Jorge Alcantara

Correct Code, Wrong Product

A coding agent can produce clean code that passes every test and still violate the customer requirement behind the work. The talk walks through the decision record that keeps that requirement attached: the source evidence, the named owner, and the constraints the agent has to respect.

Product, engineering, and developer-tool audiences working with coding agents.

• Spot the point where technically correct work loses the customer requirement.
• Build a decision record with source evidence, an owner, and explicit constraints.
• Give agents enough context to build without giving them the product decision.

Build the Harness That Outlives the Agent

Jorge shows how Zentrik divides work between people and agents, then builds a harness that the whole team can improve as models and tools change. The point is to go beyond faster coding and keep the work connected to what customers need.

Product, design, engineering, and marketing teams working with agents.

10% vs 10x: Are You Saving Time or Building a Stronger Team?

Most AI programs promise to save time. Time savings stop at 100%; a company’s capacity to create value has no equivalent ceiling. This session separates cheaper old workflows from product systems that learn, decide, and deliver better with every cycle.

Executives and product leaders deciding how AI should change the operating model.

• Distinguish task automation from a stronger product operating system.
• Find the review gates where human judgment changes the outcome, and drop the ones that only add delay.
• Measure whether AI improves the next decision, not only the time spent on this task.