On April 9, 2026, Revolut launched AIR - an AI financial assistant embedded directly in their app. Customers swipe down and ask "freeze my physical card" or "how much did I spend on subscriptions last month?" Three weeks earlier, Starling Bank shipped what it called the UK's first agentic AI financial assistant. HSBC, Barclays, and NatWest are all signing deals with AI labs to build their own.

You can ship the same thing in days using Lorikeet's SDK.

Not a chatbot. Not FAQ search. A fully embedded AI agent that takes real actions inside your product - freezing cards, setting up transfers, checking insurance - the exact capability the biggest fintechs in the world just made headlines for.

What they actually built

Strip away the press release and AIR is a straightforward architecture. A conversational interface sits inside the app. A customer types a request. The system pulls their data, reasons about it, and takes an action.

Chatbots summarise help articles. Agents do things. That distinction is what makes these launches significant, and it's what every fintech customer is about to expect as standard.

Revolut first teased this at their "Revolutionaries" event in November 2024. Starling describes theirs as the culmination of eight years of AI development. Both are UK-only, with other markets "coming soon."

How Lorikeet's SDK works

Lorikeet's SDK is a REST API that lets you embed a fully capable AI agent inside your product. Your engineering team builds the interface - a chat bubble, a swipe gesture, whatever fits. The SDK handles everything else: natural language understanding, action execution, compliance guardrails, multi-turn conversations, streaming responses.

Your product, your experience, your brand. The AI is invisible infrastructure. Exactly the architecture Revolut built for AIR - except you plug in rather than build from scratch.

The question most fintechs ask next is: how do you actually get the agent to do useful things? Connect to your systems, follow your policies, handle your edge cases?

That's where Coach comes in. Coach is Lorikeet's AI-powered setup assistant. Describe what you want your agent to handle - "when a customer asks to freeze their card, verify their identity, freeze it, and offer a replacement" - and Coach generates the workflow, the API connections to your backend, the compliance guardrails, and the test scenarios to validate it all works. What used to take weeks of configuration is now a conversation.

You don't need to worry about compliance

Lorikeet is SOC 2 and ISO certified. The patterns for financial services - data handling, PII protection, audit trails, escalation rules - are built into the platform and tested in production.

Fintechs like Airwallex, Self, Taptap Send, and Step already run Lorikeet agents across their customer support operations. They handle real financial queries with real compliance requirements every day. The platform works for fintech because it's been shaped by fintech.

You don't need to solve compliance from scratch. You configure it for your specific policies, and Coach helps you do that too.

The compounding advantage of going first

Revolut is UK-only. Starling is UK-only. The traditional banks are still in partnership announcement territory. If you're operating in the US, Europe, or APAC, your customers haven't seen this yet.

You have a window. Ship now and by the time the big players roll into your market, your agent has months of real conversations behind it. Months of edge cases handled, workflows refined, escalation paths tested. That compound advantage is hard to replicate once someone else has it.

Of course, speed alone isn't enough. Klarna learned that the hard way - their AI assistant handled 2.3 million conversations in its first month, but quality collapsed and they ended up re-hiring human agents by mid-2025. The conversational interface is the easy part. Getting the agent to reliably resolve problems across thousands of scenarios is the hard part, and it's the part you shouldn't be building from scratch.

Revolut spent a year. You don't have to.

Embed a capable agent into your existing product. Connect it to your systems. Let Coach build the workflows. Deploy in days.

Your customers don't know or care what's under the hood. They care that it works.