Shipping AI is one thing. Shipping AI that actually works in production is something else entirely. Last week, we brought together @AnthropicAI and @_StacksAI at our London office for “AI Agents in Production — Shipping AI that actually works,” a candid conversation on what it takes to bridge that gap. Adriaan Engelbrecht, Applied Engineer at Anthropic, highlighted the importance of evaluating AI models tailored to specific use cases rather than a one-size-fits-all approach. Ivan Bovyrin, Founding Engineer at Stacks, stressed that human-in-the-loop patterns are key to gathering customer feedback and improving agent performance. Both speakers emphasized that trust comes from deterministic, reliable systems. It was inspiring to be in the room alongside the engineers and builders at the forefront of Applied AI. Conversations like these continue to move the needle.

The next major frontier for AI exploration might be for the Office of the CFO. This week on The Investment Memo, @_StacksAI Co-Founder and CEO @albertmalikov joins @alex_schmitti to explain the "perfect storm" hitting finance teams, from accountant shortages, fragmented data, and workflows still anchored in the past. Stacks is taking a different approach with a strong data foundation and systems designed for accuracy. It’s a clear look at how an AI-first finance stack could help set a higher bar for efficiency and accuracy. 0:00 Accounting Workflows & Quick ROI 0:49 TIM Introduction 1:42 Why the CFO Office Is Ripe for Disruption 3:41 Monthly Close & Data Fragmentation Problem 7:31 Why AI Adoption in Finance Is Hard 11:31 Stacks Origin Story & Product Vision 19:02 Real Customer Use Cases & Impact 26:22 Series A, Culture, and Future of Finance