The AI Problem Nobody Talks About—You Can’t Even Turn Them Off | Ep. 397 with Brandon Card CEO of Terzo AI

Description

Daniel Robbins interviews Brandon Card, the CEO of Terzo AI, about the hidden financial chaos inside enterprise contracts and why AI is the only scalable way to fix it. Brandon explains how Terzo helps companies treat contracts like financial assets, not legal documents, extracting obligations and commitments with 99.9% accuracy through a hybrid model of AI plus trained human review. They also discuss why Terzo started by selling into Fortune 100 instead of SMB, how early customer pain shaped product-market fit, and why Brandon is equally focused on building community and mental health resilience as he is on revenue. Key Discussion Points Brandon explains the origin of the name Terzo, inspired by “third” in Italian and the idea of managing third-party relationships, plus the “Oracle effect” of a timeless name beyond one product. He recounts incorporating the company on March 13, 2020, then watching airports go empty as COVID hit, building on Zoom 18 hours a day through chaos and loss. Brandon describes the pain he lived at Microsoft: customers managing $50B supplier spend with no tools, contracts lost, and manual PowerPoint decks built from screenshots to summarize global agreements. He argues CLM systems were built by lawyers for drafting, not for procurement and finance workflows, leaving a massive gap no one understood until recently. Brandon shares the early breakthrough: realizing contracts are 300–400 pages long and signed by the thousands, making human-only review mathematically impossible. Terzo’s key differentiation is accuracy through humans-in-the-loop, because LLMs can miss financial context across related documents and even turn $1M into $3M in naive extraction. He explains why they started enterprise-first, taking on SOC, GDPR, and security onboarding to win Fortune 100 master service agreements rather than “credit card swipe SaaS.” Brandon shares his view on AI safety: he believes models behave like they have a “mind of their own,” and that even builders struggle to govern or fully control them. Th