Send us a text

Most enterprise knowledge lives in chaos—PDFs, email threads, scans, recordings—and for years that chaos resisted automation. We sit down with Bernard Aceituno, Co-Founder of Stack AI, to unpack how generative models and agentic systems finally tackle the messy 90% of data and convert it into measurable actions.

Bernard draws a sharp line between keyword search and real understanding, showing how reasoning traces and tool use enable AI to read, decide, and act—sending the email, filing the claim, extracting the clause, and updating the record.

We go deep on what “agentic” really means in plain language: planning steps, choosing tools, and executing with auditable logs. From banks and hospitals to defense and civil engineering, Bernard maps where AI agents are already reducing handle time and error rates across collections, payments, compliance, and business development. He also outlines near-term constraints—accuracy under pressure and context window limits—and how retrieval, external memory, and better infrastructure are closing the gap.

The conversation demystifies AGI as a reliability benchmark for specific tasks and frames superintelligence as both a risk and an opportunity frontier, especially in drug discovery and complex research.

If you’re wrestling with legacy systems, data silos, or unclear starting points, this episode offers a pragmatic roadmap: target high-friction workflows, layer agents with guardrails, measure outcomes, and expand. Ready to turn unstructured data into operational advantage and move beyond demos to dependable ROI? Follow the show, share with a teammate who owns the process you want to fix, and leave a review with the one task you’d hand to an agent first.