Susan Ferrari’s recent Martech article (link shared below) talks about a benchmark in the AI world called Humanity’s Last Exam. It was designed to be nearly impossible for AI models. When it launched in early 2025, the best models could barely scratch it. Eighteen months later, the leading model clears more than half of it.
Sit with that for a second. A test built to be unbeatable was half-solved in a year and a half.
Now ask yourself: when did your marketing operations team last re-evaluate the AI tools it standardized on?
For most B2B teams we work with, the honest answer is “when we picked them.” And that’s the problem.
The loyalty instinct is now a liability
For years, the smart play in MarTech was commitment. Pick your platform, learn it deeply, build your workflows around it, train the team, defend the investment. That made sense when platforms evolved on annual release cycles.
AI has broken that logic. The capability frontier now moves monthly, sometimes weekly. A workflow that required a specialist in January might be a checkbox feature by June. A model that couldn’t reliably segment your audience last quarter might do it better than your agency this quarter.
The teams that treat their current stack as the final answer are quietly accumulating a gap — the distance between what’s possible now and what they’re actually using. That gap is invisible on any dashboard, but it shows up everywhere: slower campaign cycles, more manual QA, higher agency costs, competitors who somehow ship faster.
The skill that compounds
Here’s the reframe that matters: the valuable skill was never the tool. It was always the ability to direct AI toward a useful outcome — a cleaner segment, a sharper nurture sequence, a faster data audit, a smarter testing plan.
That skill lives in your team, not in your vendor contract. And once you separate the skill from the tool, switching tools stops feeling like starting over. You’re pointing the same judgment at a sharper instrument.
The discipline this requires is surprisingly small:
One check-in, on a schedule. Ten minutes a week. Scan for what’s genuinely new — not incremental, new.
A simple filter. Does this let you do something you couldn’t do well before? If yes, it earns a hands-on test. If it’s just a faster version of something you’ve already covered, note it and move on.
Test, don’t read. Five minutes actually using a new capability tells you more than an hour of reviews and hot takes. You’ll know almost immediately whether it changes anything for your operation.
That’s the whole practice. It’s not a role. It’s not a committee. It’s a habit.
But here’s what the AI-skills conversation keeps missing
At RightWave, we’d add one hard-won caveat to all of this — because we see what happens when teams adopt the newest model on top of the same old foundation.
AI doesn’t neutralize bad data. It operationalizes it.
Every capability leap makes this more true, not less. A more powerful model scoring leads against a database full of duplicates, stale titles, and inconsistent firmographics doesn’t produce better lead scores. It produces confidently wrong lead scores, faster, at scale, wired directly into your routing and your reps’ follow-up queues.
So the 10-minute habit needs a companion discipline: every time you evaluate a new AI capability, evaluate whether your data is ready to feed it. The teams getting real leverage from AI aren’t just the ones testing new tools first — they’re the ones whose normalization, enrichment, and governance are strong enough that a new model can actually be trusted with production workflows on day one.
That’s the readiness gap most teams don’t see until it’s expensive.
What this looks like in practice
If you run marketing operations, here’s a simple operating rhythm we recommend:
- Weekly (10 min): One person scans for genuinely new AI capabilities relevant to your funnel — content, data, scoring, orchestration.
- When something passes the filter: Run a small hands-on test against a real (sandboxed) use case. Real data, low stakes.
- Before anything touches production: Ask the data question. Is the input clean, normalized, and current enough that this capability’s output can be trusted downstream?
- Quarterly: Compare what your stack does against what’s now possible. Retire what’s been leapfrogged.
None of this requires a bigger budget or a new hire. It requires a small habit and the willingness to open the tool instead of just reading about it.
The advantage in the next few years won’t go to the teams with the biggest AI budgets. It will go to the teams who notice what’s newly possible, whose data is ready for it, and who move while everyone else is still scheduling the evaluation meeting.
Stay close to the pace. Keep the foundation clean. That combination is the moat.
RightWave helps B2B marketing teams build the data foundation and operational discipline that make AI adoption actually pay off. If your team is evaluating new AI capabilities and wondering whether your data is ready for them, let’s talk.
Reference – https://martech.org/the-most-valuable-ai-skill-takes-10-minutes-a-week/
