6 Levels of Autonomous Work in Financial Services
Like many others, I could not understand the difference between “assistants,” “copilots,” and “agents.” I am glad I stumbled upon Sarah Hinkfuss ' enlightening article on the 6 Levels of Autonomous Work! It was like a lightbulb moment.
Product Managers love frameworks. And I am using this newfound understanding to map out how these different AI capabilities apply to jobs across financial services.
Level 0: No Automation
Fintech operations remain entirely human-driven. For example, in-person banking services such as safe deposit access or complex advisory services for wealth management.
Level 1: Worker Assistance
Examples could be automating repetitive tasks such as verifying customer information against databases or processing standard transactions.
Level 2: Partial Automation / AI Assistant
AI assistants enhance customer interactions by providing first-level support, answering FAQs, or guiding customers through application processes online. For example, chatbots answering account balances, or transaction histories, as seen in every banking app now. Other examples include initial credit assessments to pre-qualify loan requests before human review.
Level 3: Conditional Automation / AI Copilot
Use of AI in managing compliance and regulatory reporting, where AI tools track and report transactions in real-time, alerting human operators to anomalies that suggest fraud or errors
Level 4: High Automation / AI Agent
Automated wealth management services (robo-advisors) was probably the first example in this category. And the future could be AI-driven platforms like those handling automated tax filing and optimization for individuals and businesses.
Level 5: Full Automation
A fully automated digital bank that uses AI to conduct all its operations, including customer onboarding, KYC (Know Your Customer) compliance, credit scoring, loan disbursement, customer service, and even complex advisory services. Or, a fully automated insurance entity that uses AI to assess applications, manage policies, process claims, and handle customer interactions through advanced natural language processing. It's hard for me to see this now, but that's the end game with Full Automation.