How Ramp built an AI agent that can think outside of tokens | Alex Shevchenko

Alexander Shevchenko is the head of applied research at Ramp, where he leads Ramp Labs – the team behind Ramp Sheets and a steady stream of public AI engineering experiments. Ramp Sheets started as an internal process mining tool that turned Loom videos of accountants into Markov diagrams, before evolving into the agentic spreadsheet editor that shipped in November. In this conversation, Alex walks through the architecture under the hood, why Ramp biases the agent toward Excel formulas over Python code gen, and two recent Labs experiments: Latent Briefing and a user-steerable revival of Golden Gate Claude. We also discuss: • Under the hood of Ramp Sheets • Inspect, Ramp's internal coding agent, and the self-improving monitor loop it powers • Why finance professionals rejected code gen as too "black box" • Why Anthropic models tend to excel at agentic spreadsheet manipulation • The case for putting the agent outside the sandbox, not inside it • The Loom-to-Markov-diagram process mining pipeline • RLMs and how subagents can share memory in latent space • Latent Briefing and KV-cache communication between subagents • Reviving Golden Gate Claude with steering vectors on Gemma Referenced: • Alex Levinson: https://www.linkedin.com/in/alex-levinson/ • Anthropic: https://www.anthropic.com/ • Ben Geist: https://www.linkedin.com/in/benjamin-geist/ • Claude: https://www.anthropic.com/claude • Efficient Memory Sharing for Multi-Agent Systems via KV Cache Compaction (Ben Geist): https://x.com/RampLabs/status/2042660310851449223 • Gemma: https://ai.google.dev/gemma • Golden Gate Claude: https://www.anthropic.com/news/golden-gate-claude • Graphviz: https://graphviz.org/ • Inspect: https://builders.ramp.com/post/why-we-built-our-background-agent • Latent Briefing: https://x.com/RampLabs/status/2042672773747589588 • Loom: https://www.loom.com/ • Modal: https://modal.com/ • OpenAI: https://openai.com/ • Opus: https://www.anthropic.com/claude/opus • Qwen: https://qwen.ai/ • Ramp: ht