Ankur Patel from Multimodal on Building Agentic AI for the Enterprise

The Evolution of AI in Enterprise

• 💡 Ankur Patel's journey from bond trading at Bridgewater to fintech highlighted the early impact of data science and NLP.
• 🚀 The realization in 2023 was to automate larger workflows and processes with AI, moving beyond simple task automation.
• 🧠 Large Language Models (LLMs) demonstrated the ability to think and reason akin to humans, paving the way for agentic AI.

Multimodal's Vision for Agentic AI

• 🎯 Multimodal is designing AI labor and workers to perform complex tasks in financial services, starting with lending workflow automation.
• 🛠️ Unlike previous "tool" startups, Multimodal aims for end-to-end automation, with AI agents working alongside human teams.
• ✅ AI agents operate within human-defined guardrails and benefit from human feedback to refine their reasoning and decision-making.

Addressing Enterprise Challenges

• 🔒 Explainability is crucial for enterprise adoption, with AI agents providing clear logs to trace decisions and interventions.
• ⚠️ Key objections from compliance teams include information security (data storage, access) and governance (model training, bias, trust in autonomous outputs).
• 📊 Deploying AI in government faces similar hurdles to large financial institutions, including longer procurement cycles and vast data scales.

AI Pricing and ROI

• 📈 The market is shifting from seat-based SaaS to usage-based or value-oriented pricing for AI agents, reflecting their role as "AI labor."
• 💰 Multimodal prices by workflow with credit allocations, balancing budget predictability with consumption-driven value creation.
• 🤝 AI agents are primarily seen as an efficiency play to enable existing staff to do more, rather than a direct means to reduce headcount.

Strategic Insights for AI Startups

• 🔑 Founders must conduct thorough market research on pricing and understand customer-specific priorities to avoid common mistakes.
• 🚀 A sales-led motion is vital for enterprise AI, complemented by channel partnerships once initial direct sales success is established.
• 💡 Vertical AI companies are positioned to win by focusing on niche, domain-specific problems that large horizontal players struggle to address effectively.