What Northeastern's AI Ready or Risky Session Taught Me About Moving AI From Pilot to Value

Plenty of organizations have identified an opportunity for AI. The harder question is whether the business is ready to turn that opportunity into measurable value. That was the premise of AI Ready or Risky, an executive working session Northeastern University hosted on October 1 during Boston AI Week.

The format set it apart from a typical conference panel. Each executive in the room brought one active AI initiative and spent the morning pressure-testing it with Northeastern researchers and industry experts. After a fireside conversation on moving from experimentation to measurable value, small cohorts rotated through expert teams covering business value, people, data and governance. By noon, each of us had an AI Readiness Card showing where we were ready to move forward and where gaps remained.

I brought Greenough's agency-wide Claude rollout, with my focus on business readiness and people management. Our clients hire us for our strategy and judgment, so the rollout has to strengthen that zone of genius as much as it speeds up the work. My key takeaways centered on how leaders choose the right AI project, how they keep pilots from stalling and why governance is the missing piece.

Choosing the Right AI Project Starts With Enterprise Value

The first question for any AI initiative is whether it's worth pursuing at all. The answer depends on enterprise benefit, as individual productivity gains rarely make the case on their own. Choosing the right type of AI matters just as much, and generative AI isn't always the right fit. That choice starts with a real understanding of the pain points your trying to solve and the outcome you want. The products and processes you identify for automating may be the wrong targets, and sometimes the larger business model needs rethinking first.

The same discipline applies to vendors. AI-washing and agent-washing are common, so ask which underlying technology a product uses and for what purpose. Agents, which ar gaining popularity right now, make decisions, but that’s not always the best option for your project. Ask whether the task needs autonomous decisions at all and, if it does, take the next step to decide how you would manage the agent, set its guardrails and shut it down if needed.

AI Pilots Need Clear Metrics and a Decision Point

The panel put it plainly: if you're loose on your metrics, you could live in constant pilot purgatory. A strong pilot shows value in sequence and ends at a clear decision point, including when to abandon it and move on.

At Greenough, our six-week AI certification with Pete Pachal of The Media Copilot ends in a demo day where each team member presents a workflow built for a real program, so every experiment has an owner and a finish line. The efficiency those workflows free up is where the larger opportunity sits, because it gives us time to think more strategically about our clients' markets.

AI Governance Is the Missing Piece

Governance came up in every conversation. OneTrust's 2026 AI-Ready Governance Report found that 96% of organizations had at least one AI initiative slowed by oversight requirements, and only 5% said accountability was clearly defined across the AI lifecycle.

That accountability gap stayed with me. When AI goes wrong, responsibility diffuses between the engineer who built it and the manager who approved it. One speaker compared it to driving. We install seatbelts and airbags, but if you get behind the wheel impaired, the outcome is on you. Greenough's AI Charter commits us to AI-powered, human-led work, and we treat every number, name and link that comes through AI as unverified until a person checks the source.

Building Judgment in Early-Career Talent: Join Me December 1

In the session, I asked how people develop judgment when AI takes over the work they traditionally learned on. That question became Northeastern's next AI in Action conversation. AI can now take notes and build media lists, and the year of watching and learning is giving way to a faster path. It has me rethinking how I teach what good looks like and how to create safe places to fail. Learning by doing is still critical, so our training and mentorship have to change in parallel with the work.

I'll join Devra Henderson of Veteran Hiring Solutions and Tracey Spinato of the Roux Institute as a panelist for When AI Changes Entry-Level Work, How Do People Learn? on Tuesday, December 1 at 1 p.m. ET on Microsoft Teams, moderated by Katie Shorey of Live + Work in Maine. Register here.

‍