🤝 Join our FREE course on Feb 11th: How to Avoid AI Project Failures For Org Leaders

Most AI pilots don’t fail because of the tech.
They fail because orgs fund the wrong things, in the wrong order.

In this 30-minute Lightning Lesson on Maven, I break down:

• why AI pilots stall before ROI
• the decision mistakes that quietly kill momentum
• a simple way to prioritize AI bets that actually stick

→ Reserve your Free seat (limited)

Weekly finds

📰 AI Insight

1. The AI Super Bowl Showdown Anthropic’s $8M ad mocked OpenAI’s new ChatGPT ads. Sam Altman called it “dishonest,” but the internet kept the receipts. Grab the popcorn; the gloves are off.

2. OpenClaw: AI with Hands This open-source agent racked up 150k GitHub stars by actually doing tasks like booking flights. Security experts call it a “nightmare,” but developers call it “hired.” 3

3. The “SaaSpocalypse” is Here Software stocks lost $285B after Anthropic launched legal plugins. When AI can handle document review and compliance, traditional SaaS starts looking like a typewriter. 7

4. Sam Altman’s AI Successor Altman revealed a plan to eventually hand OpenAI’s leadership to an AI model. If the CEO is an LLM, does it still get a golden parachute?

5. AI Catches 27% More Cancer A Swedish study found AI spots significantly more aggressive breast cancers while halving radiologist workloads. Finally, an AI “taking jobs” that we can all celebrate. 10

THE ISSUE

The 4 Pillars of AI Transformation

I’ve never met a leadership team that didn’t think they were taking AI seriously.

I have met plenty who were six months in, $500K lighter, surrounded by impressive demos, and quietly wondering why nothing important had changed.

That gap doesn’t come from bad tools. It comes from skipping the hard, unglamorous work. The work that doesn’t fit neatly on a roadmap slide.

Everything that actually matters in AI transformation comes down to four pillars. Not as a framework. As behavior.

Let’s walk through them the way they actually show up inside real organizations.

Pillar 1: Strategic Clarity

Here’s the moment where things usually go sideways.

Someone says, “We should be doing more with AI.”
Everyone nods. A pilot gets funded. A tool gets selected.

And nobody can answer, cleanly, what success looks like.

Start with one sentence. Seriously. One.

Every AI initiative needs a single sentence that explains why it exists in business terms. Not technical terms. Not aspirational ones.

When I ask leaders for this, I usually get something like:

That’s not a why. That’s an activity.

A real one sounds more like:

Now here’s the part most people skip.

Take that sentence and share it with three peer leaders. Not your AI team. Not the vendor. People who run real parts of the business.

Ask them two questions:

• Is this clear?

• Is this compelling?

If they hesitate, even politely, you don’t have clarity yet. And if you don’t have clarity, everything downstream will wobble.

Metrics: The Fastest Way to Expose Vagueness

Once the “why” is clear, metrics should feel obvious. If they don’t, that’s a warning sign.

A lot of teams default to project metrics:

• Adoption rates

• Usage numbers

• Number of pilots launched

Those are comforting. They’re also mostly irrelevant.

What you want are business metrics that change if AI actually works.

Things like:

• Resolution time dropping from 45 minutes to 27

• $2M in annualized cost reduction

• CSAT moving 15 points

Here’s my rule of thumb:
If you can’t tell whether you’re winning within 90 days, the metric is too vague.

Write them down. Put them on one slide. If leadership can’t react to them immediately, they’re not sharp enough.

Use Success as a Filter

Before approving any AI pilot, ask one question:

If the answer is “none,” “maybe,” or “eventually,” don’t fund it.

This feels harsh. It also saves an incredible amount of money and political capital.

Most AI backlogs are full of “interesting” ideas. Very few are tied to outcomes anyone would actually fight for.

Kill early. Focus ruthlessly.

Pillar 2: Ways of Working , Where Old Habits Quietly Kill New Tech

Here’s the uncomfortable truth: agentic AI does not work inside project-era operating models.

You can’t run it like a feature release and expect transformation.

Your Roadmap Is Lying to You

Open your roadmap. Look at it honestly.

If it’s full of things like:

• “Deploy chatbot”

• “Integrate LLM into CRM”

• “Roll out AI assistant”

…it’s describing activity, not change.

Now try rewriting each item as an outcome.

Instead of:

Try:

Anything that can’t be rewritten that way isn’t ready. Delete it or rethink it.

Roadmaps should tell a story about how the business gets better, not which tools show up.

Stop Asking What Shipped. Start Asking What Changed

Most teams are trained to report progress as output.

AI doesn’t care about output. It cares about learning.

One of the simplest changes you can make is replacing:

with:

Run a short weekly learning standup:

• What worked?

• What didn’t?

• What surprised us?

• What are we doing differently next week?

Document insights, not tasks. Over time, learning velocity matters more than shipping velocity.

Status Meetings Are Where Momentum Goes to Die

If every update sounds like:

you’re not learning. You’re narrating.

Change one recurring meeting into a learning update.

No “what we did.”
Only:

It feels subtle. It’s not. This shift alone changes how people think about progress.