people can deploy code to production. This is already happening in some teams. ## AI in Lending and Underwriting Lending is Allica’s core business. The process is complex and not fully automated; many applications come via email from intermediaries and brokers. Rather than forcing customers to change their behavior, Allica adapted. They introduced an agent that can read email content, identify missing information, and request more details from the broker before the application enters the portal. By combining deterministic and non‑deterministic agents, some loan applications are now decided in under 7 to 12 minutes. The philosophy: where manual processes exist that software couldn't solve before, ask "Where can we use agents?"—with appropriate guardrails. ## AI for Relationship Banking (Enhance, Not Replace) Relationship banking is Allica’s unique selling point. The team debated how to use AI without replacing the human relationship manager. The conclusion: do not replace them with chatbots, but support them. AI can provide relationship managers with insights and context about customers, so they spend less time researching and more time driving meaningful conversations. It’s about giving them better context to move the conversation forward. ## Future Outlook: Next 12–24 Months Six months ago Allica was still experimenting. Now there’s a clear direction, especially in product‑engineering. Last year the team deployed over 3,700 projects. This year they aim to double that number—but not just for the sake of a metric. The real goal is to increase customer‑facing and internal product increments, including risk, compliance, and security. They want to maintain speed while keeping (or improving) quality. The overarching drive: better serve customers, move faster, and increase efficiency. ## Source https://www.youtube.com/watch?v=pcAtJDBO3hw