Decision Velocity: Why AI Won't Fix Your Broken Process

The Core Argument

Guided by host Sanjay Saini, Agile Coach Joel Bancroft Connors (known as the "Gorilla Coach") argued that chasing pure delivery speed via AI is a fatal trap if an organization's underlying processes are fundamentally broken. Adopting AI into a disjointed workflow is like trading a beat-up car for a sports car while continuing to drive on a pothole-filled road—you just crash faster. True business agility relies on "Decision Velocity": the ability to make a well-informed decision that sticks permanently without triggering costly rework cycles down the line.

“Speed is what you get once you remove the friction, not the goal itself. AI is tempting leaders to skip the friction removal, which just means bad decisions happen faster.” Joel Bancroft Connors (03:54)

Detailed Answers

1. Could you give a brief intro about yourself and your background?

Joel Bancroft Connors has spent the last 30 years helping organizations confront the "gorillas in the room"—the massive structural and cultural problems nobody wants to acknowledge. Ranging from solopreneurs to Fortune 500 giants, Joel consults on smoothing out complex friction points between people, processes, and technology.

2. What is decision velocity and how is it different from speed?

Speed is simply the raw pace of output. Decision velocity is defined by whether a decision actually sticks once it's made. If a team makes a rapid decision but is forced to reverse it, re-litigate it, or entirely rework the feature three handoffs later, their decision velocity is effectively zero despite their perceived execution speed.

3. Why does adding AI to a broken process lead to failure?

Joel shared a case study of a data warehouse team struggling with a 23-week lead time defined by extensive sign-offs and a 30% rework rate. If management had replaced the database engineers with AI agents, the overall delivery speed would only improve by 5-10% because the administrative roadblocks and approval bottlenecks remained completely untouched. Adding AI to a fractured system merely accelerates failure.

4. What is the difference between consent and consensus in decision making?

Consensus demands that everyone actively agrees, often dragging out meetings and forcing reluctant compliance through majority rule. Consent-based decision making relies simply on checking if anyone fundamentally *objects*. Techniques like "thumb-voting sideways" (meaning "I'll go with the group") give power to the people who hold genuine objections (thumbs down), revealing hidden roadblocks early without stalling momentum.

5. How does a friction audit expose organizational bottlenecks?

A friction audit consists of interviewing leadership, observing actual on-the-ground workflows (not just documented ones), and running targeted retrospectives to hunt for delays. The output is a highly actionable "Friction Report" that calculates exactly how much time, money, and focus an organization is bleeding due to specific bureaucratic handoffs and unaddressed technical debt.

6. What is cognitive grunt work and how does AI solve it?

Cognitive grunt work includes the mentally draining tasks we hate doing but must execute daily: scanning hundreds of emails, categorizing feedback, or sorting through Slack notifications to build a to-do list. By configuring personal AI agents (like Joel's 7 A.M. "Morning Report"), professionals can automatically filter and summarize these noise-heavy tasks, reclaiming hours of high-level strategic focus.

7. How should leaders actually start their AI transformation journey?

Before buying enterprise licenses, leaders must ask: "Where do decisions currently get stuck, reversed, or reworked?" They must map out their value streams and pinpoint the exact friction areas. Rather than commanding AI to "fix the code," leaders should feed the process constraints into the AI and ask the AI what clarifying questions it has for *them*, instantly uncovering forgotten workflow complexities.

8. Why is standardizing on a single AI tool a terrible strategy?

Just as human teams thrive on cognitive diversity, technological stacks require diverse tooling. Standardizing an entire enterprise exclusively on Copilot limits problem-solving capabilities to one isolated ecosystem. Joel advocates for utilizing a diverse toolkit—using Claude for analytical orchestration, Gemini for native integration queries, and Perplexity for deep data research—ensuring you use the right tool for the right job.

Stop optimizing broken systems. Learn how to conduct friction audits, streamline value streams, and lead authentic AI transformation in our Agile Leadership and Change Management Training Course.