Tabs & LangChain: Building Ambient AI Agents for B2B Revenue Operations

Tabs: Vertical AI for B2B Revenue

• 🎯 Tabs is a vertical AI company focused on B2B revenue intelligence, aiming to get money into bank accounts faster and report accurately.
• 💡 The company extracts key information from sales contracts to run revenue operations, improving cash flow and streamlining processes.
• 🚀 Tabs pivoted into this specific vertical after the ChatGPT moment, recognizing the commoditization of general document information extraction.

The Commercial Graph & Agentic Workflows

• 🧠 Tabs is building a "commercial graph", a unique data model that hydrates all information around merchant and customer relationships.
• 🛠️ This graph enables fully intelligent workflows to automate post-contract processing, invoice generation, and revenue collection.
• 📈 The company is transitioning from guided AI experiences to fully agentic workflows, aiming for a headless operational software experience.

Understanding Ambient AI Agents

• 🌌 Ambient agents are distinguished from background agents by being event-triggered and running autonomously in the background, rather than human-initiated.
• 💬 The vision for Tabs is a lean internal finance team supported by ambient agents handling day-to-day tasks, with communication often occurring through platforms like Slack or email.
• 🔍 The end goal is to build beautiful operational software that users rarely need to interact with directly, allowing human teams to focus on strategic work.

Challenges and Future of Agent Memory

• ⚠️ A primary challenge in building agents is information retrieval, specifically knowing what data is most relevant from a vast knowledge graph for a given task.
• 🧩 Structuring and pruning unstructured data from various sources (customer sentiment, usage, contracts) is crucial for effective agent performance.
• 🧠 Memory and learning from human interactions are considered a significant moat for vertical AI companies, allowing agents to improve and avoid repeated corrections.